├── .clang-format ├── .github └── workflows │ └── build_check.yml ├── .gitignore ├── .gitmodules ├── .pre-commit-config.yaml ├── CMakeLists.txt ├── CODEOWNERS ├── CONTRIBUTING.md ├── LICENSE ├── MANIFEST.in ├── NOTICE ├── README.md ├── docker ├── Pypi │ └── Dockerfile ├── README.md └── Tritonserver │ └── Dockerfile ├── docs ├── build.md ├── examples.md ├── guide.md ├── images │ ├── features.png │ ├── generation.png │ ├── logo.png │ ├── nmt.png │ ├── single_step.png │ ├── support.png │ └── total_time.png └── performance.md ├── examples ├── inference │ ├── benchmark_bart.sh │ ├── benchmark_gpt.sh │ ├── benchmark_quant_bart.sh │ ├── benchmark_quant_gpt.sh │ ├── cpp │ │ ├── CMakeLists.txt │ │ ├── bert_example.cc │ │ ├── decoder_example.cc.cu │ │ ├── gpt_example.cc │ │ ├── quant_bert_example.cc │ │ ├── quant_gpt_example.cc │ │ ├── quant_transformer_example.cc │ │ ├── quant_vit_example.cc │ │ ├── transformer_example.cc │ │ └── vit_example.cc │ └── python │ │ ├── README.md │ │ ├── __init__.py │ │ ├── export │ │ ├── __init__.py │ │ ├── fairseq │ │ │ ├── __init__.py │ │ │ ├── ls_fs_quant_transformer_export.py │ │ │ ├── ls_fs_transformer_export.py │ │ │ ├── ls_fs_transformer_ptq_export.py │ │ │ ├── ls_torch_fs_quant_transformer_export.py │ │ │ ├── ls_torch_fs_transformer_export.py │ │ │ ├── ls_torch_fs_transformer_ptq_export.py │ │ │ ├── native_fs_moe_transformer_export.py │ │ │ ├── native_fs_transformer_export.py │ │ │ └── native_fs_transformer_ptq_export.py │ │ ├── huggingface │ │ │ ├── __init__.py │ │ │ ├── hf_bart_export.py │ │ │ ├── hf_bert_export.py │ │ │ ├── hf_gpt2_export.py │ │ │ ├── hf_mt5_export.py │ │ │ ├── hf_t5_export.py │ │ │ ├── hf_vit_export.py │ │ │ ├── hf_xglm_export.py │ │ │ ├── ls_hf_bert_export.py │ │ │ ├── ls_hf_gpt2_export.py │ │ │ ├── ls_hf_quant_bert_export.py │ │ │ ├── ls_hf_quant_gpt2_export.py │ │ │ ├── ls_hf_quant_vit_export.py │ │ │ ├── ls_hf_vit_export.py │ │ │ ├── ls_torch_hf_quant_bert_export.py │ │ │ └── ls_torch_hf_quant_gpt2_export.py │ │ ├── ls_quant_transformer_export.py │ │ ├── ls_transformer_export.py │ │ ├── ls_transformer_ptq_export.py │ │ ├── proto │ │ │ ├── __init__.py │ │ │ ├── bert_pb2.py │ │ │ ├── gpt_pb2.py │ │ │ ├── moe_pb2.py │ │ │ ├── mt5_pb2.py │ │ │ ├── quant_transformer_pb2.py │ │ │ ├── t5_pb2.py │ │ │ └── transformer_pb2.py │ │ └── util.py │ │ ├── generate_model.py │ │ └── test │ │ ├── ls_bart.py │ │ ├── ls_bart_bench.py │ │ ├── ls_bert.py │ │ ├── ls_bert_bench.py │ │ ├── ls_fairseq.sh │ │ ├── ls_gpt2.py │ │ ├── ls_gpt2_bench.py │ │ ├── ls_mt5.py │ │ ├── ls_quant_bert.py │ │ ├── ls_quant_gpt2.py │ │ ├── ls_quant_vit.py │ │ ├── ls_t5.py │ │ ├── ls_vit.py │ │ ├── ls_vit_bench.py │ │ └── ls_xglm.py ├── training │ ├── custom │ │ ├── README.md │ │ ├── run.py │ │ ├── run.sh │ │ └── run_quant.sh │ ├── deepspeed │ │ ├── README.md │ │ ├── __init__.py │ │ ├── deepspeed_config.json │ │ ├── ds_fairseq.py │ │ ├── ds_fairseq_argument.py │ │ ├── ds_fairseq_data.py │ │ └── ds_fairseq_wmt14en2de.sh │ ├── fairseq │ │ ├── README.md │ │ ├── ls_fairseq_gcq_wmt14en2de.sh │ │ ├── ls_fairseq_quant_wmt14en2de.sh │ │ ├── ls_fairseq_wmt14en2de.sh │ │ ├── ls_finetune_bart │ │ │ ├── convert_lightseq_to_huggingface.py │ │ │ ├── convert_lightseq_to_huggingface.sh │ │ │ └── ls_fairseq_summarization_cnn_dm.sh │ │ ├── ls_preprocess_sharded_databin.sh │ │ ├── ls_torch_fairseq_quant_wmt14en2de.sh │ │ ├── ls_torch_fairseq_wmt14en2de.sh │ │ ├── native_fairseq_wmt14en2de.sh │ │ └── requirements.txt │ ├── huggingface │ │ ├── README.md │ │ ├── bart │ │ │ └── summarization │ │ │ │ ├── ls_bart_model.py │ │ │ │ ├── requirements.txt │ │ │ │ ├── run_summarization.py │ │ │ │ └── run_summarization.sh │ │ ├── bert │ │ │ ├── __init__.py │ │ │ ├── ls_hf_transformer_layer.py │ │ │ ├── task_glue │ │ │ │ ├── run_gcq_glue.py │ │ │ │ ├── run_gcq_glue.sh │ │ │ │ ├── run_glue.py │ │ │ │ ├── run_glue.sh │ │ │ │ └── run_quant_glue.sh │ │ │ ├── task_ner │ │ │ │ ├── run_gcq_ner.py │ │ │ │ ├── run_gcq_ner.sh │ │ │ │ ├── run_ner.py │ │ │ │ ├── run_ner.sh │ │ │ │ └── run_quant_ner.sh │ │ │ └── task_qa │ │ │ │ ├── run_gcq_qa.py │ │ │ │ ├── run_gcq_qa.sh │ │ │ │ ├── run_qa.py │ │ │ │ ├── run_qa.sh │ │ │ │ ├── run_quant_qa.sh │ │ │ │ ├── trainer_gcq_qa.py │ │ │ │ ├── trainer_qa.py │ │ │ │ └── utils_qa.py │ │ ├── gcq │ │ │ ├── __init__.py │ │ │ ├── cli_utils.py │ │ │ └── ls_hf_gcq_trainer.py │ │ ├── gpt │ │ │ ├── __init__.py │ │ │ ├── ls_hf_gpt_layer.py │ │ │ ├── requirements.txt │ │ │ ├── run_clm.py │ │ │ ├── run_clm.sh │ │ │ ├── run_gcq_clm.py │ │ │ ├── run_gcq_clm.sh │ │ │ └── run_quant_clm.sh │ │ └── vit │ │ │ ├── __init__.py │ │ │ ├── ls_hf_vit_encoder_layer.py │ │ │ ├── run_gcq_vit.py │ │ │ ├── run_gcq_vit.sh │ │ │ ├── run_quant_vit.sh │ │ │ ├── run_vit.py │ │ │ └── run_vit.sh │ └── neurst │ │ ├── README.md │ │ └── __init__.py └── triton_backend │ ├── README.md │ ├── model_repo │ ├── bert_example │ │ ├── 1 │ │ │ └── .gitignore │ │ └── config.pbtxt │ ├── gpt_example │ │ ├── 1 │ │ │ └── .gitignore │ │ └── config.pbtxt │ └── transformer_example │ │ ├── 1 │ │ └── .gitignore │ │ └── config.pbtxt │ └── transformer_client_example.py ├── lightseq ├── __init__.py ├── csrc │ ├── .gitignore │ ├── example │ │ ├── CMakeLists.txt │ │ ├── bert_example.cc │ │ ├── gpt_example.cc │ │ ├── llama_example.cc │ │ └── transformer_example.cc │ ├── export │ │ ├── __init__.py │ │ ├── hf_llama_export.py │ │ └── util.py │ ├── kernels │ │ ├── arm │ │ │ ├── CMakeLists.txt │ │ │ ├── gemm.cc │ │ │ ├── includes │ │ │ │ ├── kernel_headers.h │ │ │ │ └── utils.h │ │ │ └── utils.cc │ │ ├── cuda │ │ │ ├── CMakeLists.txt │ │ │ ├── crf.cu │ │ │ ├── cross_entropy.cu │ │ │ ├── cublas_algo_map.cpp │ │ │ ├── cublas_wrappers.cpp │ │ │ ├── cublas_wrappers.cu │ │ │ ├── cuda_util.cu │ │ │ ├── dropout_kernels.cu │ │ │ ├── embKernels.cc.cu │ │ │ ├── embedding_kernels.cu │ │ │ ├── fused_adam_kernel.cu │ │ │ ├── gemm_test.cpp │ │ │ ├── general_kernels.cu │ │ │ ├── gptKernels.cc.cu │ │ │ ├── includes │ │ │ │ ├── block_reduce.h │ │ │ │ ├── common.h │ │ │ │ ├── cublas_algo_map.h │ │ │ │ ├── cublas_wrappers.h │ │ │ │ ├── cuda_util.h │ │ │ │ ├── embKernels.h │ │ │ │ ├── fused_adam_kernel.h │ │ │ │ ├── gptKernels.h │ │ │ │ ├── kernel_headers.h │ │ │ │ ├── kernels.h │ │ │ │ ├── llama_kernels.h │ │ │ │ ├── ls_cub.cuh │ │ │ │ ├── multi_tensor_apply.cuh │ │ │ │ ├── transformerKernels.h │ │ │ │ └── util.h │ │ │ ├── llama_kernels.cu │ │ │ ├── normalize_kernels.cu │ │ │ ├── quantize_kernels.cu │ │ │ ├── softmax_kernels.cu │ │ │ ├── softmax_kernels_new.cu │ │ │ ├── transform_kernels.cu │ │ │ ├── transform_kernels_new.cu │ │ │ ├── transformerKernels.cc.cu │ │ │ └── util.cc.cu │ │ └── x86 │ │ │ ├── CMakeLists.txt │ │ │ ├── gemm.cpp │ │ │ ├── includes │ │ │ ├── kernel_headers.h │ │ │ ├── kernels.h │ │ │ └── util.h │ │ │ └── util.cc │ ├── layers │ │ ├── cross_entropy_layer.cpp │ │ ├── includes │ │ │ ├── cross_entropy_layer.h │ │ │ ├── quant_linear_layer.h │ │ │ ├── transformer_decoder_layer.h │ │ │ ├── transformer_embedding_layer.h │ │ │ └── transformer_encoder_layer.h │ │ ├── quant_linear_layer.cpp │ │ ├── transformer_decoder_layer.cpp │ │ ├── transformer_embedding_layer.cpp │ │ └── transformer_encoder_layer.cpp │ ├── layers_new │ │ ├── CMakeLists.txt │ │ ├── crf_layer.cpp │ │ ├── dec_enc_attention_layer.cpp │ │ ├── dec_self_attention_layer.cpp │ │ ├── encdec_kv_layer.cpp │ │ ├── feed_forward_layer.cpp │ │ ├── generator_layer.cpp │ │ ├── gpt_attention_layer.cpp │ │ ├── gpt_layer.cpp │ │ ├── includes │ │ │ ├── crf_layer.h │ │ │ ├── dec_enc_attention_layer.h │ │ │ ├── dec_self_attention_layer.h │ │ │ ├── encdec_kv_layer.h │ │ │ ├── feed_forward_layer.h │ │ │ ├── generator_layer.h │ │ │ ├── gpt_attention_layer.h │ │ │ ├── gpt_layer.h │ │ │ ├── launch_dec_emb_layer.h │ │ │ ├── launch_enc_emb_layer.h │ │ │ ├── launch_gpt_emb_layer.h │ │ │ ├── launch_llama_emb_layer.h │ │ │ ├── linear_layer.h │ │ │ ├── llama_attention_layer.h │ │ │ ├── llama_layer.h │ │ │ ├── llama_mlp_layer.h │ │ │ ├── lyr_normalize_layer.h │ │ │ ├── multihead_attention_layer.h │ │ │ ├── rms_norm_layer.h │ │ │ ├── sample_layer.h │ │ │ ├── sdpa_layer.h │ │ │ ├── transformer_decoder_layer.h │ │ │ └── transformer_encoder_layer.h │ │ ├── launch_gpt_emb_layer.cpp │ │ ├── linear_layer.cpp │ │ ├── llama_attention_layer.cpp │ │ ├── llama_layer.cpp │ │ ├── llama_mlp_layer.cpp │ │ ├── multihead_attention_layer.cpp │ │ ├── sample_layer.cpp │ │ ├── sdpa_layer.cpp │ │ ├── transformer_decoder_layer.cpp │ │ └── transformer_encoder_layer.cpp │ ├── lsflow │ │ ├── CMakeLists.txt │ │ ├── README.md │ │ ├── allocator.cpp │ │ ├── context.cpp │ │ ├── includes │ │ │ ├── allocator.h │ │ │ ├── context.h │ │ │ ├── declaration.h │ │ │ ├── layer.h │ │ │ ├── lsflow_util.h │ │ │ ├── manager.h │ │ │ ├── node.h │ │ │ ├── shape.h │ │ │ └── tensor.h │ │ ├── layer.cpp │ │ ├── lsflow_util.cpp │ │ ├── manager.cpp │ │ ├── node.cpp │ │ ├── operator.cpp │ │ ├── shape.cpp │ │ ├── tensor.cpp │ │ └── variable.cpp │ ├── models │ │ ├── CMakeLists.txt │ │ ├── bert.cc │ │ ├── bert_crf.cc │ │ ├── gpt.cc │ │ ├── includes │ │ │ ├── bert.h │ │ │ ├── bert_crf.h │ │ │ ├── gpt.h │ │ │ ├── llama.h │ │ │ ├── model_base.h │ │ │ ├── model_util.h │ │ │ └── transformer.h │ │ ├── llama.cc │ │ ├── model_util.cc │ │ ├── test_layer.cc │ │ └── transformer.cu │ ├── ops │ │ └── includes │ │ │ ├── context.h │ │ │ ├── dropout.h │ │ │ ├── feed_forward.h │ │ │ ├── normalize_layer.h │ │ │ ├── softmax.h │ │ │ └── strided_batch_gemm.h │ ├── ops_new │ │ ├── CMakeLists.txt │ │ ├── act_elewise_product.cpp │ │ ├── beam_search_topk.cu │ │ ├── bias_act_dropout.cpp │ │ ├── bias_add_transform_20314.cpp │ │ ├── bias_dropout_residual.cpp │ │ ├── concat3_dim1.cpp │ │ ├── crf.cpp │ │ ├── dropout.cpp │ │ ├── fuse_add2_op.cpp │ │ ├── fuse_rotary_position_qkv.cpp │ │ ├── includes │ │ │ ├── act_elewise_product.h │ │ │ ├── beam_search_topk.h │ │ │ ├── bias_act_dropout.h │ │ │ ├── bias_add_transform_20314.h │ │ │ ├── bias_dropout_residual.h │ │ │ ├── concat3_dim1.h │ │ │ ├── crf.h │ │ │ ├── dropout.h │ │ │ ├── fuse_add2_op.h │ │ │ ├── fuse_rotary_position_qkv.h │ │ │ ├── launch_dec_emb_op.h │ │ │ ├── launch_enc_emb.h │ │ │ ├── launch_gpt_emb.h │ │ │ ├── launch_llama_emb.h │ │ │ ├── layer_normalize.h │ │ │ ├── linear.h │ │ │ ├── rms_layer_norm.h │ │ │ ├── sampling.h │ │ │ ├── softmax.h │ │ │ ├── split_head_op.h │ │ │ ├── strided_batch_gemm.h │ │ │ └── transform_0213.h │ │ ├── launch_dec_emb_op.cpp │ │ ├── launch_enc_emb.cpp │ │ ├── launch_gpt_emb.cpp │ │ ├── launch_llama_emb.cpp │ │ ├── layer_normalize.cpp │ │ ├── linear.cpp │ │ ├── rms_layer_norm.cpp │ │ ├── sampling.cc.cu │ │ ├── softmax.cpp │ │ ├── split_head_op.cpp │ │ ├── strided_batch_gemm.cpp │ │ └── transform_0213.cpp │ ├── proto │ │ ├── CMakeLists.txt │ │ ├── bert.proto │ │ ├── bert_crf.proto │ │ ├── bert_crf_weight.cc │ │ ├── bert_weight.cc │ │ ├── gpt.proto │ │ ├── gpt_weight.cc │ │ ├── includes │ │ │ ├── bert_crf_weight.h │ │ │ ├── bert_weight.h │ │ │ ├── gpt_weight.h │ │ │ ├── hdf5_util.h │ │ │ ├── llama_weight.h │ │ │ ├── proto_headers.h │ │ │ ├── proto_util.h │ │ │ ├── test_model_weight.h │ │ │ └── transformer_weight.h │ │ ├── llama.proto │ │ ├── llama_weight.cc │ │ ├── proto_util.cc │ │ ├── transformer.proto │ │ └── transformer_weight.cc │ ├── pybind │ │ ├── CMakeLists.txt │ │ ├── pybind_adam.cpp │ │ ├── pybind_kernel_cuda.cpp │ │ ├── pybind_kernel_x86.cpp │ │ ├── pybind_layer.cpp │ │ ├── pybind_layer_new.cpp │ │ ├── pybind_model.cpp │ │ └── pybind_op.cpp │ ├── pytorch │ │ ├── __init__.py │ │ ├── builder │ │ │ ├── __init__.py │ │ │ ├── builder.py │ │ │ ├── cuda_kernel_builder.py │ │ │ ├── cuda_layer_builder.py │ │ │ └── x86_kernel_builder.py │ │ ├── layer_base.py │ │ ├── pytorch_quantization │ │ │ ├── __init__.py │ │ │ ├── calib │ │ │ │ ├── __init__.py │ │ │ │ ├── calibrator.py │ │ │ │ ├── histogram.py │ │ │ │ └── max.py │ │ │ ├── nn │ │ │ │ ├── __init__.py │ │ │ │ ├── _functions │ │ │ │ │ ├── __init__.py │ │ │ │ │ └── quant_rnn.py │ │ │ │ ├── functional.py │ │ │ │ └── modules │ │ │ │ │ ├── __init__.py │ │ │ │ │ ├── _utils.py │ │ │ │ │ ├── clip.py │ │ │ │ │ ├── quant_bert.py │ │ │ │ │ ├── quant_conv.py │ │ │ │ │ ├── quant_instancenorm.py │ │ │ │ │ ├── quant_linear.py │ │ │ │ │ ├── quant_pooling.py │ │ │ │ │ ├── quant_rnn.py │ │ │ │ │ └── tensor_quantizer.py │ │ │ ├── optim │ │ │ │ ├── __init__.py │ │ │ │ └── helper.py │ │ │ ├── quant_modules.py │ │ │ ├── tensor_quant.py │ │ │ ├── utils │ │ │ │ ├── __init__.py │ │ │ │ ├── quant_logging.py │ │ │ │ └── reduce_amax.py │ │ │ └── version.py │ │ ├── quantization.py │ │ ├── sdpa_layers.py │ │ ├── torch_transformer_layers.py │ │ ├── transformer_decoder_layer.py │ │ ├── transformer_encoder_layer.py │ │ └── util.py │ ├── tensorflow │ │ └── README.md │ ├── tests │ │ ├── __init__.py │ │ ├── cuda │ │ │ ├── __init__.py │ │ │ ├── fairseq_layers.py │ │ │ ├── test_kernel.py │ │ │ ├── test_layer.py │ │ │ ├── test_ls_small_layer.py │ │ │ └── torch_crf.py │ │ ├── util.py │ │ └── x86 │ │ │ └── test_kernel.py │ └── triton_backend │ │ ├── CMakeLists.txt │ │ ├── cmake │ │ └── TutorialMinimalBackendConfig.cmake.in │ │ └── src │ │ ├── libtriton_minimal.ldscript │ │ ├── lightseq_backend.cc │ │ ├── triton_model.h │ │ └── triton_utils.h ├── inference │ ├── kernels │ │ ├── CMakeLists.txt │ │ ├── common.h │ │ ├── embKernels.cc.cu │ │ ├── embKernels.h │ │ ├── embKernels_int8.cc.cu │ │ ├── embKernels_int8.h │ │ ├── gptKernels.cc.cu │ │ ├── gptKernels.h │ │ ├── gptKernels_int8.cc.cu │ │ ├── gptKernels_int8.h │ │ ├── moeKernels.cc.cu │ │ ├── moeKernels.h │ │ ├── multilgKernels.cc.cu │ │ ├── multilgKernels.h │ │ ├── t5EmbKernels.cc.cu │ │ ├── t5EmbKernels.h │ │ ├── t5Kernels.cc.cu │ │ ├── t5Kernels.h │ │ ├── transformerKernels.cc.cu │ │ ├── transformerKernels.h │ │ ├── transformerKernels_int8.cc.cu │ │ └── transformerKernels_int8.h │ ├── model │ │ ├── CMakeLists.txt │ │ ├── bert_encoder.cc.cu │ │ ├── bert_encoder.h │ │ ├── cublas_algo_map.cc │ │ ├── cublas_algo_map.h │ │ ├── cublas_helper.cc │ │ ├── cublas_helper.h │ │ ├── decoder.cc.cu │ │ ├── decoder.h │ │ ├── encoder.cc.cu │ │ ├── encoder.h │ │ ├── gpt_encoder.cc.cu │ │ ├── gpt_encoder.h │ │ ├── moe_decoder.cc.cu │ │ ├── moe_decoder.h │ │ ├── moe_encoder.cc.cu │ │ ├── moe_encoder.h │ │ ├── mt5_decoder.cc.cu │ │ ├── mt5_decoder.h │ │ ├── mt5_encoder.cc.cu │ │ ├── mt5_encoder.h │ │ ├── quant_bert_encoder.cc.cu │ │ ├── quant_bert_encoder.h │ │ ├── quant_decoder.cc.cu │ │ ├── quant_decoder.h │ │ ├── quant_encoder.cc.cu │ │ ├── quant_encoder.h │ │ ├── quant_gpt_encoder.cc.cu │ │ ├── quant_gpt_encoder.h │ │ ├── quant_vit_encoder.cc.cu │ │ ├── quant_vit_encoder.h │ │ ├── t5_decoder.cc.cu │ │ ├── t5_decoder.h │ │ ├── t5_encoder.cc.cu │ │ ├── t5_encoder.h │ │ ├── vit_encoder.cc.cu │ │ └── vit_encoder.h │ ├── proto │ │ ├── CMakeLists.txt │ │ ├── bert.proto │ │ ├── bert_weight.cc │ │ ├── bert_weight.h │ │ ├── gpt.proto │ │ ├── gpt_weight.cc │ │ ├── gpt_weight.h │ │ ├── moe.proto │ │ ├── moe_weight.cc │ │ ├── moe_weight.h │ │ ├── mt5.proto │ │ ├── mt5_weight.cc │ │ ├── mt5_weight.h │ │ ├── quant_bert.proto │ │ ├── quant_bert_weight.cc │ │ ├── quant_bert_weight.h │ │ ├── quant_gpt.proto │ │ ├── quant_gpt_weight.cc │ │ ├── quant_gpt_weight.h │ │ ├── quant_transformer.proto │ │ ├── quant_transformer_weight.cc │ │ ├── quant_transformer_weight.h │ │ ├── quant_vit.proto │ │ ├── quant_vit_weight.cc │ │ ├── quant_vit_weight.h │ │ ├── t5.proto │ │ ├── t5_weight.cc │ │ ├── t5_weight.h │ │ ├── transformer.proto │ │ ├── transformer_weight.cc │ │ ├── transformer_weight.h │ │ ├── vit.proto │ │ ├── vit_weight.cc │ │ └── vit_weight.h │ ├── pywrapper │ │ ├── CMakeLists.txt │ │ ├── bert.cc │ │ ├── bert.h │ │ ├── gpt.cc │ │ ├── gpt.h │ │ ├── model_base.h │ │ ├── moe.cc │ │ ├── moe.h │ │ ├── mt5.cc │ │ ├── mt5.h │ │ ├── quant_bert.cc │ │ ├── quant_bert.h │ │ ├── quant_gpt.cc │ │ ├── quant_gpt.h │ │ ├── quant_transformer.cc │ │ ├── quant_transformer.h │ │ ├── quant_vit.cc │ │ ├── quant_vit.h │ │ ├── t5.cc │ │ ├── t5.h │ │ ├── transformer.cc │ │ ├── transformer.h │ │ ├── transformer_decoder.cc.cu │ │ ├── vit.cc │ │ ├── vit.h │ │ └── wrapper.cc │ ├── server │ │ ├── CMakeLists.txt │ │ ├── custom.h │ │ ├── decoder_generate_server.cc.cu │ │ ├── generate_server.cc.cu │ │ ├── gpt_generate_server.cc.cu │ │ ├── gptlm_server.cc.cu │ │ ├── libserver.ldscript │ │ ├── model_config.h │ │ ├── model_config.proto │ │ ├── model_config_cuda.h │ │ ├── moe_server.cc.cu │ │ └── transformer_server.cc.cu │ ├── tools │ │ ├── CMakeLists.txt │ │ ├── util.cc.cu │ │ └── util.h │ └── triton_backend │ │ ├── CMakeLists.txt │ │ ├── cmake │ │ └── TutorialMinimalBackendConfig.cmake.in │ │ └── src │ │ ├── libtriton_minimal.ldscript │ │ ├── lightseq_backend.cc │ │ ├── triton_model.h │ │ └── triton_utils.h └── training │ ├── __init__.py │ ├── cli │ ├── __init__.py │ ├── fs_modules │ │ ├── __init__.py │ │ ├── ls_adam.py │ │ ├── ls_bart.py │ │ ├── ls_fs_transformer_decoder_layer.py │ │ ├── ls_label_smoothed_cross_entropy.py │ │ ├── ls_transformer.py │ │ └── ls_translation.py │ ├── lightseq_deepspeed_cli.py │ ├── lightseq_fairseq_generate_cli.py │ ├── lightseq_fairseq_train_cli.py │ ├── lightseq_fairseq_validate_cli.py │ └── lightseq_infer_cli.py │ ├── csrc │ └── ops │ │ └── includes │ │ └── strided_batch_gemm.h │ ├── gcq │ ├── __init__.py │ ├── gcq.py │ ├── ls_fs_gcq_train.py │ └── ls_fs_gcq_trainer.py │ ├── ops │ ├── __init__.py │ ├── pytorch │ │ ├── __init__.py │ │ ├── adam.py │ │ ├── builder │ │ │ ├── __init__.py │ │ │ ├── adam_builder.py │ │ │ ├── builder.py │ │ │ ├── kernel_builder.py │ │ │ ├── layer_builder.py │ │ │ ├── operator_builder.py │ │ │ └── transformer_builder.py │ │ ├── cross_entropy_layer.py │ │ ├── export.py │ │ ├── export_quant.py │ │ ├── gemm_test.py │ │ ├── gpt_layer.py │ │ ├── layer_base.py │ │ ├── quant_linear_layer.py │ │ ├── quantization.py │ │ ├── torch_transformer_layers.py │ │ ├── transformer.py │ │ ├── transformer_decoder_layer.py │ │ ├── transformer_decoder_layer_new.py │ │ ├── transformer_embedding_layer.py │ │ ├── transformer_encoder_layer.py │ │ ├── transformer_encoder_layer_new.py │ │ └── util.py │ └── tensorflow │ │ ├── README.md │ │ └── __init__.py │ └── pytorch_quantization │ ├── __init__.py │ ├── calib │ ├── __init__.py │ ├── calibrator.py │ ├── histogram.py │ └── max.py │ ├── nn │ ├── __init__.py │ ├── _functions │ │ ├── __init__.py │ │ └── quant_rnn.py │ ├── functional.py │ └── modules │ │ ├── __init__.py │ │ ├── _utils.py │ │ ├── clip.py │ │ ├── quant_bert.py │ │ ├── quant_conv.py │ │ ├── quant_instancenorm.py │ │ ├── quant_linear.py │ │ ├── quant_pooling.py │ │ ├── quant_rnn.py │ │ └── tensor_quantizer.py │ ├── optim │ ├── __init__.py │ └── helper.py │ ├── quant_modules.py │ ├── tensor_quant.py │ ├── utils │ ├── __init__.py │ ├── quant_logging.py │ └── reduce_amax.py │ └── version.py ├── setup.py └── tests ├── __init__.py ├── cublas ├── build.sh ├── gemm.cu ├── gemm.h ├── test.cpp └── util.h ├── fairseq_layers.py ├── gemm_test ├── configs │ ├── igemm_A10.cfg │ ├── igemm_A100.cfg │ ├── igemm_A30.cfg │ └── igemm_T4.cfg ├── gemm_test.py └── gemm_test.sh ├── huggingface └── test_gpt.py ├── quant_test.py ├── test_ls_kernels.py ├── test_ls_layers.py ├── test_ls_layers_new.py ├── test_ls_ops.py ├── torch_crf.py ├── tvm └── matmul_autotune.py └── util.py /.clang-format: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/.clang-format -------------------------------------------------------------------------------- /.github/workflows/build_check.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/.github/workflows/build_check.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/.gitignore -------------------------------------------------------------------------------- /.gitmodules: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/.gitmodules -------------------------------------------------------------------------------- /.pre-commit-config.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/.pre-commit-config.yaml -------------------------------------------------------------------------------- /CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/CMakeLists.txt -------------------------------------------------------------------------------- /CODEOWNERS: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/CODEOWNERS -------------------------------------------------------------------------------- /CONTRIBUTING.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/CONTRIBUTING.md -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/LICENSE -------------------------------------------------------------------------------- /MANIFEST.in: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/MANIFEST.in -------------------------------------------------------------------------------- /NOTICE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/NOTICE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/README.md -------------------------------------------------------------------------------- /docker/Pypi/Dockerfile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docker/Pypi/Dockerfile -------------------------------------------------------------------------------- /docker/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docker/README.md -------------------------------------------------------------------------------- /docker/Tritonserver/Dockerfile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docker/Tritonserver/Dockerfile -------------------------------------------------------------------------------- /docs/build.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/build.md -------------------------------------------------------------------------------- /docs/examples.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/examples.md -------------------------------------------------------------------------------- /docs/guide.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/guide.md -------------------------------------------------------------------------------- /docs/images/features.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/features.png -------------------------------------------------------------------------------- /docs/images/generation.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/generation.png -------------------------------------------------------------------------------- /docs/images/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/logo.png -------------------------------------------------------------------------------- /docs/images/nmt.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/nmt.png -------------------------------------------------------------------------------- /docs/images/single_step.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/single_step.png -------------------------------------------------------------------------------- /docs/images/support.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/support.png -------------------------------------------------------------------------------- /docs/images/total_time.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/images/total_time.png -------------------------------------------------------------------------------- /docs/performance.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/docs/performance.md -------------------------------------------------------------------------------- /examples/inference/benchmark_bart.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/benchmark_bart.sh -------------------------------------------------------------------------------- /examples/inference/benchmark_gpt.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/benchmark_gpt.sh -------------------------------------------------------------------------------- /examples/inference/benchmark_quant_bart.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/benchmark_quant_bart.sh -------------------------------------------------------------------------------- /examples/inference/benchmark_quant_gpt.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/benchmark_quant_gpt.sh -------------------------------------------------------------------------------- /examples/inference/cpp/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/CMakeLists.txt -------------------------------------------------------------------------------- /examples/inference/cpp/bert_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/bert_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/decoder_example.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/decoder_example.cc.cu -------------------------------------------------------------------------------- /examples/inference/cpp/gpt_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/gpt_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/quant_bert_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/quant_bert_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/quant_gpt_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/quant_gpt_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/quant_transformer_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/quant_transformer_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/quant_vit_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/quant_vit_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/transformer_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/transformer_example.cc -------------------------------------------------------------------------------- /examples/inference/cpp/vit_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/cpp/vit_example.cc -------------------------------------------------------------------------------- /examples/inference/python/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/README.md -------------------------------------------------------------------------------- /examples/inference/python/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/inference/python/export/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/ls_fs_quant_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/ls_fs_quant_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/ls_fs_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/ls_fs_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/ls_fs_transformer_ptq_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/ls_fs_transformer_ptq_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/ls_torch_fs_quant_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/ls_torch_fs_quant_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/ls_torch_fs_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/ls_torch_fs_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/ls_torch_fs_transformer_ptq_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/ls_torch_fs_transformer_ptq_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/native_fs_moe_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/native_fs_moe_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/native_fs_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/native_fs_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/fairseq/native_fs_transformer_ptq_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/fairseq/native_fs_transformer_ptq_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_bart_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_bart_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_bert_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_bert_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_gpt2_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_gpt2_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_mt5_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_mt5_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_t5_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_t5_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_vit_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_vit_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/hf_xglm_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/hf_xglm_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_hf_bert_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_hf_bert_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_hf_gpt2_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_hf_gpt2_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_hf_quant_bert_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_hf_quant_bert_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_hf_quant_gpt2_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_hf_quant_gpt2_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_hf_quant_vit_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_hf_quant_vit_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_hf_vit_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_hf_vit_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_torch_hf_quant_bert_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_torch_hf_quant_bert_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/huggingface/ls_torch_hf_quant_gpt2_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/huggingface/ls_torch_hf_quant_gpt2_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/ls_quant_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/ls_quant_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/ls_transformer_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/ls_transformer_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/ls_transformer_ptq_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/ls_transformer_ptq_export.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/inference/python/export/proto/bert_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/bert_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/gpt_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/gpt_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/moe_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/moe_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/mt5_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/mt5_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/quant_transformer_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/quant_transformer_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/t5_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/t5_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/proto/transformer_pb2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/proto/transformer_pb2.py -------------------------------------------------------------------------------- /examples/inference/python/export/util.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/export/util.py -------------------------------------------------------------------------------- /examples/inference/python/generate_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/generate_model.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_bart.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_bart.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_bart_bench.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_bart_bench.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_bert.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_bert.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_bert_bench.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_bert_bench.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_fairseq.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_fairseq.sh -------------------------------------------------------------------------------- /examples/inference/python/test/ls_gpt2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_gpt2.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_gpt2_bench.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_gpt2_bench.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_mt5.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_mt5.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_quant_bert.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_quant_bert.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_quant_gpt2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_quant_gpt2.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_quant_vit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_quant_vit.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_t5.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_t5.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_vit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_vit.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_vit_bench.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_vit_bench.py -------------------------------------------------------------------------------- /examples/inference/python/test/ls_xglm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/inference/python/test/ls_xglm.py -------------------------------------------------------------------------------- /examples/training/custom/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/custom/README.md -------------------------------------------------------------------------------- /examples/training/custom/run.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/custom/run.py -------------------------------------------------------------------------------- /examples/training/custom/run.sh: -------------------------------------------------------------------------------- 1 | python3 examples/training/custom/run.py 2 | -------------------------------------------------------------------------------- /examples/training/custom/run_quant.sh: -------------------------------------------------------------------------------- 1 | python3 examples/training/custom/run.py --enable_quant 2 | -------------------------------------------------------------------------------- /examples/training/deepspeed/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/deepspeed/README.md -------------------------------------------------------------------------------- /examples/training/deepspeed/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/training/deepspeed/deepspeed_config.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/deepspeed/deepspeed_config.json -------------------------------------------------------------------------------- /examples/training/deepspeed/ds_fairseq.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/deepspeed/ds_fairseq.py -------------------------------------------------------------------------------- /examples/training/deepspeed/ds_fairseq_argument.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/deepspeed/ds_fairseq_argument.py -------------------------------------------------------------------------------- /examples/training/deepspeed/ds_fairseq_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/deepspeed/ds_fairseq_data.py -------------------------------------------------------------------------------- /examples/training/deepspeed/ds_fairseq_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/deepspeed/ds_fairseq_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/README.md -------------------------------------------------------------------------------- /examples/training/fairseq/ls_fairseq_gcq_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_fairseq_gcq_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_fairseq_quant_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_fairseq_quant_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_fairseq_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_fairseq_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_finetune_bart/convert_lightseq_to_huggingface.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_finetune_bart/convert_lightseq_to_huggingface.py -------------------------------------------------------------------------------- /examples/training/fairseq/ls_finetune_bart/convert_lightseq_to_huggingface.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_finetune_bart/convert_lightseq_to_huggingface.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_finetune_bart/ls_fairseq_summarization_cnn_dm.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_finetune_bart/ls_fairseq_summarization_cnn_dm.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_preprocess_sharded_databin.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_preprocess_sharded_databin.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_torch_fairseq_quant_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_torch_fairseq_quant_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/ls_torch_fairseq_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/ls_torch_fairseq_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/native_fairseq_wmt14en2de.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/native_fairseq_wmt14en2de.sh -------------------------------------------------------------------------------- /examples/training/fairseq/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/fairseq/requirements.txt -------------------------------------------------------------------------------- /examples/training/huggingface/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/README.md -------------------------------------------------------------------------------- /examples/training/huggingface/bart/summarization/ls_bart_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bart/summarization/ls_bart_model.py -------------------------------------------------------------------------------- /examples/training/huggingface/bart/summarization/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bart/summarization/requirements.txt -------------------------------------------------------------------------------- /examples/training/huggingface/bart/summarization/run_summarization.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bart/summarization/run_summarization.py -------------------------------------------------------------------------------- /examples/training/huggingface/bart/summarization/run_summarization.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bart/summarization/run_summarization.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/training/huggingface/bert/ls_hf_transformer_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/ls_hf_transformer_layer.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_glue/run_gcq_glue.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_glue/run_gcq_glue.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_glue/run_gcq_glue.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_glue/run_gcq_glue.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_glue/run_glue.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_glue/run_glue.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_glue/run_glue.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_glue/run_glue.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_glue/run_quant_glue.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_glue/run_quant_glue.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_ner/run_gcq_ner.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_ner/run_gcq_ner.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_ner/run_gcq_ner.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_ner/run_gcq_ner.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_ner/run_ner.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_ner/run_ner.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_ner/run_ner.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_ner/run_ner.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_ner/run_quant_ner.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_ner/run_quant_ner.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/run_gcq_qa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/run_gcq_qa.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/run_gcq_qa.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/run_gcq_qa.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/run_qa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/run_qa.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/run_qa.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/run_qa.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/run_quant_qa.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/run_quant_qa.sh -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/trainer_gcq_qa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/trainer_gcq_qa.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/trainer_qa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/trainer_qa.py -------------------------------------------------------------------------------- /examples/training/huggingface/bert/task_qa/utils_qa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/bert/task_qa/utils_qa.py -------------------------------------------------------------------------------- /examples/training/huggingface/gcq/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gcq/__init__.py -------------------------------------------------------------------------------- /examples/training/huggingface/gcq/cli_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gcq/cli_utils.py -------------------------------------------------------------------------------- /examples/training/huggingface/gcq/ls_hf_gcq_trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gcq/ls_hf_gcq_trainer.py -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/ls_hf_gpt_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/ls_hf_gpt_layer.py -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/requirements.txt -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/run_clm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/run_clm.py -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/run_clm.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/run_clm.sh -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/run_gcq_clm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/run_gcq_clm.py -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/run_gcq_clm.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/run_gcq_clm.sh -------------------------------------------------------------------------------- /examples/training/huggingface/gpt/run_quant_clm.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/gpt/run_quant_clm.sh -------------------------------------------------------------------------------- /examples/training/huggingface/vit/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/training/huggingface/vit/ls_hf_vit_encoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/vit/ls_hf_vit_encoder_layer.py -------------------------------------------------------------------------------- /examples/training/huggingface/vit/run_gcq_vit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/vit/run_gcq_vit.py -------------------------------------------------------------------------------- /examples/training/huggingface/vit/run_gcq_vit.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/vit/run_gcq_vit.sh -------------------------------------------------------------------------------- /examples/training/huggingface/vit/run_quant_vit.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/vit/run_quant_vit.sh -------------------------------------------------------------------------------- /examples/training/huggingface/vit/run_vit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/vit/run_vit.py -------------------------------------------------------------------------------- /examples/training/huggingface/vit/run_vit.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/huggingface/vit/run_vit.sh -------------------------------------------------------------------------------- /examples/training/neurst/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/training/neurst/README.md -------------------------------------------------------------------------------- /examples/training/neurst/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/triton_backend/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/triton_backend/README.md -------------------------------------------------------------------------------- /examples/triton_backend/model_repo/bert_example/1/.gitignore: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/triton_backend/model_repo/bert_example/config.pbtxt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/triton_backend/model_repo/bert_example/config.pbtxt -------------------------------------------------------------------------------- /examples/triton_backend/model_repo/gpt_example/1/.gitignore: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/triton_backend/model_repo/gpt_example/config.pbtxt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/triton_backend/model_repo/gpt_example/config.pbtxt -------------------------------------------------------------------------------- /examples/triton_backend/model_repo/transformer_example/1/.gitignore: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /examples/triton_backend/model_repo/transformer_example/config.pbtxt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/triton_backend/model_repo/transformer_example/config.pbtxt -------------------------------------------------------------------------------- /examples/triton_backend/transformer_client_example.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/examples/triton_backend/transformer_client_example.py -------------------------------------------------------------------------------- /lightseq/__init__.py: -------------------------------------------------------------------------------- 1 | __version__ = "3.0.1" 2 | -------------------------------------------------------------------------------- /lightseq/csrc/.gitignore: -------------------------------------------------------------------------------- 1 | build/* 2 | -------------------------------------------------------------------------------- /lightseq/csrc/example/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/example/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/example/bert_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/example/bert_example.cc -------------------------------------------------------------------------------- /lightseq/csrc/example/gpt_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/example/gpt_example.cc -------------------------------------------------------------------------------- /lightseq/csrc/example/llama_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/example/llama_example.cc -------------------------------------------------------------------------------- /lightseq/csrc/example/transformer_example.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/example/transformer_example.cc -------------------------------------------------------------------------------- /lightseq/csrc/export/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /lightseq/csrc/export/hf_llama_export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/export/hf_llama_export.py -------------------------------------------------------------------------------- /lightseq/csrc/export/util.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/export/util.py -------------------------------------------------------------------------------- /lightseq/csrc/kernels/arm/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/arm/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/kernels/arm/gemm.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/arm/gemm.cc -------------------------------------------------------------------------------- /lightseq/csrc/kernels/arm/includes/kernel_headers.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/arm/includes/kernel_headers.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/arm/includes/utils.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/arm/includes/utils.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/arm/utils.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/arm/utils.cc -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/crf.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/crf.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/cross_entropy.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/cross_entropy.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/cublas_algo_map.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/cublas_algo_map.cpp -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/cublas_wrappers.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/cublas_wrappers.cpp -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/cublas_wrappers.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/cublas_wrappers.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/cuda_util.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/cuda_util.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/dropout_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/dropout_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/embKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/embKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/embedding_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/embedding_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/fused_adam_kernel.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/fused_adam_kernel.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/gemm_test.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/gemm_test.cpp -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/general_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/general_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/gptKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/gptKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/block_reduce.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/block_reduce.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/common.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/common.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/cublas_algo_map.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/cublas_algo_map.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/cublas_wrappers.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/cublas_wrappers.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/cuda_util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/cuda_util.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/embKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/embKernels.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/fused_adam_kernel.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/fused_adam_kernel.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/gptKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/gptKernels.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/kernel_headers.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/kernel_headers.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/kernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/kernels.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/llama_kernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/llama_kernels.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/ls_cub.cuh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/ls_cub.cuh -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/multi_tensor_apply.cuh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/multi_tensor_apply.cuh -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/transformerKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/transformerKernels.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/includes/util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/includes/util.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/llama_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/llama_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/normalize_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/normalize_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/quantize_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/quantize_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/softmax_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/softmax_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/softmax_kernels_new.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/softmax_kernels_new.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/transform_kernels.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/transform_kernels.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/transform_kernels_new.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/transform_kernels_new.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/transformerKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/transformerKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/cuda/util.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/cuda/util.cc.cu -------------------------------------------------------------------------------- /lightseq/csrc/kernels/x86/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/x86/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/kernels/x86/gemm.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/x86/gemm.cpp -------------------------------------------------------------------------------- /lightseq/csrc/kernels/x86/includes/kernel_headers.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/x86/includes/kernel_headers.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/x86/includes/kernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/x86/includes/kernels.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/x86/includes/util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/x86/includes/util.h -------------------------------------------------------------------------------- /lightseq/csrc/kernels/x86/util.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/kernels/x86/util.cc -------------------------------------------------------------------------------- /lightseq/csrc/layers/cross_entropy_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/cross_entropy_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers/includes/cross_entropy_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/includes/cross_entropy_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers/includes/quant_linear_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/includes/quant_linear_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers/includes/transformer_decoder_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/includes/transformer_decoder_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers/includes/transformer_embedding_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/includes/transformer_embedding_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers/includes/transformer_encoder_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/includes/transformer_encoder_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers/quant_linear_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/quant_linear_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers/transformer_decoder_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/transformer_decoder_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers/transformer_embedding_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/transformer_embedding_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers/transformer_encoder_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers/transformer_encoder_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/crf_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/crf_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/dec_enc_attention_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/dec_enc_attention_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/dec_self_attention_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/dec_self_attention_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/encdec_kv_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/encdec_kv_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/feed_forward_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/feed_forward_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/generator_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/generator_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/gpt_attention_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/gpt_attention_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/gpt_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/gpt_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/crf_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/crf_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/dec_enc_attention_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/dec_enc_attention_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/dec_self_attention_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/dec_self_attention_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/encdec_kv_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/encdec_kv_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/feed_forward_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/feed_forward_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/generator_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/generator_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/gpt_attention_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/gpt_attention_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/gpt_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/gpt_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/launch_dec_emb_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/launch_dec_emb_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/launch_enc_emb_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/launch_enc_emb_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/launch_gpt_emb_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/launch_gpt_emb_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/launch_llama_emb_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/launch_llama_emb_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/linear_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/linear_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/llama_attention_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/llama_attention_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/llama_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/llama_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/llama_mlp_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/llama_mlp_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/lyr_normalize_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/lyr_normalize_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/multihead_attention_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/multihead_attention_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/rms_norm_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/rms_norm_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/sample_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/sample_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/sdpa_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/sdpa_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/transformer_decoder_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/transformer_decoder_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/includes/transformer_encoder_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/includes/transformer_encoder_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/launch_gpt_emb_layer.cpp: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/linear_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/linear_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/llama_attention_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/llama_attention_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/llama_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/llama_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/llama_mlp_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/llama_mlp_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/multihead_attention_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/multihead_attention_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/sample_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/sample_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/sdpa_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/sdpa_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/transformer_decoder_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/transformer_decoder_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/layers_new/transformer_encoder_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/layers_new/transformer_encoder_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/README.md: -------------------------------------------------------------------------------- 1 | LsFlow is a extremely clean implement of computation graph. 2 | -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/allocator.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/allocator.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/context.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/context.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/allocator.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/allocator.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/context.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/context.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/declaration.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/declaration.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/layer.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/lsflow_util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/lsflow_util.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/manager.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/manager.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/node.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/node.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/shape.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/shape.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/includes/tensor.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/includes/tensor.h -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/lsflow_util.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/lsflow_util.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/manager.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/manager.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/node.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/node.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/operator.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/operator.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/shape.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/shape.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/tensor.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/tensor.cpp -------------------------------------------------------------------------------- /lightseq/csrc/lsflow/variable.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/lsflow/variable.cpp -------------------------------------------------------------------------------- /lightseq/csrc/models/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/models/bert.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/bert.cc -------------------------------------------------------------------------------- /lightseq/csrc/models/bert_crf.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/bert_crf.cc -------------------------------------------------------------------------------- /lightseq/csrc/models/gpt.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/gpt.cc -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/bert.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/bert.h -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/bert_crf.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/bert_crf.h -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/gpt.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/gpt.h -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/llama.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/llama.h -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/model_base.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/model_base.h -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/model_util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/model_util.h -------------------------------------------------------------------------------- /lightseq/csrc/models/includes/transformer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/includes/transformer.h -------------------------------------------------------------------------------- /lightseq/csrc/models/llama.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/llama.cc -------------------------------------------------------------------------------- /lightseq/csrc/models/model_util.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/model_util.cc -------------------------------------------------------------------------------- /lightseq/csrc/models/test_layer.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/test_layer.cc -------------------------------------------------------------------------------- /lightseq/csrc/models/transformer.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/models/transformer.cu -------------------------------------------------------------------------------- /lightseq/csrc/ops/includes/context.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops/includes/context.h -------------------------------------------------------------------------------- /lightseq/csrc/ops/includes/dropout.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops/includes/dropout.h -------------------------------------------------------------------------------- /lightseq/csrc/ops/includes/feed_forward.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops/includes/feed_forward.h -------------------------------------------------------------------------------- /lightseq/csrc/ops/includes/normalize_layer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops/includes/normalize_layer.h -------------------------------------------------------------------------------- /lightseq/csrc/ops/includes/softmax.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops/includes/softmax.h -------------------------------------------------------------------------------- /lightseq/csrc/ops/includes/strided_batch_gemm.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops/includes/strided_batch_gemm.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/act_elewise_product.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/act_elewise_product.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/beam_search_topk.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/beam_search_topk.cu -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/bias_act_dropout.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/bias_act_dropout.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/bias_add_transform_20314.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/bias_add_transform_20314.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/bias_dropout_residual.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/bias_dropout_residual.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/concat3_dim1.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/concat3_dim1.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/crf.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/crf.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/dropout.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/dropout.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/fuse_add2_op.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/fuse_add2_op.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/fuse_rotary_position_qkv.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/fuse_rotary_position_qkv.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/act_elewise_product.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/act_elewise_product.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/beam_search_topk.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/beam_search_topk.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/bias_act_dropout.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/bias_act_dropout.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/bias_add_transform_20314.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/bias_add_transform_20314.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/bias_dropout_residual.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/bias_dropout_residual.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/concat3_dim1.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/concat3_dim1.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/crf.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/crf.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/dropout.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/dropout.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/fuse_add2_op.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/fuse_add2_op.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/fuse_rotary_position_qkv.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/fuse_rotary_position_qkv.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/launch_dec_emb_op.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/launch_dec_emb_op.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/launch_enc_emb.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/launch_enc_emb.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/launch_gpt_emb.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/launch_gpt_emb.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/launch_llama_emb.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/launch_llama_emb.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/layer_normalize.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/layer_normalize.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/linear.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/linear.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/rms_layer_norm.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/rms_layer_norm.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/sampling.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/sampling.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/softmax.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/softmax.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/split_head_op.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/split_head_op.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/strided_batch_gemm.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/strided_batch_gemm.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/includes/transform_0213.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/includes/transform_0213.h -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/launch_dec_emb_op.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/launch_dec_emb_op.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/launch_enc_emb.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/launch_enc_emb.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/launch_gpt_emb.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/launch_gpt_emb.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/launch_llama_emb.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/launch_llama_emb.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/layer_normalize.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/layer_normalize.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/linear.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/linear.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/rms_layer_norm.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/rms_layer_norm.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/sampling.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/sampling.cc.cu -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/softmax.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/softmax.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/split_head_op.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/split_head_op.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/strided_batch_gemm.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/strided_batch_gemm.cpp -------------------------------------------------------------------------------- /lightseq/csrc/ops_new/transform_0213.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/ops_new/transform_0213.cpp -------------------------------------------------------------------------------- /lightseq/csrc/proto/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/proto/bert.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/bert.proto -------------------------------------------------------------------------------- /lightseq/csrc/proto/bert_crf.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/bert_crf.proto -------------------------------------------------------------------------------- /lightseq/csrc/proto/bert_crf_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/bert_crf_weight.cc -------------------------------------------------------------------------------- /lightseq/csrc/proto/bert_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/bert_weight.cc -------------------------------------------------------------------------------- /lightseq/csrc/proto/gpt.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/gpt.proto -------------------------------------------------------------------------------- /lightseq/csrc/proto/gpt_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/gpt_weight.cc -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/bert_crf_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/bert_crf_weight.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/bert_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/bert_weight.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/gpt_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/gpt_weight.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/hdf5_util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/hdf5_util.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/llama_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/llama_weight.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/proto_headers.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/proto_headers.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/proto_util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/proto_util.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/test_model_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/test_model_weight.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/includes/transformer_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/includes/transformer_weight.h -------------------------------------------------------------------------------- /lightseq/csrc/proto/llama.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/llama.proto -------------------------------------------------------------------------------- /lightseq/csrc/proto/llama_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/llama_weight.cc -------------------------------------------------------------------------------- /lightseq/csrc/proto/proto_util.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/proto_util.cc -------------------------------------------------------------------------------- /lightseq/csrc/proto/transformer.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/transformer.proto -------------------------------------------------------------------------------- /lightseq/csrc/proto/transformer_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/proto/transformer_weight.cc -------------------------------------------------------------------------------- /lightseq/csrc/pybind/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_adam.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_adam.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_kernel_cuda.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_kernel_cuda.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_kernel_x86.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_kernel_x86.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_layer.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_layer.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_layer_new.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_layer_new.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_model.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_model.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pybind/pybind_op.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pybind/pybind_op.cpp -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/builder/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/builder/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/builder/builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/builder/builder.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/builder/cuda_kernel_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/builder/cuda_kernel_builder.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/builder/cuda_layer_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/builder/cuda_layer_builder.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/builder/x86_kernel_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/builder/x86_kernel_builder.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/layer_base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/layer_base.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/calib/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/calib/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/calib/calibrator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/calib/calibrator.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/calib/histogram.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/calib/histogram.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/calib/max.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/calib/max.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/_functions/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/_functions/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/_functions/quant_rnn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/_functions/quant_rnn.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/functional.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/functional.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/_utils.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/clip.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/clip.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_bert.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_bert.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_conv.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_conv.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_instancenorm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_instancenorm.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_linear.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_linear.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_pooling.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_pooling.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_rnn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/quant_rnn.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/nn/modules/tensor_quantizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/nn/modules/tensor_quantizer.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/optim/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/optim/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/optim/helper.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/optim/helper.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/quant_modules.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/quant_modules.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/tensor_quant.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/tensor_quant.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/utils/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/utils/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/utils/quant_logging.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/utils/quant_logging.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/utils/reduce_amax.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/utils/reduce_amax.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/pytorch_quantization/version.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/pytorch_quantization/version.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/quantization.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/quantization.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/sdpa_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/sdpa_layers.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/torch_transformer_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/torch_transformer_layers.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/transformer_decoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/transformer_decoder_layer.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/transformer_encoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/transformer_encoder_layer.py -------------------------------------------------------------------------------- /lightseq/csrc/pytorch/util.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/pytorch/util.py -------------------------------------------------------------------------------- /lightseq/csrc/tensorflow/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tensorflow/README.md -------------------------------------------------------------------------------- /lightseq/csrc/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /lightseq/csrc/tests/cuda/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/cuda/__init__.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/cuda/fairseq_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/cuda/fairseq_layers.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/cuda/test_kernel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/cuda/test_kernel.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/cuda/test_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/cuda/test_layer.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/cuda/test_ls_small_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/cuda/test_ls_small_layer.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/cuda/torch_crf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/cuda/torch_crf.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/util.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/util.py -------------------------------------------------------------------------------- /lightseq/csrc/tests/x86/test_kernel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/tests/x86/test_kernel.py -------------------------------------------------------------------------------- /lightseq/csrc/triton_backend/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/triton_backend/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/csrc/triton_backend/cmake/TutorialMinimalBackendConfig.cmake.in: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/triton_backend/cmake/TutorialMinimalBackendConfig.cmake.in -------------------------------------------------------------------------------- /lightseq/csrc/triton_backend/src/libtriton_minimal.ldscript: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/triton_backend/src/libtriton_minimal.ldscript -------------------------------------------------------------------------------- /lightseq/csrc/triton_backend/src/lightseq_backend.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/triton_backend/src/lightseq_backend.cc -------------------------------------------------------------------------------- /lightseq/csrc/triton_backend/src/triton_model.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/triton_backend/src/triton_model.h -------------------------------------------------------------------------------- /lightseq/csrc/triton_backend/src/triton_utils.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/csrc/triton_backend/src/triton_utils.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/kernels/common.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/common.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/embKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/embKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/embKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/embKernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/embKernels_int8.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/embKernels_int8.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/embKernels_int8.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/embKernels_int8.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/gptKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/gptKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/gptKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/gptKernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/gptKernels_int8.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/gptKernels_int8.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/gptKernels_int8.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/gptKernels_int8.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/moeKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/moeKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/moeKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/moeKernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/multilgKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/multilgKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/multilgKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/multilgKernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/t5EmbKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/t5EmbKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/t5EmbKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/t5EmbKernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/t5Kernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/t5Kernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/t5Kernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/t5Kernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/transformerKernels.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/transformerKernels.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/transformerKernels.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/transformerKernels.h -------------------------------------------------------------------------------- /lightseq/inference/kernels/transformerKernels_int8.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/transformerKernels_int8.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/kernels/transformerKernels_int8.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/kernels/transformerKernels_int8.h -------------------------------------------------------------------------------- /lightseq/inference/model/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/model/bert_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/bert_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/bert_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/bert_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/cublas_algo_map.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/cublas_algo_map.cc -------------------------------------------------------------------------------- /lightseq/inference/model/cublas_algo_map.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/cublas_algo_map.h -------------------------------------------------------------------------------- /lightseq/inference/model/cublas_helper.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/cublas_helper.cc -------------------------------------------------------------------------------- /lightseq/inference/model/cublas_helper.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/cublas_helper.h -------------------------------------------------------------------------------- /lightseq/inference/model/decoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/decoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/decoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/decoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/gpt_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/gpt_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/gpt_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/gpt_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/moe_decoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/moe_decoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/moe_decoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/moe_decoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/moe_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/moe_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/moe_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/moe_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/mt5_decoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/mt5_decoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/mt5_decoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/mt5_decoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/mt5_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/mt5_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/mt5_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/mt5_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/quant_bert_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_bert_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/quant_bert_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_bert_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/quant_decoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_decoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/quant_decoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_decoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/quant_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/quant_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/quant_gpt_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_gpt_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/quant_gpt_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_gpt_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/quant_vit_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_vit_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/quant_vit_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/quant_vit_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/t5_decoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/t5_decoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/t5_decoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/t5_decoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/t5_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/t5_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/t5_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/t5_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/model/vit_encoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/vit_encoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/model/vit_encoder.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/model/vit_encoder.h -------------------------------------------------------------------------------- /lightseq/inference/proto/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/proto/bert.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/bert.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/bert_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/bert_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/bert_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/bert_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/gpt.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/gpt.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/gpt_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/gpt_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/gpt_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/gpt_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/moe.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/moe.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/moe_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/moe_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/moe_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/moe_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/mt5.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/mt5.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/mt5_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/mt5_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/mt5_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/mt5_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_bert.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_bert.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_bert_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_bert_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_bert_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_bert_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_gpt.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_gpt.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_gpt_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_gpt_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_gpt_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_gpt_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_transformer.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_transformer.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_transformer_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_transformer_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_transformer_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_transformer_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_vit.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_vit.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_vit_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_vit_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/quant_vit_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/quant_vit_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/t5.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/t5.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/t5_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/t5_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/t5_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/t5_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/transformer.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/transformer.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/transformer_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/transformer_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/transformer_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/transformer_weight.h -------------------------------------------------------------------------------- /lightseq/inference/proto/vit.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/vit.proto -------------------------------------------------------------------------------- /lightseq/inference/proto/vit_weight.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/vit_weight.cc -------------------------------------------------------------------------------- /lightseq/inference/proto/vit_weight.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/proto/vit_weight.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/bert.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/bert.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/bert.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/bert.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/gpt.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/gpt.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/gpt.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/gpt.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/model_base.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/model_base.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/moe.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/moe.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/moe.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/moe.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/mt5.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/mt5.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/mt5.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/mt5.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_bert.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_bert.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_bert.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_bert.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_gpt.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_gpt.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_gpt.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_gpt.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_transformer.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_transformer.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_transformer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_transformer.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_vit.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_vit.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/quant_vit.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/quant_vit.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/t5.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/t5.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/t5.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/t5.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/transformer.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/transformer.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/transformer.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/transformer.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/transformer_decoder.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/transformer_decoder.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/vit.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/vit.cc -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/vit.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/vit.h -------------------------------------------------------------------------------- /lightseq/inference/pywrapper/wrapper.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/pywrapper/wrapper.cc -------------------------------------------------------------------------------- /lightseq/inference/server/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/server/custom.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/custom.h -------------------------------------------------------------------------------- /lightseq/inference/server/decoder_generate_server.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/decoder_generate_server.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/server/generate_server.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/generate_server.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/server/gpt_generate_server.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/gpt_generate_server.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/server/gptlm_server.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/gptlm_server.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/server/libserver.ldscript: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/libserver.ldscript -------------------------------------------------------------------------------- /lightseq/inference/server/model_config.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/model_config.h -------------------------------------------------------------------------------- /lightseq/inference/server/model_config.proto: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/model_config.proto -------------------------------------------------------------------------------- /lightseq/inference/server/model_config_cuda.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/model_config_cuda.h -------------------------------------------------------------------------------- /lightseq/inference/server/moe_server.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/moe_server.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/server/transformer_server.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/server/transformer_server.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/tools/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/tools/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/tools/util.cc.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/tools/util.cc.cu -------------------------------------------------------------------------------- /lightseq/inference/tools/util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/tools/util.h -------------------------------------------------------------------------------- /lightseq/inference/triton_backend/CMakeLists.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/triton_backend/CMakeLists.txt -------------------------------------------------------------------------------- /lightseq/inference/triton_backend/cmake/TutorialMinimalBackendConfig.cmake.in: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/triton_backend/cmake/TutorialMinimalBackendConfig.cmake.in -------------------------------------------------------------------------------- /lightseq/inference/triton_backend/src/libtriton_minimal.ldscript: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/triton_backend/src/libtriton_minimal.ldscript -------------------------------------------------------------------------------- /lightseq/inference/triton_backend/src/lightseq_backend.cc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/triton_backend/src/lightseq_backend.cc -------------------------------------------------------------------------------- /lightseq/inference/triton_backend/src/triton_model.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/triton_backend/src/triton_model.h -------------------------------------------------------------------------------- /lightseq/inference/triton_backend/src/triton_utils.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/inference/triton_backend/src/triton_utils.h -------------------------------------------------------------------------------- /lightseq/training/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/__init__.py -------------------------------------------------------------------------------- /lightseq/training/cli/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/__init__.py -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/ls_adam.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/ls_adam.py -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/ls_bart.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/ls_bart.py -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/ls_fs_transformer_decoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/ls_fs_transformer_decoder_layer.py -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/ls_label_smoothed_cross_entropy.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/ls_label_smoothed_cross_entropy.py -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/ls_transformer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/ls_transformer.py -------------------------------------------------------------------------------- /lightseq/training/cli/fs_modules/ls_translation.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/fs_modules/ls_translation.py -------------------------------------------------------------------------------- /lightseq/training/cli/lightseq_deepspeed_cli.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/lightseq_deepspeed_cli.py -------------------------------------------------------------------------------- /lightseq/training/cli/lightseq_fairseq_generate_cli.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/lightseq_fairseq_generate_cli.py -------------------------------------------------------------------------------- /lightseq/training/cli/lightseq_fairseq_train_cli.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/lightseq_fairseq_train_cli.py -------------------------------------------------------------------------------- /lightseq/training/cli/lightseq_fairseq_validate_cli.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/lightseq_fairseq_validate_cli.py -------------------------------------------------------------------------------- /lightseq/training/cli/lightseq_infer_cli.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/cli/lightseq_infer_cli.py -------------------------------------------------------------------------------- /lightseq/training/csrc/ops/includes/strided_batch_gemm.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/csrc/ops/includes/strided_batch_gemm.h -------------------------------------------------------------------------------- /lightseq/training/gcq/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/gcq/__init__.py -------------------------------------------------------------------------------- /lightseq/training/gcq/gcq.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/gcq/gcq.py -------------------------------------------------------------------------------- /lightseq/training/gcq/ls_fs_gcq_train.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/gcq/ls_fs_gcq_train.py -------------------------------------------------------------------------------- /lightseq/training/gcq/ls_fs_gcq_trainer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/gcq/ls_fs_gcq_trainer.py -------------------------------------------------------------------------------- /lightseq/training/ops/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/__init__.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/adam.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/adam.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/__init__.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/adam_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/adam_builder.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/builder.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/kernel_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/kernel_builder.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/layer_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/layer_builder.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/operator_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/operator_builder.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/builder/transformer_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/builder/transformer_builder.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/cross_entropy_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/cross_entropy_layer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/export.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/export.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/export_quant.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/export_quant.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/gemm_test.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/gemm_test.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/gpt_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/gpt_layer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/layer_base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/layer_base.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/quant_linear_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/quant_linear_layer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/quantization.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/quantization.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/torch_transformer_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/torch_transformer_layers.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/transformer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/transformer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/transformer_decoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/transformer_decoder_layer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/transformer_decoder_layer_new.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/transformer_decoder_layer_new.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/transformer_embedding_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/transformer_embedding_layer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/transformer_encoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/transformer_encoder_layer.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/transformer_encoder_layer_new.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/transformer_encoder_layer_new.py -------------------------------------------------------------------------------- /lightseq/training/ops/pytorch/util.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/pytorch/util.py -------------------------------------------------------------------------------- /lightseq/training/ops/tensorflow/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/ops/tensorflow/README.md -------------------------------------------------------------------------------- /lightseq/training/ops/tensorflow/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/calib/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/calib/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/calib/calibrator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/calib/calibrator.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/calib/histogram.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/calib/histogram.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/calib/max.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/calib/max.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/_functions/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/_functions/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/_functions/quant_rnn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/_functions/quant_rnn.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/functional.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/functional.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/_utils.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/clip.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/clip.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/quant_bert.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/quant_bert.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/quant_conv.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/quant_conv.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/quant_instancenorm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/quant_instancenorm.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/quant_linear.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/quant_linear.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/quant_pooling.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/quant_pooling.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/quant_rnn.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/quant_rnn.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/nn/modules/tensor_quantizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/nn/modules/tensor_quantizer.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/optim/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/optim/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/optim/helper.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/optim/helper.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/quant_modules.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/quant_modules.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/tensor_quant.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/tensor_quant.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/utils/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/utils/__init__.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/utils/quant_logging.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/utils/quant_logging.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/utils/reduce_amax.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/utils/reduce_amax.py -------------------------------------------------------------------------------- /lightseq/training/pytorch_quantization/version.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/lightseq/training/pytorch_quantization/version.py -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/setup.py -------------------------------------------------------------------------------- /tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tests/cublas/build.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/cublas/build.sh -------------------------------------------------------------------------------- /tests/cublas/gemm.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/cublas/gemm.cu -------------------------------------------------------------------------------- /tests/cublas/gemm.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/cublas/gemm.h -------------------------------------------------------------------------------- /tests/cublas/test.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/cublas/test.cpp -------------------------------------------------------------------------------- /tests/cublas/util.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/cublas/util.h -------------------------------------------------------------------------------- /tests/fairseq_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/fairseq_layers.py -------------------------------------------------------------------------------- /tests/gemm_test/configs/igemm_A10.cfg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/gemm_test/configs/igemm_A10.cfg -------------------------------------------------------------------------------- /tests/gemm_test/configs/igemm_A100.cfg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/gemm_test/configs/igemm_A100.cfg -------------------------------------------------------------------------------- /tests/gemm_test/configs/igemm_A30.cfg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/gemm_test/configs/igemm_A30.cfg -------------------------------------------------------------------------------- /tests/gemm_test/configs/igemm_T4.cfg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/gemm_test/configs/igemm_T4.cfg -------------------------------------------------------------------------------- /tests/gemm_test/gemm_test.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/gemm_test/gemm_test.py -------------------------------------------------------------------------------- /tests/gemm_test/gemm_test.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/gemm_test/gemm_test.sh -------------------------------------------------------------------------------- /tests/huggingface/test_gpt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/huggingface/test_gpt.py -------------------------------------------------------------------------------- /tests/quant_test.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/quant_test.py -------------------------------------------------------------------------------- /tests/test_ls_kernels.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/test_ls_kernels.py -------------------------------------------------------------------------------- /tests/test_ls_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/test_ls_layers.py -------------------------------------------------------------------------------- /tests/test_ls_layers_new.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/test_ls_layers_new.py -------------------------------------------------------------------------------- /tests/test_ls_ops.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/test_ls_ops.py -------------------------------------------------------------------------------- /tests/torch_crf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/torch_crf.py -------------------------------------------------------------------------------- /tests/tvm/matmul_autotune.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/tvm/matmul_autotune.py -------------------------------------------------------------------------------- /tests/util.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/bytedance/lightseq/HEAD/tests/util.py --------------------------------------------------------------------------------