├── .github └── workflows │ ├── ci.md │ └── main.yml ├── .gitignore ├── .gitlab-ci.yml ├── LICENSE ├── MANIFEST.in ├── Makefile ├── README.md ├── config_tasks ├── task_boolq.sh ├── task_cb.sh ├── task_copa.sh ├── task_multirc.sh ├── task_record.sh ├── task_rte.sh ├── task_wic.sh ├── task_wsc.sh └── task_wsc_generative.sh ├── configs └── glm-130b │ └── glm-130b.sh ├── evaluation ├── datasets.py ├── evaluate.py ├── main.py ├── metrics.py └── utils.py ├── examples ├── create_embeddings.sh ├── curriculum_learning │ ├── README.md │ ├── ds_config_cl.json │ └── pretrain_gpt_cl.sh ├── evaluate_ict_zeroshot_nq.sh ├── evaluate_zeroshot_gpt.sh ├── finetune_mnli_distributed.sh ├── finetune_race_distributed.sh ├── generate_text.sh ├── merge_mp_bert.sh ├── pretrain_bert.sh ├── pretrain_bert_distributed.sh ├── pretrain_bert_distributed_with_mp.sh ├── pretrain_gpt.sh ├── pretrain_gpt3_175B.sh ├── pretrain_gpt_distributed.sh ├── pretrain_gpt_distributed_with_mp.sh ├── pretrain_gpt_multilingual.sh ├── pretrain_gpt_single_node.sh ├── pretrain_gpt_tiny.sh ├── pretrain_ict.sh ├── pretrain_t5.sh ├── pretrain_t5_distributed.sh └── pretrain_t5_distributed_with_mp.sh ├── glm ├── __init__.py ├── collator.py ├── data.py ├── datasets.py ├── generate.py ├── generation │ └── utils.py └── model.py ├── images └── cases_april2021.png ├── megatron ├── __init__.py ├── arguments.py ├── checkpointing.py ├── data │ ├── Makefile │ ├── __init__.py │ ├── autoaugment.py │ ├── bert_dataset.py │ ├── biencoder_dataset_utils.py │ ├── blendable_dataset.py │ ├── data_samplers.py │ ├── dataset_utils.py │ ├── distdata.py │ ├── gpt_dataset.py │ ├── helpers.cpp │ ├── ict_dataset.py │ ├── indexed_dataset.py │ ├── orqa_wiki_dataset.py │ ├── realm_dataset_utils.py │ ├── realm_index.py │ ├── t5_dataset.py │ ├── test │ │ ├── test_indexed_dataset.py │ │ └── test_preprocess_data.sh │ └── vit_dataset.py ├── enums.py ├── fp16_deprecated │ └── loss_scaler.py ├── fused_kernels │ ├── __init__.py │ ├── compat.h │ ├── layer_norm_cuda.cpp │ ├── layer_norm_cuda_kernel.cu │ ├── rotary_positional_embedding.cpp │ ├── rotary_positional_embedding.cu │ ├── scaled_masked_softmax.cpp │ ├── scaled_masked_softmax.h │ ├── scaled_masked_softmax_cuda.cu │ ├── scaled_upper_triang_masked_softmax.cpp │ ├── scaled_upper_triang_masked_softmax.h │ ├── scaled_upper_triang_masked_softmax_cuda.cu │ ├── tests │ │ ├── __init__.py │ │ └── test_fused_kernels.py │ └── type_shim.h ├── global_vars.py ├── indexer.py ├── initialize.py ├── learning_rates.py ├── logging.py ├── memory.py ├── microbatches.py ├── model │ ├── __init__.py │ ├── bert_model.py │ ├── biencoder_model.py │ ├── classification.py │ ├── distributed.py │ ├── fused_bias_gelu.py │ ├── fused_layer_norm.py │ ├── fused_softmax.py │ ├── gau.py │ ├── glu_activations.py │ ├── gpt_model.py │ ├── language_model.py │ ├── module.py │ ├── multiple_choice.py │ ├── positional_embeddings.py │ ├── realm_model.py │ ├── t5_model.py │ ├── transformer.py │ ├── utils.py │ └── vit_model.py ├── mpu │ ├── __init__.py │ ├── cross_entropy.py │ ├── data.py │ ├── initialize.py │ ├── layers.py │ ├── mappings.py │ ├── random.py │ ├── tests │ │ ├── __init__.py │ │ ├── commons.py │ │ ├── test_cross_entropy.py │ │ ├── test_data.py │ │ ├── test_initialize.py │ │ ├── test_layers.py │ │ └── test_random.py │ └── utils.py ├── optimizer │ ├── __init__.py │ ├── clip_grads.py │ ├── grad_scaler.py │ └── optimizer.py ├── p2p_communication.py ├── package_info.py ├── schedules.py ├── testing_utils.py ├── text_generation_utils.py ├── tokenizer │ ├── __init__.py │ ├── bert_tokenization.py │ ├── gpt2_tokenization.py │ ├── ice_tokenizer.py │ └── tokenizer.py ├── training.py └── utils.py ├── patch ├── pipe_engine.py └── pipe_engine_0.6.5.py ├── pretrain_bert.py ├── pretrain_glm.py ├── pretrain_gpt.py ├── pretrain_ict.py ├── pretrain_prefix_lm.py ├── pretrain_t5.py ├── pretrain_vit.py ├── pyproject.toml ├── requirements.txt ├── run.sh ├── run_bf16.sh ├── run_fp16.sh ├── scripts ├── finetune │ ├── grid-record.sh │ ├── grid-rte.sh │ ├── grid-wsc.sh │ └── superglue.sh ├── generate │ └── benchmark.sh ├── p-tuning │ ├── grid-boolq.sh │ ├── grid-cb.sh │ ├── grid-multirc.sh │ ├── grid-rte.sh │ ├── grid-wic.sh │ ├── grid-wsc.sh │ └── superglue.sh └── submit_gpu.sh ├── setup.cfg ├── setup.py ├── tasks ├── data_utils.py ├── ensemble_classifier.py ├── eval_utils.py ├── finetune_utils.py ├── glue │ ├── data.py │ ├── finetune.py │ ├── mnli.py │ └── qqp.py ├── main.py ├── orqa │ ├── evaluate_orqa.py │ ├── evaluate_utils.py │ └── natural_questions │ │ ├── nq.py │ │ ├── qa_utils.py │ │ └── tokenizers.py ├── race │ ├── data.py │ └── finetune.py ├── superglue │ ├── data_utils.py │ ├── dataset.py │ ├── eval_utils.py │ ├── evaluate.py │ ├── finetune.py │ └── pvp.py ├── vision │ ├── classification.py │ ├── eval_utils.py │ ├── finetune_utils.py │ └── main.py ├── zeroshot_glm │ ├── datasets.py │ └── evaluate.py └── zeroshot_gpt │ ├── datasets.py │ ├── detokenizer.py │ └── evaluate.py ├── tests ├── README.md ├── conftest.py ├── data │ └── gpt2 │ │ ├── gpt2-tiny-merges.txt │ │ ├── gpt2-tiny-vocab.json │ │ ├── meg-gpt2-openwebtext_text_document.bin │ │ ├── meg-gpt2-openwebtext_text_document.idx │ │ └── openwebtext-1000.jsonl ├── ds_config.json ├── ds_config_cl.json ├── test_activations.py ├── test_basic.py ├── test_model.py ├── test_preprocessing.py ├── test_rotary.py ├── test_tensor_parallel.py ├── test_training.py └── tools │ ├── README.md │ ├── openwebtext-to-jsonl.py │ └── shrink-tokenizer.py └── tools ├── README.md ├── convert_checkpoint ├── README.md ├── deepspeed_checkpoint.py ├── deepspeed_to_megatron.py ├── deepspeed_to_transformers.py ├── inspect_checkpoint.py └── inspect_deepspeed_checkpoint.py ├── create_doc_index.py ├── generate_samples_gpt.py ├── linter.py ├── logs ├── rescale-logs.py ├── tb-rename-events.py └── tb-rescale-scalars.py ├── merge_mp_partitions.py ├── merge_preprocessed_data.py ├── openwebtext ├── README.md ├── add_id.py ├── blacklist_urls.py ├── cleanup_dataset.py ├── cleanup_fix_dataset.py ├── filter_ngrams.py ├── find_duplicates.py ├── group_duplicate_url.py ├── merge_jsons.py └── remove_group_duplicates.py ├── preprocess_data.py ├── preprocess_data_dist.py ├── preprocess_data_many_cores.py ├── sample_idxs_to_text.py └── tb ├── tb-remove-events-by-group.py ├── tb-remove-events-by-tag.py └── tb-rename-events.py /.github/workflows/ci.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/.github/workflows/ci.md -------------------------------------------------------------------------------- /.github/workflows/main.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/.github/workflows/main.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/.gitignore -------------------------------------------------------------------------------- /.gitlab-ci.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/.gitlab-ci.yml -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/LICENSE -------------------------------------------------------------------------------- /MANIFEST.in: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/MANIFEST.in -------------------------------------------------------------------------------- /Makefile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/Makefile -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/README.md -------------------------------------------------------------------------------- /config_tasks/task_boolq.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_boolq.sh -------------------------------------------------------------------------------- /config_tasks/task_cb.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_cb.sh -------------------------------------------------------------------------------- /config_tasks/task_copa.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_copa.sh -------------------------------------------------------------------------------- /config_tasks/task_multirc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_multirc.sh -------------------------------------------------------------------------------- /config_tasks/task_record.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_record.sh -------------------------------------------------------------------------------- /config_tasks/task_rte.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_rte.sh -------------------------------------------------------------------------------- /config_tasks/task_wic.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_wic.sh -------------------------------------------------------------------------------- /config_tasks/task_wsc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_wsc.sh -------------------------------------------------------------------------------- /config_tasks/task_wsc_generative.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/config_tasks/task_wsc_generative.sh -------------------------------------------------------------------------------- /configs/glm-130b/glm-130b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/configs/glm-130b/glm-130b.sh -------------------------------------------------------------------------------- /evaluation/datasets.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/evaluation/datasets.py -------------------------------------------------------------------------------- /evaluation/evaluate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/evaluation/evaluate.py -------------------------------------------------------------------------------- /evaluation/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/evaluation/main.py -------------------------------------------------------------------------------- /evaluation/metrics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/evaluation/metrics.py -------------------------------------------------------------------------------- /evaluation/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/evaluation/utils.py -------------------------------------------------------------------------------- /examples/create_embeddings.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/create_embeddings.sh -------------------------------------------------------------------------------- /examples/curriculum_learning/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/curriculum_learning/README.md -------------------------------------------------------------------------------- /examples/curriculum_learning/ds_config_cl.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/curriculum_learning/ds_config_cl.json -------------------------------------------------------------------------------- /examples/curriculum_learning/pretrain_gpt_cl.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/curriculum_learning/pretrain_gpt_cl.sh -------------------------------------------------------------------------------- /examples/evaluate_ict_zeroshot_nq.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/evaluate_ict_zeroshot_nq.sh -------------------------------------------------------------------------------- /examples/evaluate_zeroshot_gpt.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/evaluate_zeroshot_gpt.sh -------------------------------------------------------------------------------- /examples/finetune_mnli_distributed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/finetune_mnli_distributed.sh -------------------------------------------------------------------------------- /examples/finetune_race_distributed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/finetune_race_distributed.sh -------------------------------------------------------------------------------- /examples/generate_text.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/generate_text.sh -------------------------------------------------------------------------------- /examples/merge_mp_bert.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/merge_mp_bert.sh -------------------------------------------------------------------------------- /examples/pretrain_bert.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_bert.sh -------------------------------------------------------------------------------- /examples/pretrain_bert_distributed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_bert_distributed.sh -------------------------------------------------------------------------------- /examples/pretrain_bert_distributed_with_mp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_bert_distributed_with_mp.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt3_175B.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt3_175B.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt_distributed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt_distributed.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt_distributed_with_mp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt_distributed_with_mp.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt_multilingual.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt_multilingual.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt_single_node.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt_single_node.sh -------------------------------------------------------------------------------- /examples/pretrain_gpt_tiny.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_gpt_tiny.sh -------------------------------------------------------------------------------- /examples/pretrain_ict.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_ict.sh -------------------------------------------------------------------------------- /examples/pretrain_t5.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_t5.sh -------------------------------------------------------------------------------- /examples/pretrain_t5_distributed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_t5_distributed.sh -------------------------------------------------------------------------------- /examples/pretrain_t5_distributed_with_mp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/examples/pretrain_t5_distributed_with_mp.sh -------------------------------------------------------------------------------- /glm/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/__init__.py -------------------------------------------------------------------------------- /glm/collator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/collator.py -------------------------------------------------------------------------------- /glm/data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/data.py -------------------------------------------------------------------------------- /glm/datasets.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/datasets.py -------------------------------------------------------------------------------- /glm/generate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/generate.py -------------------------------------------------------------------------------- /glm/generation/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/generation/utils.py -------------------------------------------------------------------------------- /glm/model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/glm/model.py -------------------------------------------------------------------------------- /images/cases_april2021.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/images/cases_april2021.png -------------------------------------------------------------------------------- /megatron/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/__init__.py -------------------------------------------------------------------------------- /megatron/arguments.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/arguments.py -------------------------------------------------------------------------------- /megatron/checkpointing.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/checkpointing.py -------------------------------------------------------------------------------- /megatron/data/Makefile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/Makefile -------------------------------------------------------------------------------- /megatron/data/__init__.py: -------------------------------------------------------------------------------- 1 | from . import indexed_dataset 2 | -------------------------------------------------------------------------------- /megatron/data/autoaugment.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/autoaugment.py -------------------------------------------------------------------------------- /megatron/data/bert_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/bert_dataset.py -------------------------------------------------------------------------------- /megatron/data/biencoder_dataset_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/biencoder_dataset_utils.py -------------------------------------------------------------------------------- /megatron/data/blendable_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/blendable_dataset.py -------------------------------------------------------------------------------- /megatron/data/data_samplers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/data_samplers.py -------------------------------------------------------------------------------- /megatron/data/dataset_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/dataset_utils.py -------------------------------------------------------------------------------- /megatron/data/distdata.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/distdata.py -------------------------------------------------------------------------------- /megatron/data/gpt_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/gpt_dataset.py -------------------------------------------------------------------------------- /megatron/data/helpers.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/helpers.cpp -------------------------------------------------------------------------------- /megatron/data/ict_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/ict_dataset.py -------------------------------------------------------------------------------- /megatron/data/indexed_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/indexed_dataset.py -------------------------------------------------------------------------------- /megatron/data/orqa_wiki_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/orqa_wiki_dataset.py -------------------------------------------------------------------------------- /megatron/data/realm_dataset_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/realm_dataset_utils.py -------------------------------------------------------------------------------- /megatron/data/realm_index.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/realm_index.py -------------------------------------------------------------------------------- /megatron/data/t5_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/t5_dataset.py -------------------------------------------------------------------------------- /megatron/data/test/test_indexed_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/test/test_indexed_dataset.py -------------------------------------------------------------------------------- /megatron/data/test/test_preprocess_data.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/test/test_preprocess_data.sh -------------------------------------------------------------------------------- /megatron/data/vit_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/data/vit_dataset.py -------------------------------------------------------------------------------- /megatron/enums.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/enums.py -------------------------------------------------------------------------------- /megatron/fp16_deprecated/loss_scaler.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fp16_deprecated/loss_scaler.py -------------------------------------------------------------------------------- /megatron/fused_kernels/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/__init__.py -------------------------------------------------------------------------------- /megatron/fused_kernels/compat.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/compat.h -------------------------------------------------------------------------------- /megatron/fused_kernels/layer_norm_cuda.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/layer_norm_cuda.cpp -------------------------------------------------------------------------------- /megatron/fused_kernels/layer_norm_cuda_kernel.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/layer_norm_cuda_kernel.cu -------------------------------------------------------------------------------- /megatron/fused_kernels/rotary_positional_embedding.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/rotary_positional_embedding.cpp -------------------------------------------------------------------------------- /megatron/fused_kernels/rotary_positional_embedding.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/rotary_positional_embedding.cu -------------------------------------------------------------------------------- /megatron/fused_kernels/scaled_masked_softmax.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/scaled_masked_softmax.cpp -------------------------------------------------------------------------------- /megatron/fused_kernels/scaled_masked_softmax.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/scaled_masked_softmax.h -------------------------------------------------------------------------------- /megatron/fused_kernels/scaled_masked_softmax_cuda.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/scaled_masked_softmax_cuda.cu -------------------------------------------------------------------------------- /megatron/fused_kernels/scaled_upper_triang_masked_softmax.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/scaled_upper_triang_masked_softmax.cpp -------------------------------------------------------------------------------- /megatron/fused_kernels/scaled_upper_triang_masked_softmax.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/scaled_upper_triang_masked_softmax.h -------------------------------------------------------------------------------- /megatron/fused_kernels/scaled_upper_triang_masked_softmax_cuda.cu: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/scaled_upper_triang_masked_softmax_cuda.cu -------------------------------------------------------------------------------- /megatron/fused_kernels/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /megatron/fused_kernels/tests/test_fused_kernels.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/tests/test_fused_kernels.py -------------------------------------------------------------------------------- /megatron/fused_kernels/type_shim.h: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/fused_kernels/type_shim.h -------------------------------------------------------------------------------- /megatron/global_vars.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/global_vars.py -------------------------------------------------------------------------------- /megatron/indexer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/indexer.py -------------------------------------------------------------------------------- /megatron/initialize.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/initialize.py -------------------------------------------------------------------------------- /megatron/learning_rates.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/learning_rates.py -------------------------------------------------------------------------------- /megatron/logging.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/logging.py -------------------------------------------------------------------------------- /megatron/memory.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/memory.py -------------------------------------------------------------------------------- /megatron/microbatches.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/microbatches.py -------------------------------------------------------------------------------- /megatron/model/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/__init__.py -------------------------------------------------------------------------------- /megatron/model/bert_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/bert_model.py -------------------------------------------------------------------------------- /megatron/model/biencoder_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/biencoder_model.py -------------------------------------------------------------------------------- /megatron/model/classification.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/classification.py -------------------------------------------------------------------------------- /megatron/model/distributed.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/distributed.py -------------------------------------------------------------------------------- /megatron/model/fused_bias_gelu.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/fused_bias_gelu.py -------------------------------------------------------------------------------- /megatron/model/fused_layer_norm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/fused_layer_norm.py -------------------------------------------------------------------------------- /megatron/model/fused_softmax.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/fused_softmax.py -------------------------------------------------------------------------------- /megatron/model/gau.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/gau.py -------------------------------------------------------------------------------- /megatron/model/glu_activations.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/glu_activations.py -------------------------------------------------------------------------------- /megatron/model/gpt_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/gpt_model.py -------------------------------------------------------------------------------- /megatron/model/language_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/language_model.py -------------------------------------------------------------------------------- /megatron/model/module.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/module.py -------------------------------------------------------------------------------- /megatron/model/multiple_choice.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/multiple_choice.py -------------------------------------------------------------------------------- /megatron/model/positional_embeddings.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/positional_embeddings.py -------------------------------------------------------------------------------- /megatron/model/realm_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/realm_model.py -------------------------------------------------------------------------------- /megatron/model/t5_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/t5_model.py -------------------------------------------------------------------------------- /megatron/model/transformer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/transformer.py -------------------------------------------------------------------------------- /megatron/model/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/utils.py -------------------------------------------------------------------------------- /megatron/model/vit_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/model/vit_model.py -------------------------------------------------------------------------------- /megatron/mpu/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/__init__.py -------------------------------------------------------------------------------- /megatron/mpu/cross_entropy.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/cross_entropy.py -------------------------------------------------------------------------------- /megatron/mpu/data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/data.py -------------------------------------------------------------------------------- /megatron/mpu/initialize.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/initialize.py -------------------------------------------------------------------------------- /megatron/mpu/layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/layers.py -------------------------------------------------------------------------------- /megatron/mpu/mappings.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/mappings.py -------------------------------------------------------------------------------- /megatron/mpu/random.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/random.py -------------------------------------------------------------------------------- /megatron/mpu/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /megatron/mpu/tests/commons.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/tests/commons.py -------------------------------------------------------------------------------- /megatron/mpu/tests/test_cross_entropy.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/tests/test_cross_entropy.py -------------------------------------------------------------------------------- /megatron/mpu/tests/test_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/tests/test_data.py -------------------------------------------------------------------------------- /megatron/mpu/tests/test_initialize.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/tests/test_initialize.py -------------------------------------------------------------------------------- /megatron/mpu/tests/test_layers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/tests/test_layers.py -------------------------------------------------------------------------------- /megatron/mpu/tests/test_random.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/tests/test_random.py -------------------------------------------------------------------------------- /megatron/mpu/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/mpu/utils.py -------------------------------------------------------------------------------- /megatron/optimizer/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/optimizer/__init__.py -------------------------------------------------------------------------------- /megatron/optimizer/clip_grads.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/optimizer/clip_grads.py -------------------------------------------------------------------------------- /megatron/optimizer/grad_scaler.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/optimizer/grad_scaler.py -------------------------------------------------------------------------------- /megatron/optimizer/optimizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/optimizer/optimizer.py -------------------------------------------------------------------------------- /megatron/p2p_communication.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/p2p_communication.py -------------------------------------------------------------------------------- /megatron/package_info.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/package_info.py -------------------------------------------------------------------------------- /megatron/schedules.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/schedules.py -------------------------------------------------------------------------------- /megatron/testing_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/testing_utils.py -------------------------------------------------------------------------------- /megatron/text_generation_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/text_generation_utils.py -------------------------------------------------------------------------------- /megatron/tokenizer/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/tokenizer/__init__.py -------------------------------------------------------------------------------- /megatron/tokenizer/bert_tokenization.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/tokenizer/bert_tokenization.py -------------------------------------------------------------------------------- /megatron/tokenizer/gpt2_tokenization.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/tokenizer/gpt2_tokenization.py -------------------------------------------------------------------------------- /megatron/tokenizer/ice_tokenizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/tokenizer/ice_tokenizer.py -------------------------------------------------------------------------------- /megatron/tokenizer/tokenizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/tokenizer/tokenizer.py -------------------------------------------------------------------------------- /megatron/training.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/training.py -------------------------------------------------------------------------------- /megatron/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/megatron/utils.py -------------------------------------------------------------------------------- /patch/pipe_engine.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/patch/pipe_engine.py -------------------------------------------------------------------------------- /patch/pipe_engine_0.6.5.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/patch/pipe_engine_0.6.5.py -------------------------------------------------------------------------------- /pretrain_bert.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_bert.py -------------------------------------------------------------------------------- /pretrain_glm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_glm.py -------------------------------------------------------------------------------- /pretrain_gpt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_gpt.py -------------------------------------------------------------------------------- /pretrain_ict.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_ict.py -------------------------------------------------------------------------------- /pretrain_prefix_lm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_prefix_lm.py -------------------------------------------------------------------------------- /pretrain_t5.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_t5.py -------------------------------------------------------------------------------- /pretrain_vit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/pretrain_vit.py -------------------------------------------------------------------------------- /pyproject.toml: -------------------------------------------------------------------------------- 1 | [tool.black] 2 | line-length = 119 3 | target-version = ['py35'] -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/requirements.txt -------------------------------------------------------------------------------- /run.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/run.sh -------------------------------------------------------------------------------- /run_bf16.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/run_bf16.sh -------------------------------------------------------------------------------- /run_fp16.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/run_fp16.sh -------------------------------------------------------------------------------- /scripts/finetune/grid-record.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/finetune/grid-record.sh -------------------------------------------------------------------------------- /scripts/finetune/grid-rte.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/finetune/grid-rte.sh -------------------------------------------------------------------------------- /scripts/finetune/grid-wsc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/finetune/grid-wsc.sh -------------------------------------------------------------------------------- /scripts/finetune/superglue.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/finetune/superglue.sh -------------------------------------------------------------------------------- /scripts/generate/benchmark.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/generate/benchmark.sh -------------------------------------------------------------------------------- /scripts/p-tuning/grid-boolq.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/grid-boolq.sh -------------------------------------------------------------------------------- /scripts/p-tuning/grid-cb.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/grid-cb.sh -------------------------------------------------------------------------------- /scripts/p-tuning/grid-multirc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/grid-multirc.sh -------------------------------------------------------------------------------- /scripts/p-tuning/grid-rte.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/grid-rte.sh -------------------------------------------------------------------------------- /scripts/p-tuning/grid-wic.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/grid-wic.sh -------------------------------------------------------------------------------- /scripts/p-tuning/grid-wsc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/grid-wsc.sh -------------------------------------------------------------------------------- /scripts/p-tuning/superglue.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/p-tuning/superglue.sh -------------------------------------------------------------------------------- /scripts/submit_gpu.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/scripts/submit_gpu.sh -------------------------------------------------------------------------------- /setup.cfg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/setup.cfg -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/setup.py -------------------------------------------------------------------------------- /tasks/data_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/data_utils.py -------------------------------------------------------------------------------- /tasks/ensemble_classifier.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/ensemble_classifier.py -------------------------------------------------------------------------------- /tasks/eval_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/eval_utils.py -------------------------------------------------------------------------------- /tasks/finetune_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/finetune_utils.py -------------------------------------------------------------------------------- /tasks/glue/data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/glue/data.py -------------------------------------------------------------------------------- /tasks/glue/finetune.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/glue/finetune.py -------------------------------------------------------------------------------- /tasks/glue/mnli.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/glue/mnli.py -------------------------------------------------------------------------------- /tasks/glue/qqp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/glue/qqp.py -------------------------------------------------------------------------------- /tasks/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/main.py -------------------------------------------------------------------------------- /tasks/orqa/evaluate_orqa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/orqa/evaluate_orqa.py -------------------------------------------------------------------------------- /tasks/orqa/evaluate_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/orqa/evaluate_utils.py -------------------------------------------------------------------------------- /tasks/orqa/natural_questions/nq.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/orqa/natural_questions/nq.py -------------------------------------------------------------------------------- /tasks/orqa/natural_questions/qa_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/orqa/natural_questions/qa_utils.py -------------------------------------------------------------------------------- /tasks/orqa/natural_questions/tokenizers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/orqa/natural_questions/tokenizers.py -------------------------------------------------------------------------------- /tasks/race/data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/race/data.py -------------------------------------------------------------------------------- /tasks/race/finetune.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/race/finetune.py -------------------------------------------------------------------------------- /tasks/superglue/data_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/superglue/data_utils.py -------------------------------------------------------------------------------- /tasks/superglue/dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/superglue/dataset.py -------------------------------------------------------------------------------- /tasks/superglue/eval_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/superglue/eval_utils.py -------------------------------------------------------------------------------- /tasks/superglue/evaluate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/superglue/evaluate.py -------------------------------------------------------------------------------- /tasks/superglue/finetune.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/superglue/finetune.py -------------------------------------------------------------------------------- /tasks/superglue/pvp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/superglue/pvp.py -------------------------------------------------------------------------------- /tasks/vision/classification.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/vision/classification.py -------------------------------------------------------------------------------- /tasks/vision/eval_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/vision/eval_utils.py -------------------------------------------------------------------------------- /tasks/vision/finetune_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/vision/finetune_utils.py -------------------------------------------------------------------------------- /tasks/vision/main.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/vision/main.py -------------------------------------------------------------------------------- /tasks/zeroshot_glm/datasets.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/zeroshot_glm/datasets.py -------------------------------------------------------------------------------- /tasks/zeroshot_glm/evaluate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/zeroshot_glm/evaluate.py -------------------------------------------------------------------------------- /tasks/zeroshot_gpt/datasets.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/zeroshot_gpt/datasets.py -------------------------------------------------------------------------------- /tasks/zeroshot_gpt/detokenizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/zeroshot_gpt/detokenizer.py -------------------------------------------------------------------------------- /tasks/zeroshot_gpt/evaluate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tasks/zeroshot_gpt/evaluate.py -------------------------------------------------------------------------------- /tests/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/README.md -------------------------------------------------------------------------------- /tests/conftest.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/conftest.py -------------------------------------------------------------------------------- /tests/data/gpt2/gpt2-tiny-merges.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/data/gpt2/gpt2-tiny-merges.txt -------------------------------------------------------------------------------- /tests/data/gpt2/gpt2-tiny-vocab.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/data/gpt2/gpt2-tiny-vocab.json -------------------------------------------------------------------------------- /tests/data/gpt2/meg-gpt2-openwebtext_text_document.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/data/gpt2/meg-gpt2-openwebtext_text_document.bin -------------------------------------------------------------------------------- /tests/data/gpt2/meg-gpt2-openwebtext_text_document.idx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/data/gpt2/meg-gpt2-openwebtext_text_document.idx -------------------------------------------------------------------------------- /tests/data/gpt2/openwebtext-1000.jsonl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/data/gpt2/openwebtext-1000.jsonl -------------------------------------------------------------------------------- /tests/ds_config.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/ds_config.json -------------------------------------------------------------------------------- /tests/ds_config_cl.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/ds_config_cl.json -------------------------------------------------------------------------------- /tests/test_activations.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_activations.py -------------------------------------------------------------------------------- /tests/test_basic.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_basic.py -------------------------------------------------------------------------------- /tests/test_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_model.py -------------------------------------------------------------------------------- /tests/test_preprocessing.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_preprocessing.py -------------------------------------------------------------------------------- /tests/test_rotary.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_rotary.py -------------------------------------------------------------------------------- /tests/test_tensor_parallel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_tensor_parallel.py -------------------------------------------------------------------------------- /tests/test_training.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/test_training.py -------------------------------------------------------------------------------- /tests/tools/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/tools/README.md -------------------------------------------------------------------------------- /tests/tools/openwebtext-to-jsonl.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/tools/openwebtext-to-jsonl.py -------------------------------------------------------------------------------- /tests/tools/shrink-tokenizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tests/tools/shrink-tokenizer.py -------------------------------------------------------------------------------- /tools/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/README.md -------------------------------------------------------------------------------- /tools/convert_checkpoint/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/convert_checkpoint/README.md -------------------------------------------------------------------------------- /tools/convert_checkpoint/deepspeed_checkpoint.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/convert_checkpoint/deepspeed_checkpoint.py -------------------------------------------------------------------------------- /tools/convert_checkpoint/deepspeed_to_megatron.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/convert_checkpoint/deepspeed_to_megatron.py -------------------------------------------------------------------------------- /tools/convert_checkpoint/deepspeed_to_transformers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/convert_checkpoint/deepspeed_to_transformers.py -------------------------------------------------------------------------------- /tools/convert_checkpoint/inspect_checkpoint.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/convert_checkpoint/inspect_checkpoint.py -------------------------------------------------------------------------------- /tools/convert_checkpoint/inspect_deepspeed_checkpoint.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/convert_checkpoint/inspect_deepspeed_checkpoint.py -------------------------------------------------------------------------------- /tools/create_doc_index.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/create_doc_index.py -------------------------------------------------------------------------------- /tools/generate_samples_gpt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/generate_samples_gpt.py -------------------------------------------------------------------------------- /tools/linter.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/linter.py -------------------------------------------------------------------------------- /tools/logs/rescale-logs.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/logs/rescale-logs.py -------------------------------------------------------------------------------- /tools/logs/tb-rename-events.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/logs/tb-rename-events.py -------------------------------------------------------------------------------- /tools/logs/tb-rescale-scalars.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/logs/tb-rescale-scalars.py -------------------------------------------------------------------------------- /tools/merge_mp_partitions.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/merge_mp_partitions.py -------------------------------------------------------------------------------- /tools/merge_preprocessed_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/merge_preprocessed_data.py -------------------------------------------------------------------------------- /tools/openwebtext/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/README.md -------------------------------------------------------------------------------- /tools/openwebtext/add_id.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/add_id.py -------------------------------------------------------------------------------- /tools/openwebtext/blacklist_urls.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/blacklist_urls.py -------------------------------------------------------------------------------- /tools/openwebtext/cleanup_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/cleanup_dataset.py -------------------------------------------------------------------------------- /tools/openwebtext/cleanup_fix_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/cleanup_fix_dataset.py -------------------------------------------------------------------------------- /tools/openwebtext/filter_ngrams.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/filter_ngrams.py -------------------------------------------------------------------------------- /tools/openwebtext/find_duplicates.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/find_duplicates.py -------------------------------------------------------------------------------- /tools/openwebtext/group_duplicate_url.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/group_duplicate_url.py -------------------------------------------------------------------------------- /tools/openwebtext/merge_jsons.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/merge_jsons.py -------------------------------------------------------------------------------- /tools/openwebtext/remove_group_duplicates.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/openwebtext/remove_group_duplicates.py -------------------------------------------------------------------------------- /tools/preprocess_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/preprocess_data.py -------------------------------------------------------------------------------- /tools/preprocess_data_dist.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/preprocess_data_dist.py -------------------------------------------------------------------------------- /tools/preprocess_data_many_cores.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/preprocess_data_many_cores.py -------------------------------------------------------------------------------- /tools/sample_idxs_to_text.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/sample_idxs_to_text.py -------------------------------------------------------------------------------- /tools/tb/tb-remove-events-by-group.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/tb/tb-remove-events-by-group.py -------------------------------------------------------------------------------- /tools/tb/tb-remove-events-by-tag.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/tb/tb-remove-events-by-tag.py -------------------------------------------------------------------------------- /tools/tb/tb-rename-events.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wangguojim/LargeScale/HEAD/tools/tb/tb-rename-events.py --------------------------------------------------------------------------------