├── .gitignore ├── .vscode └── settings.json ├── README.md ├── docs ├── Dataset.md ├── FAQ.md ├── LLM_finetuning.md ├── images │ ├── feature-based_FN.png │ ├── feature-based_FN_2.png │ ├── featurebased_FN_.png │ └── full-param-FN.png ├── inference.md ├── multi_gpu.md └── single_gpu.md ├── examples ├── Getting_to_know_Llama.ipynb ├── README.md ├── chat_completion │ ├── chat_completion.py │ └── chats.json ├── code_llama │ ├── code_completion_example.py │ ├── code_completion_prompt.txt │ ├── code_infilling_example.py │ └── code_infilling_prompt.txt ├── custom_dataset.py ├── finetuning.py ├── hf_text_generation_inference │ ├── README.md │ └── merge_lora_weights.py ├── inference.py ├── multi_node.slurm ├── quickstart.ipynb ├── samsum_prompt.txt └── vllm │ └── inference.py ├── install.sh ├── megatron_lm ├── megatron │ ├── core │ │ ├── .gitkeep │ │ └── datasets │ │ │ ├── Makefile │ │ │ ├── blended_dataset.py │ │ │ ├── blended_megatron_dataset_builder.py │ │ │ ├── blended_megatron_dataset_config.py │ │ │ ├── gpt_dataset.py │ │ │ ├── helpers.cpp │ │ │ ├── indexed_dataset.py │ │ │ ├── megatron_dataset.py │ │ │ └── utils.py │ ├── data │ │ ├── bert_dataset.py │ │ ├── data_samplers.py │ │ └── dataset_utils.py │ ├── global_vars.py │ └── tokenizer │ │ ├── __init__.py │ │ └── tokenizer.py └── tools │ └── preprocess_data.py ├── requirements.txt ├── scripts ├── abci │ ├── 13b │ │ ├── llama2-13b_a100_16node.sh │ │ ├── llama2-13b_a100_2node.sh │ │ ├── llama2-13b_a100_2node_offload.sh │ │ ├── llama2-13b_a100_32node.sh │ │ └── llama2-13b_v100_64node_ja_wiki.sh │ ├── 70b │ │ └── llama2-70b_a100.sh │ ├── 7b │ │ ├── llama2-7b_a100_4node.sh │ │ └── llama2-7b_v100_64node.sh │ ├── instruction │ │ ├── 13B │ │ │ ├── convert │ │ │ │ ├── convert-llama-2-13b-base-clueweb.sh │ │ │ │ ├── convert-llama-2-13b-base-gpt-4.sh │ │ │ │ ├── convert-llama-2-13b-base-ichikara.sh │ │ │ │ ├── convert-llama-2-13b-base-llm-jp.sh │ │ │ │ ├── convert-llama-2-13b-base.sh │ │ │ │ ├── convert-meta-llama-2-13b-base.sh │ │ │ │ └── convert-stockmark-13b.sh │ │ │ ├── instruct-llama-2-13b-base-cc-gpt-4.sh │ │ │ ├── instruct-llama-2-13b-base-cc-ichikara.sh │ │ │ ├── instruct-llama-2-13b-base-cc.sh │ │ │ ├── instruct-llama-2-13b-base-clueweb.sh │ │ │ ├── instruct-llama-2-13b-base-llm-jp.sh │ │ │ ├── instruct-llm-jp-13b.sh │ │ │ ├── instruct-meta-llama-2-13b-base.sh │ │ │ ├── instruct-plamo-13b.sh │ │ │ ├── instruct-stockmark-13b.sh │ │ │ └── upload │ │ │ │ ├── upload-abci-llama-2-13b-base-cc-gpt-4.sh │ │ │ │ ├── upload-abci-llama-2-13b-base-cc.sh │ │ │ │ ├── upload-abci-llama-2-13b-base-clueweb.sh │ │ │ │ ├── upload-abci-llama-2-13b-base-llm-jp.sh │ │ │ │ ├── upload-meta-llama-2-13b-base.sh │ │ │ │ └── upload-stockmark-13b.sh │ │ ├── 70B │ │ │ ├── convert │ │ │ │ ├── convert-llama-2-70b-base-cc.sh │ │ │ │ ├── convert-llama-2-70b-base-extended-gpt-4.sh │ │ │ │ ├── convert-llama-2-70b-base-extended-ichikara.sh │ │ │ │ ├── convert-llama-2-70b-base-extended.sh │ │ │ │ └── convert-meta-llama-2-70b.sh │ │ │ ├── instruct-llama-2-base-cc-70b.sh │ │ │ ├── instruct-llama-2-base-extended-cc-70b-gpt-4.sh │ │ │ ├── instruct-llama-2-base-extended-cc-70b-ichikara.sh │ │ │ ├── instruct-llama-2-base-extended-cc-70b.sh │ │ │ ├── instruct-meta-llama-2-70b.sh │ │ │ └── upload │ │ │ │ ├── upload-abci-llama-2-70b-base-cc.sh │ │ │ │ ├── upload-abci-llama-2-70b-base-extended-cc-gpt-4.sh │ │ │ │ └── upload-abci-llama-2-70b-base-extended-cc-ichikara.sh │ │ ├── 7B │ │ │ ├── convert │ │ │ │ ├── convert-cyberagent-calm-2-7b.sh │ │ │ │ ├── convert-elyza-llama-2-7b.sh │ │ │ │ ├── convert-llama-2-7b-base-cc.sh │ │ │ │ ├── convert-llama-2-7b-base-extended-gpt-4.sh │ │ │ │ ├── convert-llama-2-7b-base-extended-ichikara.sh │ │ │ │ ├── convert-llama-2-7b-base-extended.sh │ │ │ │ ├── convert-meta-llama-2-7b.sh │ │ │ │ └── convert-stability-beta-llama-2-7b.sh │ │ │ ├── instruct-cyberagent-calm-2-7b.sh │ │ │ ├── instruct-elyza-llama-2-7b.sh │ │ │ ├── instruct-llama-2-7b-base-cc.sh │ │ │ ├── instruct-llama-2-7b-base-extended-cc-gpt4.sh │ │ │ ├── instruct-llama-2-7b-base-extended-cc-ichikara.sh │ │ │ ├── instruct-llama-2-7b-base-extended-cc.sh │ │ │ ├── instruct-meta-llama-2-7b-base.sh │ │ │ ├── instruct-stability-llama-2-7b.sh │ │ │ ├── instruct-stabilty-stablem-alpha-7b.sh │ │ │ └── upload │ │ │ │ ├── upload-cyberagent-calm-2-7b.sh │ │ │ │ ├── upload-elyza-llama-2-7b.sh │ │ │ │ ├── upload-llama-2-7b-base-cc.sh │ │ │ │ ├── upload-llama-2-7b-base-extended-cc-gpt-4.sh │ │ │ │ ├── upload-llama-2-7b-base-extended-cc-ichikara.sh │ │ │ │ ├── upload-llama-2-7b-base-extended-cc.sh │ │ │ │ ├── upload-meta-llama-2-7b-base.sh │ │ │ │ └── upload-stablelm-beta-llama-2-7b.sh │ │ ├── hh-rlhf-49k-ja.py │ │ ├── merge_jsonl.py │ │ ├── oasst1-89k-ja.py │ │ └── split_dataset.py │ ├── llm_jp_tokenize.py │ ├── mistral │ │ ├── mistral-7b-SWA.sh │ │ ├── mistral-7b-blockexp-40.sh │ │ ├── mistral-7b-blockexp-64.sh │ │ ├── mistral-7b-nve-algebraic-stack.sh │ │ ├── mistral-7b-ve-algebraic_stack.sh │ │ ├── mistral-7b-ve-vault.sh │ │ ├── mistral-7b-ve.sh │ │ ├── mistral-7b.sh │ │ ├── mistral-cc-en-ja.sh │ │ └── mistral-cc.sh │ ├── mixtral │ │ ├── mixtral-7bx8.sh │ │ ├── mixtral-8x7B-abci.sh │ │ └── mixtral-config.json │ ├── parallel_corpus │ │ └── Llama-2-7b-base.sh │ ├── pubmed │ │ ├── Llama-2-13b-base-extended.sh │ │ ├── dataset_index.sh │ │ └── dataset_split.sh │ ├── reserve │ │ └── auto_reserve.py │ ├── tokenize.sh │ ├── tokenize_en_arxiv.sh │ ├── tokenize_en_books.sh │ └── tokenize_ja_wiki.sh ├── index.sh ├── kotoba │ └── upload │ │ ├── dataset_upload.py │ │ ├── upload-all-base.sh │ │ ├── upload-all-chat.sh │ │ ├── upload.py │ │ └── upload.sh └── llama │ ├── instruction_tuning │ └── llama2-7b-alpaca.sh │ └── next_token │ └── llama2-7b-wikipedia.sh ├── src └── llama_recipes │ ├── arguments.py │ ├── data │ ├── __init__.py │ ├── concatenator.py │ └── sampler.py │ ├── datasets │ ├── pretrain_dataset.py │ └── utils.py │ ├── finetuning.py │ ├── get_fsdp.py │ ├── get_model_decoder_layer.py │ ├── get_models.py │ ├── inference │ ├── __init__.py │ ├── chat_utils.py │ ├── checkpoint_converter_fsdp_hf.py │ ├── model_utils.py │ └── safety_utils.py │ ├── optimizer.py │ ├── policies │ ├── __init__.py │ ├── activation_checkpointing_functions.py │ ├── anyprecision_optimizer.py │ ├── mixed_precision.py │ └── wrapping.py │ └── utils │ ├── __init__.py │ ├── checkpoint.py │ ├── dataset_utils.py │ ├── distributed.py │ ├── fsdp_utils.py │ ├── memory_utils.py │ ├── random.py │ ├── sequence_length_warmup.py │ ├── train_utils.py │ └── wandb_utils.py ├── tests ├── conftest.py ├── datasets │ ├── test_custom_dataset.py │ ├── test_grammar_datasets.py │ └── test_samsum_datasets.py ├── test_batching.py ├── test_finetuning.py └── test_sampler.py └── tools ├── README.md ├── checkpoint-convert ├── convert_ckpt.py ├── convert_fsdp.py ├── scripts │ ├── abci │ │ ├── convert_ckpt.sh │ │ ├── convert_deepspeed.sh │ │ └── convert_fsdp.sh │ ├── convert.sh │ └── ylab │ │ ├── check.py │ │ ├── check.sh │ │ ├── convert_ckpt.sh │ │ └── convert_deepspeed.sh └── zero_to_fp32.py ├── dataset └── pubmed_dataset_builder.py ├── inference ├── inference-abci.sh ├── inference-mixtral.py ├── inference-ylab.sh ├── inference.py └── inference.sh ├── install.sh ├── model-upload ├── upload.py └── upload.sh ├── pre-process ├── convert_json_jsonl.py ├── index_dataset.py └── scripts │ └── index.sh └── requirements.txt /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/.gitignore -------------------------------------------------------------------------------- /.vscode/settings.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/.vscode/settings.json -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/README.md -------------------------------------------------------------------------------- /docs/Dataset.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/Dataset.md -------------------------------------------------------------------------------- /docs/FAQ.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/FAQ.md -------------------------------------------------------------------------------- /docs/LLM_finetuning.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/LLM_finetuning.md -------------------------------------------------------------------------------- /docs/images/feature-based_FN.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/images/feature-based_FN.png -------------------------------------------------------------------------------- /docs/images/feature-based_FN_2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/images/feature-based_FN_2.png -------------------------------------------------------------------------------- /docs/images/featurebased_FN_.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/images/featurebased_FN_.png -------------------------------------------------------------------------------- /docs/images/full-param-FN.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/images/full-param-FN.png -------------------------------------------------------------------------------- /docs/inference.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/inference.md -------------------------------------------------------------------------------- /docs/multi_gpu.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/multi_gpu.md -------------------------------------------------------------------------------- /docs/single_gpu.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/docs/single_gpu.md -------------------------------------------------------------------------------- /examples/Getting_to_know_Llama.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/Getting_to_know_Llama.ipynb -------------------------------------------------------------------------------- /examples/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/README.md -------------------------------------------------------------------------------- /examples/chat_completion/chat_completion.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/chat_completion/chat_completion.py -------------------------------------------------------------------------------- /examples/chat_completion/chats.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/chat_completion/chats.json -------------------------------------------------------------------------------- /examples/code_llama/code_completion_example.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/code_llama/code_completion_example.py -------------------------------------------------------------------------------- /examples/code_llama/code_completion_prompt.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/code_llama/code_completion_prompt.txt -------------------------------------------------------------------------------- /examples/code_llama/code_infilling_example.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/code_llama/code_infilling_example.py -------------------------------------------------------------------------------- /examples/code_llama/code_infilling_prompt.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/code_llama/code_infilling_prompt.txt -------------------------------------------------------------------------------- /examples/custom_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/custom_dataset.py -------------------------------------------------------------------------------- /examples/finetuning.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/finetuning.py -------------------------------------------------------------------------------- /examples/hf_text_generation_inference/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/hf_text_generation_inference/README.md -------------------------------------------------------------------------------- /examples/hf_text_generation_inference/merge_lora_weights.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/hf_text_generation_inference/merge_lora_weights.py -------------------------------------------------------------------------------- /examples/inference.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/inference.py -------------------------------------------------------------------------------- /examples/multi_node.slurm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/multi_node.slurm -------------------------------------------------------------------------------- /examples/quickstart.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/quickstart.ipynb -------------------------------------------------------------------------------- /examples/samsum_prompt.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/samsum_prompt.txt -------------------------------------------------------------------------------- /examples/vllm/inference.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/examples/vllm/inference.py -------------------------------------------------------------------------------- /install.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/install.sh -------------------------------------------------------------------------------- /megatron_lm/megatron/core/.gitkeep: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/Makefile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/Makefile -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/blended_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/blended_dataset.py -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/blended_megatron_dataset_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/blended_megatron_dataset_builder.py -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/blended_megatron_dataset_config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/blended_megatron_dataset_config.py -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/gpt_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/gpt_dataset.py -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/helpers.cpp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/helpers.cpp -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/indexed_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/indexed_dataset.py -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/megatron_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/megatron_dataset.py -------------------------------------------------------------------------------- /megatron_lm/megatron/core/datasets/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/core/datasets/utils.py -------------------------------------------------------------------------------- /megatron_lm/megatron/data/bert_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/data/bert_dataset.py -------------------------------------------------------------------------------- /megatron_lm/megatron/data/data_samplers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/data/data_samplers.py -------------------------------------------------------------------------------- /megatron_lm/megatron/data/dataset_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/data/dataset_utils.py -------------------------------------------------------------------------------- /megatron_lm/megatron/global_vars.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/global_vars.py -------------------------------------------------------------------------------- /megatron_lm/megatron/tokenizer/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/tokenizer/__init__.py -------------------------------------------------------------------------------- /megatron_lm/megatron/tokenizer/tokenizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/megatron/tokenizer/tokenizer.py -------------------------------------------------------------------------------- /megatron_lm/tools/preprocess_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/megatron_lm/tools/preprocess_data.py -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/requirements.txt -------------------------------------------------------------------------------- /scripts/abci/13b/llama2-13b_a100_16node.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/13b/llama2-13b_a100_16node.sh -------------------------------------------------------------------------------- /scripts/abci/13b/llama2-13b_a100_2node.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/13b/llama2-13b_a100_2node.sh -------------------------------------------------------------------------------- /scripts/abci/13b/llama2-13b_a100_2node_offload.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/13b/llama2-13b_a100_2node_offload.sh -------------------------------------------------------------------------------- /scripts/abci/13b/llama2-13b_a100_32node.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/13b/llama2-13b_a100_32node.sh -------------------------------------------------------------------------------- /scripts/abci/13b/llama2-13b_v100_64node_ja_wiki.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/13b/llama2-13b_v100_64node_ja_wiki.sh -------------------------------------------------------------------------------- /scripts/abci/70b/llama2-70b_a100.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/70b/llama2-70b_a100.sh -------------------------------------------------------------------------------- /scripts/abci/7b/llama2-7b_a100_4node.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/7b/llama2-7b_a100_4node.sh -------------------------------------------------------------------------------- /scripts/abci/7b/llama2-7b_v100_64node.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/7b/llama2-7b_v100_64node.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-clueweb.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-clueweb.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-llm-jp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-llama-2-13b-base-llm-jp.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-llama-2-13b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-llama-2-13b-base.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-meta-llama-2-13b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-meta-llama-2-13b-base.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/convert/convert-stockmark-13b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/convert/convert-stockmark-13b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-llama-2-13b-base-cc-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-llama-2-13b-base-cc-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-llama-2-13b-base-cc-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-llama-2-13b-base-cc-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-llama-2-13b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-llama-2-13b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-llama-2-13b-base-clueweb.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-llama-2-13b-base-clueweb.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-llama-2-13b-base-llm-jp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-llama-2-13b-base-llm-jp.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-llm-jp-13b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-llm-jp-13b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-meta-llama-2-13b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-meta-llama-2-13b-base.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-plamo-13b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-plamo-13b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/instruct-stockmark-13b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/instruct-stockmark-13b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-cc-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-cc-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-clueweb.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-clueweb.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-llm-jp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/upload/upload-abci-llama-2-13b-base-llm-jp.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/upload/upload-meta-llama-2-13b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/upload/upload-meta-llama-2-13b-base.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/13B/upload/upload-stockmark-13b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/13B/upload/upload-stockmark-13b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-extended-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-extended-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-extended-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-extended-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-extended.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/convert/convert-llama-2-70b-base-extended.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/convert/convert-meta-llama-2-70b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/convert/convert-meta-llama-2-70b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/instruct-llama-2-base-cc-70b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/instruct-llama-2-base-cc-70b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/instruct-llama-2-base-extended-cc-70b-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/instruct-llama-2-base-extended-cc-70b-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/instruct-llama-2-base-extended-cc-70b-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/instruct-llama-2-base-extended-cc-70b-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/instruct-llama-2-base-extended-cc-70b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/instruct-llama-2-base-extended-cc-70b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/instruct-meta-llama-2-70b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/instruct-meta-llama-2-70b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/upload/upload-abci-llama-2-70b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/upload/upload-abci-llama-2-70b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/upload/upload-abci-llama-2-70b-base-extended-cc-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/upload/upload-abci-llama-2-70b-base-extended-cc-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/70B/upload/upload-abci-llama-2-70b-base-extended-cc-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/70B/upload/upload-abci-llama-2-70b-base-extended-cc-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-cyberagent-calm-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-cyberagent-calm-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-elyza-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-elyza-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-extended-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-extended-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-extended-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-extended-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-extended.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-llama-2-7b-base-extended.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-meta-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-meta-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/convert/convert-stability-beta-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/convert/convert-stability-beta-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-cyberagent-calm-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-cyberagent-calm-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-elyza-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-elyza-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-llama-2-7b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-llama-2-7b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-llama-2-7b-base-extended-cc-gpt4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-llama-2-7b-base-extended-cc-gpt4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-llama-2-7b-base-extended-cc-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-llama-2-7b-base-extended-cc-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-llama-2-7b-base-extended-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-llama-2-7b-base-extended-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-meta-llama-2-7b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-meta-llama-2-7b-base.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-stability-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-stability-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/instruct-stabilty-stablem-alpha-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/instruct-stabilty-stablem-alpha-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-cyberagent-calm-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-cyberagent-calm-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-elyza-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-elyza-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-extended-cc-gpt-4.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-extended-cc-gpt-4.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-extended-cc-ichikara.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-extended-cc-ichikara.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-extended-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-llama-2-7b-base-extended-cc.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-meta-llama-2-7b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-meta-llama-2-7b-base.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/7B/upload/upload-stablelm-beta-llama-2-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/7B/upload/upload-stablelm-beta-llama-2-7b.sh -------------------------------------------------------------------------------- /scripts/abci/instruction/hh-rlhf-49k-ja.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/hh-rlhf-49k-ja.py -------------------------------------------------------------------------------- /scripts/abci/instruction/merge_jsonl.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/merge_jsonl.py -------------------------------------------------------------------------------- /scripts/abci/instruction/oasst1-89k-ja.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/oasst1-89k-ja.py -------------------------------------------------------------------------------- /scripts/abci/instruction/split_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/instruction/split_dataset.py -------------------------------------------------------------------------------- /scripts/abci/llm_jp_tokenize.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/llm_jp_tokenize.py -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-SWA.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-SWA.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-blockexp-40.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-blockexp-40.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-blockexp-64.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-blockexp-64.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-nve-algebraic-stack.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-nve-algebraic-stack.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-ve-algebraic_stack.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-ve-algebraic_stack.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-ve-vault.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-ve-vault.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b-ve.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b-ve.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-7b.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-7b.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-cc-en-ja.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-cc-en-ja.sh -------------------------------------------------------------------------------- /scripts/abci/mistral/mistral-cc.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mistral/mistral-cc.sh -------------------------------------------------------------------------------- /scripts/abci/mixtral/mixtral-7bx8.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mixtral/mixtral-7bx8.sh -------------------------------------------------------------------------------- /scripts/abci/mixtral/mixtral-8x7B-abci.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mixtral/mixtral-8x7B-abci.sh -------------------------------------------------------------------------------- /scripts/abci/mixtral/mixtral-config.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/mixtral/mixtral-config.json -------------------------------------------------------------------------------- /scripts/abci/parallel_corpus/Llama-2-7b-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/parallel_corpus/Llama-2-7b-base.sh -------------------------------------------------------------------------------- /scripts/abci/pubmed/Llama-2-13b-base-extended.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/pubmed/Llama-2-13b-base-extended.sh -------------------------------------------------------------------------------- /scripts/abci/pubmed/dataset_index.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/pubmed/dataset_index.sh -------------------------------------------------------------------------------- /scripts/abci/pubmed/dataset_split.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/pubmed/dataset_split.sh -------------------------------------------------------------------------------- /scripts/abci/reserve/auto_reserve.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/reserve/auto_reserve.py -------------------------------------------------------------------------------- /scripts/abci/tokenize.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/tokenize.sh -------------------------------------------------------------------------------- /scripts/abci/tokenize_en_arxiv.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/tokenize_en_arxiv.sh -------------------------------------------------------------------------------- /scripts/abci/tokenize_en_books.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/tokenize_en_books.sh -------------------------------------------------------------------------------- /scripts/abci/tokenize_ja_wiki.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/abci/tokenize_ja_wiki.sh -------------------------------------------------------------------------------- /scripts/index.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/index.sh -------------------------------------------------------------------------------- /scripts/kotoba/upload/dataset_upload.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/kotoba/upload/dataset_upload.py -------------------------------------------------------------------------------- /scripts/kotoba/upload/upload-all-base.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/kotoba/upload/upload-all-base.sh -------------------------------------------------------------------------------- /scripts/kotoba/upload/upload-all-chat.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/kotoba/upload/upload-all-chat.sh -------------------------------------------------------------------------------- /scripts/kotoba/upload/upload.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/kotoba/upload/upload.py -------------------------------------------------------------------------------- /scripts/kotoba/upload/upload.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/kotoba/upload/upload.sh -------------------------------------------------------------------------------- /scripts/llama/instruction_tuning/llama2-7b-alpaca.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/llama/instruction_tuning/llama2-7b-alpaca.sh -------------------------------------------------------------------------------- /scripts/llama/next_token/llama2-7b-wikipedia.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/scripts/llama/next_token/llama2-7b-wikipedia.sh -------------------------------------------------------------------------------- /src/llama_recipes/arguments.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/arguments.py -------------------------------------------------------------------------------- /src/llama_recipes/data/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/data/__init__.py -------------------------------------------------------------------------------- /src/llama_recipes/data/concatenator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/data/concatenator.py -------------------------------------------------------------------------------- /src/llama_recipes/data/sampler.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/data/sampler.py -------------------------------------------------------------------------------- /src/llama_recipes/datasets/pretrain_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/datasets/pretrain_dataset.py -------------------------------------------------------------------------------- /src/llama_recipes/datasets/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/datasets/utils.py -------------------------------------------------------------------------------- /src/llama_recipes/finetuning.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/finetuning.py -------------------------------------------------------------------------------- /src/llama_recipes/get_fsdp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/get_fsdp.py -------------------------------------------------------------------------------- /src/llama_recipes/get_model_decoder_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/get_model_decoder_layer.py -------------------------------------------------------------------------------- /src/llama_recipes/get_models.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/get_models.py -------------------------------------------------------------------------------- /src/llama_recipes/inference/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/inference/__init__.py -------------------------------------------------------------------------------- /src/llama_recipes/inference/chat_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/inference/chat_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/inference/checkpoint_converter_fsdp_hf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/inference/checkpoint_converter_fsdp_hf.py -------------------------------------------------------------------------------- /src/llama_recipes/inference/model_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/inference/model_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/inference/safety_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/inference/safety_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/optimizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/optimizer.py -------------------------------------------------------------------------------- /src/llama_recipes/policies/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/policies/__init__.py -------------------------------------------------------------------------------- /src/llama_recipes/policies/activation_checkpointing_functions.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/policies/activation_checkpointing_functions.py -------------------------------------------------------------------------------- /src/llama_recipes/policies/anyprecision_optimizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/policies/anyprecision_optimizer.py -------------------------------------------------------------------------------- /src/llama_recipes/policies/mixed_precision.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/policies/mixed_precision.py -------------------------------------------------------------------------------- /src/llama_recipes/policies/wrapping.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/policies/wrapping.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/__init__.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/checkpoint.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/checkpoint.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/dataset_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/dataset_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/distributed.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/distributed.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/fsdp_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/fsdp_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/memory_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/memory_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/random.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/random.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/sequence_length_warmup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/sequence_length_warmup.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/train_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/train_utils.py -------------------------------------------------------------------------------- /src/llama_recipes/utils/wandb_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/src/llama_recipes/utils/wandb_utils.py -------------------------------------------------------------------------------- /tests/conftest.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/conftest.py -------------------------------------------------------------------------------- /tests/datasets/test_custom_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/datasets/test_custom_dataset.py -------------------------------------------------------------------------------- /tests/datasets/test_grammar_datasets.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/datasets/test_grammar_datasets.py -------------------------------------------------------------------------------- /tests/datasets/test_samsum_datasets.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/datasets/test_samsum_datasets.py -------------------------------------------------------------------------------- /tests/test_batching.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/test_batching.py -------------------------------------------------------------------------------- /tests/test_finetuning.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/test_finetuning.py -------------------------------------------------------------------------------- /tests/test_sampler.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tests/test_sampler.py -------------------------------------------------------------------------------- /tools/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/README.md -------------------------------------------------------------------------------- /tools/checkpoint-convert/convert_ckpt.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/convert_ckpt.py -------------------------------------------------------------------------------- /tools/checkpoint-convert/convert_fsdp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/convert_fsdp.py -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/abci/convert_ckpt.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/abci/convert_ckpt.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/abci/convert_deepspeed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/abci/convert_deepspeed.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/abci/convert_fsdp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/abci/convert_fsdp.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/convert.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/convert.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/ylab/check.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/ylab/check.py -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/ylab/check.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/ylab/check.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/ylab/convert_ckpt.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/ylab/convert_ckpt.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/scripts/ylab/convert_deepspeed.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/scripts/ylab/convert_deepspeed.sh -------------------------------------------------------------------------------- /tools/checkpoint-convert/zero_to_fp32.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/checkpoint-convert/zero_to_fp32.py -------------------------------------------------------------------------------- /tools/dataset/pubmed_dataset_builder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/dataset/pubmed_dataset_builder.py -------------------------------------------------------------------------------- /tools/inference/inference-abci.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/inference/inference-abci.sh -------------------------------------------------------------------------------- /tools/inference/inference-mixtral.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/inference/inference-mixtral.py -------------------------------------------------------------------------------- /tools/inference/inference-ylab.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/inference/inference-ylab.sh -------------------------------------------------------------------------------- /tools/inference/inference.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/inference/inference.py -------------------------------------------------------------------------------- /tools/inference/inference.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/inference/inference.sh -------------------------------------------------------------------------------- /tools/install.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/install.sh -------------------------------------------------------------------------------- /tools/model-upload/upload.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/model-upload/upload.py -------------------------------------------------------------------------------- /tools/model-upload/upload.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/model-upload/upload.sh -------------------------------------------------------------------------------- /tools/pre-process/convert_json_jsonl.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/pre-process/convert_json_jsonl.py -------------------------------------------------------------------------------- /tools/pre-process/index_dataset.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/pre-process/index_dataset.py -------------------------------------------------------------------------------- /tools/pre-process/scripts/index.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/pre-process/scripts/index.sh -------------------------------------------------------------------------------- /tools/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/okoge-kaz/moe-recipes/HEAD/tools/requirements.txt --------------------------------------------------------------------------------