├── .github └── FUNDING.yml ├── .gitignore ├── LICENSE ├── README.md ├── README_ja.md ├── air_llm ├── LICENSE ├── README.md ├── __init__.py ├── airllm │ ├── __init__.py │ ├── airllm.py │ ├── airllm_baichuan.py │ ├── airllm_base.py │ ├── airllm_chatglm.py │ ├── airllm_internlm.py │ ├── airllm_llama_mlx.py │ ├── airllm_mistral.py │ ├── airllm_mixtral.py │ ├── airllm_qwen.py │ ├── airllm_qwen2.py │ ├── auto_model.py │ ├── persist │ │ ├── __init__.py │ │ ├── mlx_model_persister.py │ │ ├── model_persister.py │ │ └── safetensor_model_persister.py │ ├── profiler.py │ ├── tokenization_baichuan.py │ └── utils.py ├── examples │ ├── run_all_types_of_models.ipynb │ ├── run_llama3.1_405B.ipynb │ └── run_on_macos.ipynb ├── inference_example.py ├── setup.py └── tests │ ├── __init__.py │ ├── test_automodel.py │ ├── test_compression.py │ └── test_notebooks │ ├── test_compression.ipynb │ ├── test_mixtral.ipynb │ ├── test_mlx.ipynb │ ├── test_models_transformer_4_35_2.ipynb │ ├── test_models_transformer_4_36_2_torch_2_1_2.ipynb │ └── test_sealllm.ipynb ├── anima_100k ├── 120_topics_en.jsonl ├── README.md ├── extened_longchat_topiced_conversations.json ├── gen_longchat_lines_retrieval_eval_dataset.ipynb ├── gen_longchat_topics_retrieval_eval_dataset_extended.ipynb ├── longer_training.py ├── modeling_flash_llama.py └── run_longer_training.sh ├── anima_logo.png ├── assets ├── 8millionwaystodie.jpeg ├── airllm2_time_improvement.png ├── airllm_logo.png ├── airllm_logo_sm.png ├── anima_q1.png ├── anima_q2.png ├── gpt4_q1.png ├── gpt4_q2.png ├── oom.png ├── wechat_group.png └── wechat_pub_account.jpg ├── data ├── gpt4_translate_vicuna_eval_set.ipynb └── translated_vicuna_eval_set.json ├── eval └── elo_tournanment_all_models_on_translated_vicuna.ipynb ├── examples └── inferrence.ipynb ├── requirements.txt ├── rlhf ├── DPO_loss.png ├── README.md ├── RLHF.png ├── pre_post_dpo_model_output_belle_eval_1k.csv ├── qlora_dpo.py ├── run_dpo_training.sh └── wandb_eval_loss_chart.png ├── scripts └── test_cn_dataset_lenghts.py └── training ├── README.md ├── README_en.md ├── qlora.py ├── run_Amina_training.sh └── run_finetune_raining_based_on_Anima.sh /.github/FUNDING.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/.github/FUNDING.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/.gitignore -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/README.md -------------------------------------------------------------------------------- /README_ja.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/README_ja.md -------------------------------------------------------------------------------- /air_llm/LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/LICENSE -------------------------------------------------------------------------------- /air_llm/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/README.md -------------------------------------------------------------------------------- /air_llm/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /air_llm/airllm/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/__init__.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_baichuan.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_baichuan.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_base.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_chatglm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_chatglm.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_internlm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_internlm.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_llama_mlx.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_llama_mlx.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_mistral.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_mistral.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_mixtral.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_mixtral.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_qwen.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_qwen.py -------------------------------------------------------------------------------- /air_llm/airllm/airllm_qwen2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/airllm_qwen2.py -------------------------------------------------------------------------------- /air_llm/airllm/auto_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/auto_model.py -------------------------------------------------------------------------------- /air_llm/airllm/persist/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/persist/__init__.py -------------------------------------------------------------------------------- /air_llm/airllm/persist/mlx_model_persister.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/persist/mlx_model_persister.py -------------------------------------------------------------------------------- /air_llm/airllm/persist/model_persister.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/persist/model_persister.py -------------------------------------------------------------------------------- /air_llm/airllm/persist/safetensor_model_persister.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/persist/safetensor_model_persister.py -------------------------------------------------------------------------------- /air_llm/airllm/profiler.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/profiler.py -------------------------------------------------------------------------------- /air_llm/airllm/tokenization_baichuan.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/tokenization_baichuan.py -------------------------------------------------------------------------------- /air_llm/airllm/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/airllm/utils.py -------------------------------------------------------------------------------- /air_llm/examples/run_all_types_of_models.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/examples/run_all_types_of_models.ipynb -------------------------------------------------------------------------------- /air_llm/examples/run_llama3.1_405B.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/examples/run_llama3.1_405B.ipynb -------------------------------------------------------------------------------- /air_llm/examples/run_on_macos.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/examples/run_on_macos.ipynb -------------------------------------------------------------------------------- /air_llm/inference_example.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/inference_example.py -------------------------------------------------------------------------------- /air_llm/setup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/setup.py -------------------------------------------------------------------------------- /air_llm/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /air_llm/tests/test_automodel.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_automodel.py -------------------------------------------------------------------------------- /air_llm/tests/test_compression.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_compression.py -------------------------------------------------------------------------------- /air_llm/tests/test_notebooks/test_compression.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_notebooks/test_compression.ipynb -------------------------------------------------------------------------------- /air_llm/tests/test_notebooks/test_mixtral.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_notebooks/test_mixtral.ipynb -------------------------------------------------------------------------------- /air_llm/tests/test_notebooks/test_mlx.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_notebooks/test_mlx.ipynb -------------------------------------------------------------------------------- /air_llm/tests/test_notebooks/test_models_transformer_4_35_2.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_notebooks/test_models_transformer_4_35_2.ipynb -------------------------------------------------------------------------------- /air_llm/tests/test_notebooks/test_models_transformer_4_36_2_torch_2_1_2.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_notebooks/test_models_transformer_4_36_2_torch_2_1_2.ipynb -------------------------------------------------------------------------------- /air_llm/tests/test_notebooks/test_sealllm.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/air_llm/tests/test_notebooks/test_sealllm.ipynb -------------------------------------------------------------------------------- /anima_100k/120_topics_en.jsonl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/120_topics_en.jsonl -------------------------------------------------------------------------------- /anima_100k/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/README.md -------------------------------------------------------------------------------- /anima_100k/extened_longchat_topiced_conversations.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/extened_longchat_topiced_conversations.json -------------------------------------------------------------------------------- /anima_100k/gen_longchat_lines_retrieval_eval_dataset.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/gen_longchat_lines_retrieval_eval_dataset.ipynb -------------------------------------------------------------------------------- /anima_100k/gen_longchat_topics_retrieval_eval_dataset_extended.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/gen_longchat_topics_retrieval_eval_dataset_extended.ipynb -------------------------------------------------------------------------------- /anima_100k/longer_training.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/longer_training.py -------------------------------------------------------------------------------- /anima_100k/modeling_flash_llama.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/modeling_flash_llama.py -------------------------------------------------------------------------------- /anima_100k/run_longer_training.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_100k/run_longer_training.sh -------------------------------------------------------------------------------- /anima_logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/anima_logo.png -------------------------------------------------------------------------------- /assets/8millionwaystodie.jpeg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/8millionwaystodie.jpeg -------------------------------------------------------------------------------- /assets/airllm2_time_improvement.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/airllm2_time_improvement.png -------------------------------------------------------------------------------- /assets/airllm_logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/airllm_logo.png -------------------------------------------------------------------------------- /assets/airllm_logo_sm.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/airllm_logo_sm.png -------------------------------------------------------------------------------- /assets/anima_q1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/anima_q1.png -------------------------------------------------------------------------------- /assets/anima_q2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/anima_q2.png -------------------------------------------------------------------------------- /assets/gpt4_q1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/gpt4_q1.png -------------------------------------------------------------------------------- /assets/gpt4_q2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/gpt4_q2.png -------------------------------------------------------------------------------- /assets/oom.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/oom.png -------------------------------------------------------------------------------- /assets/wechat_group.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/wechat_group.png -------------------------------------------------------------------------------- /assets/wechat_pub_account.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/assets/wechat_pub_account.jpg -------------------------------------------------------------------------------- /data/gpt4_translate_vicuna_eval_set.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/data/gpt4_translate_vicuna_eval_set.ipynb -------------------------------------------------------------------------------- /data/translated_vicuna_eval_set.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/data/translated_vicuna_eval_set.json -------------------------------------------------------------------------------- /eval/elo_tournanment_all_models_on_translated_vicuna.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/eval/elo_tournanment_all_models_on_translated_vicuna.ipynb -------------------------------------------------------------------------------- /examples/inferrence.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/examples/inferrence.ipynb -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/requirements.txt -------------------------------------------------------------------------------- /rlhf/DPO_loss.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/DPO_loss.png -------------------------------------------------------------------------------- /rlhf/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/README.md -------------------------------------------------------------------------------- /rlhf/RLHF.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/RLHF.png -------------------------------------------------------------------------------- /rlhf/pre_post_dpo_model_output_belle_eval_1k.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/pre_post_dpo_model_output_belle_eval_1k.csv -------------------------------------------------------------------------------- /rlhf/qlora_dpo.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/qlora_dpo.py -------------------------------------------------------------------------------- /rlhf/run_dpo_training.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/run_dpo_training.sh -------------------------------------------------------------------------------- /rlhf/wandb_eval_loss_chart.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/rlhf/wandb_eval_loss_chart.png -------------------------------------------------------------------------------- /scripts/test_cn_dataset_lenghts.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/scripts/test_cn_dataset_lenghts.py -------------------------------------------------------------------------------- /training/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/training/README.md -------------------------------------------------------------------------------- /training/README_en.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/training/README_en.md -------------------------------------------------------------------------------- /training/qlora.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/training/qlora.py -------------------------------------------------------------------------------- /training/run_Amina_training.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/training/run_Amina_training.sh -------------------------------------------------------------------------------- /training/run_finetune_raining_based_on_Anima.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/0xSojalSec/airllm/HEAD/training/run_finetune_raining_based_on_Anima.sh --------------------------------------------------------------------------------