DeepSeek-671B-SFT-Guide
An open-source solution for full parameter fine-tuning of DeepSeek-V3/R1 671B, including complete code and scripts from training to inference, as well as some practical experiences and conclusions. (DeepSeek-V3/R1 满血版 671B 全参数微调的开源解决方案,包含从训练到推理的完整代码和脚本,以及实践中积累一些经验和结论。)
File Explorer
- convert_pth_to_hf.sh
- sft_deepseek.py
- sft_deepseek.sh
- sft_deepseek_pdsh.sh
- vllm_deploy_slurm.sh
- __init__.py
- alpaca.py
- arxiv.py
- code_alpaca.py
- colorist.py
- lawyer.py
- medical.py
- moss_003_sft.py
- oasst1.py
- open_orca.py
- sql.py
- tiny_codes.py
- wizardlm.py
- __init__.py
- model.py
- training_args.py
- baichuan2_13b_base_qlora_alpaca_e3.py
- baichuan2_13b_base_qlora_alpaca_enzh_e3.py
- baichuan2_13b_base_qlora_alpaca_enzh_oasst1_e3.py
- baichuan2_13b_base_qlora_alpaca_zh_e3.py
- baichuan2_13b_base_qlora_arxiv_gentitle_e3.py
- baichuan2_13b_base_qlora_code_alpaca_e3.py
- baichuan2_13b_base_qlora_colorist_e5.py
- baichuan2_13b_base_qlora_lawyer_e3.py
- baichuan2_13b_base_qlora_oasst1_512_e3.py
- baichuan2_13b_base_qlora_oasst1_e3.py
- baichuan2_13b_base_qlora_open_platypus_e3.py
- baichuan2_13b_base_qlora_sql_e3.py
- baichuan2_13b_chat_qlora_alpaca_e3.py
- baichuan2_13b_chat_qlora_alpaca_enzh_e3.py
- baichuan2_13b_chat_qlora_alpaca_enzh_oasst1_e3.py
- baichuan2_13b_chat_qlora_alpaca_zh_e3.py
- baichuan2_13b_chat_qlora_code_alpaca_e3.py
- baichuan2_13b_chat_qlora_lawyer_e3.py
- baichuan2_13b_chat_qlora_oasst1_512_e3.py
- baichuan2_13b_chat_qlora_oasst1_e3.py
- baichuan2_13b_chat_qlora_open_platypus_e3.py
- baichuan2_7b_base_qlora_alpaca_e3.py
- baichuan2_7b_base_qlora_alpaca_enzh_e3.py
- baichuan2_7b_base_qlora_alpaca_enzh_oasst1_e3.py
- baichuan2_7b_base_qlora_alpaca_zh_e3.py
- baichuan2_7b_base_qlora_arxiv_gentitle_e3.py
- baichuan2_7b_base_qlora_code_alpaca_e3.py
- baichuan2_7b_base_qlora_colorist_e5.py
- baichuan2_7b_base_qlora_lawyer_e3.py
- baichuan2_7b_base_qlora_oasst1_512_e3.py
- baichuan2_7b_base_qlora_oasst1_e3.py
- baichuan2_7b_base_qlora_open_platypus_e3.py
- baichuan2_7b_base_qlora_sql_e3.py
- baichuan2_7b_chat_qlora_alpaca_e3.py
- baichuan2_7b_chat_qlora_alpaca_enzh_e3.py
- baichuan2_7b_chat_qlora_alpaca_enzh_oasst1_e3.py
- baichuan2_7b_chat_qlora_alpaca_zh_e3.py
- baichuan2_7b_chat_qlora_code_alpaca_e3.py
- baichuan2_7b_chat_qlora_lawyer_e3.py
- baichuan2_7b_chat_qlora_oasst1_512_e3.py
- baichuan2_7b_chat_qlora_oasst1_e3.py
- baichuan2_7b_chat_qlora_open_platypus_e3.py
- baichuan_13b_base_qlora_alpaca_e3.py
- baichuan_13b_base_qlora_alpaca_enzh_e3.py
- baichuan_13b_base_qlora_alpaca_enzh_oasst1_e3.py
- baichuan_13b_base_qlora_alpaca_zh_e3.py
- baichuan_13b_base_qlora_arxiv_gentitle_e3.py
- baichuan_13b_base_qlora_code_alpaca_e3.py
- baichuan_13b_base_qlora_colorist_e5.py
- baichuan_13b_base_qlora_lawyer_e3.py
- baichuan_13b_base_qlora_medical_e1.py
- baichuan_13b_base_qlora_moss_sft_all_e1.py
- baichuan_13b_base_qlora_moss_sft_all_e2_gpu8.py
- baichuan_13b_base_qlora_moss_sft_plugins_e1.py
- baichuan_13b_base_qlora_oasst1_512_e3.py
- baichuan_13b_base_qlora_oasst1_e3.py
- baichuan_13b_base_qlora_open_platypus_e3.py
- baichuan_13b_base_qlora_openorca_e1.py
- baichuan_13b_base_qlora_sql_e3.py
- baichuan_13b_base_qlora_tiny_codes_e1.py
- baichuan_13b_chat_qlora_alpaca_e3.py
- baichuan_13b_chat_qlora_alpaca_enzh_e3.py
- baichuan_13b_chat_qlora_alpaca_enzh_oasst1_e3.py
- baichuan_13b_chat_qlora_alpaca_zh_e3.py
- baichuan_13b_chat_qlora_arxiv_gentitle_e3.py
- baichuan_13b_chat_qlora_code_alpaca_e3.py
- baichuan_13b_chat_qlora_colorist_e5.py
- baichuan_13b_chat_qlora_lawyer_e3.py
- baichuan_13b_chat_qlora_medical_e1.py
- baichuan_13b_chat_qlora_oasst1_512_e3.py
- baichuan_13b_chat_qlora_oasst1_e3.py
- baichuan_13b_chat_qlora_open_platypus_e3.py
- baichuan_13b_chat_qlora_openorca_e1.py
- baichuan_13b_chat_qlora_sql_e3.py
- baichuan_13b_chat_qlora_tiny_codes_e1.py
- baichuan_7b_qlora_alpaca_e3.py
- baichuan_7b_qlora_alpaca_enzh_e3.py
- baichuan_7b_qlora_alpaca_enzh_oasst1_e3.py
- baichuan_7b_qlora_alpaca_zh_e3.py
- baichuan_7b_qlora_arxiv_gentitle_e3.py
- baichuan_7b_qlora_code_alpaca_e3.py
- baichuan_7b_qlora_colorist_e5.py
- baichuan_7b_qlora_lawyer_e3.py
- baichuan_7b_qlora_medical_e1.py
- baichuan_7b_qlora_moss_sft_all_e1.py
- baichuan_7b_qlora_moss_sft_all_e2_gpu8.py
- baichuan_7b_qlora_moss_sft_plugins_e1.py
- baichuan_7b_qlora_oasst1_512_e3.py
- baichuan_7b_qlora_oasst1_e3.py
- baichuan_7b_qlora_open_platypus_e3.py
- baichuan_7b_qlora_openorca_e1.py
- baichuan_7b_qlora_sql_e3.py
- baichuan_7b_qlora_tiny_codes_e1.py
- chatglm2_6b_qlora_alpaca_e3.py
- chatglm2_6b_qlora_alpaca_enzh_e3.py
- chatglm2_6b_qlora_alpaca_enzh_oasst1_e3.py
- chatglm2_6b_qlora_alpaca_zh_e3.py
- chatglm2_6b_qlora_arxiv_gentitle_e3.py
- chatglm2_6b_qlora_code_alpaca_e3.py
- chatglm2_6b_qlora_colorist_e5.py
- chatglm2_6b_qlora_lawyer_e3.py
- chatglm2_6b_qlora_medical_e1.py
- chatglm2_6b_qlora_oasst1_512_e3.py
- chatglm2_6b_qlora_oasst1_e3.py
- chatglm2_6b_qlora_open_platypus_e3.py
- chatglm2_6b_qlora_openorca_e1.py
- chatglm2_6b_qlora_sql_e3.py
- chatglm2_6b_qlora_tiny_codes_e1.py
- chatglm3_6b_qlora_alpaca_e3.py
- chatglm3_6b_qlora_alpaca_enzh_e3.py
- chatglm3_6b_qlora_alpaca_enzh_oasst1_e3.py
- chatglm3_6b_qlora_alpaca_zh_e3.py
- chatglm3_6b_qlora_arxiv_gentitle_e3.py
- chatglm3_6b_qlora_code_alpaca_e3.py
- chatglm3_6b_qlora_colorist_e5.py
- chatglm3_6b_qlora_lawyer_e3.py
- chatglm3_6b_qlora_medical_e1.py
- chatglm3_6b_qlora_oasst1_512_e3.py
- chatglm3_6b_qlora_oasst1_e3.py
- chatglm3_6b_qlora_open_platypus_e3.py
- chatglm3_6b_qlora_openorca_e1.py
- chatglm3_6b_qlora_sql_e3.py
- chatglm3_6b_qlora_tiny_codes_e1.py
- chatglm3_6b_base_qlora_alpaca_e3.py
- chatglm3_6b_base_qlora_alpaca_enzh_e3.py
- chatglm3_6b_base_qlora_alpaca_enzh_oasst1_e3.py
- chatglm3_6b_base_qlora_alpaca_zh_e3.py
- chatglm3_6b_base_qlora_arxiv_gentitle_e3.py
- chatglm3_6b_base_qlora_code_alpaca_e3.py
- chatglm3_6b_base_qlora_colorist_e5.py
- chatglm3_6b_base_qlora_lawyer_e3.py
- chatglm3_6b_base_qlora_medical_e1.py
- chatglm3_6b_base_qlora_oasst1_512_e3.py
- chatglm3_6b_base_qlora_oasst1_e3.py
- chatglm3_6b_base_qlora_open_platypus_e3.py
- chatglm3_6b_base_qlora_openorca_e1.py
- chatglm3_6b_base_qlora_sql_e3.py
- chatglm3_6b_base_qlora_tiny_codes_e1.py
- cohere_100b_128k_sp32.py
- baichuan2_13b_base_full_custom_pretrain_e1.py
- baichuan2_7b_base_full_custom_pretrain_e1.py
- chatglm2_6b_full_custom_pretrain_e1.py
- chatglm3_6b_full_custom_pretrain_e1.py
- deepseek_moe_16b_base_full_custom_pretrain_e1.py
- gemma_2b_full_custom_pretrain_e1.py
- gemma_7b_full_custom_pretrain_e1.py
- internlm2_1_8b_full_custom_pretrain_e1.py
- internlm2_20b_full_custom_pretrain_e1.py
- internlm2_7b_full_custom_pretrain_e1.py
- llama2_70b_full_custom_pretrain_e1.py
- llama2_7b_full_custom_pretrain_e1.py
- minicpm_1b_full_custom_pretrain_e1.py
- minicpm_2b_full_custom_pretrain_e1.py
- mistral_7b_full_custom_pretrain_e1.py
- mixtral_8x7b_full_custom_pretrain_e1.py
- qwen1_5_0_5b_full_custom_pretrain_e1.py
- qwen1_5_14b_full_custom_pretrain_e1.py
- qwen1_5_1_8b_full_custom_pretrain_e1.py
- qwen1_5_4b_full_custom_pretrain_e1.py
- qwen1_5_72b_full_custom_pretrain_e1.py
- qwen1_5_7b_full_custom_pretrain_e1.py
- qwen_1_8b_full_custom_pretrain_e1.py
- qwen_72b_full_custom_pretrain_e1.py
- qwen_7b_full_custom_pretrain_e1.py
- starcoder_full_custom_pretrain_e1.py
- yi_34b_full_custom_pretrain_e1.py
- yi_6b_full_custom_pretrain_e1.py
- zephyr_7b_beta_full_custom_pretrain_e1.py
- baichuan2_13b_chat_qlora_custom_sft_e1.py
- baichuan2_7b_chat_qlora_custom_sft_e1.py
- baichuan_13b_chat_qlora_custom_sft_e1.py
- baichuan_7b_qlora_custom_sft_e1.py
- chatglm2_6b_qlora_custom_sft_e1.py
- chatglm3_6b_qlora_custom_sft_e1.py
- deepseek_moe_16b_chat_qlora_custom_sft_e1.py
- deepseekcoder_6_7b_instruct_qlora_custom_sft_e1.py
- gemma_2b_it_qlora_custom_sft_e1.py
- gemma_2b_qlora_custom_sft_e1.py
- gemma_7b_it_qlora_custom_sft_e1.py
- gemma_7b_qlora_custom_sft_e1.py
- internlm2_chat_1_8b_qlora_custom_sft_e1.py
- internlm2_chat_20b_qlora_custom_sft_e1.py
- internlm2_chat_7b_qlora_custom_sft_e1.py
- llama2_70b_qlora_custom_sft_e1.py
- llama2_7b_chat_qlora_custom_sft_e1.py
- minicpm_1b_full_custom_pretrain_e1.py
- minicpm_2b_full_custom_pretrain_e1.py
- mistral_7b_full_finetune_custom_sft_e1.py
- mixtral_8x7b_instruct_qlora_custom_sft_e1.py
- qwen1_5_0_5b_chat_qlora_custom_sft_e1.py
- qwen1_5_14b_chat_qlora_custom_sft_e1.py
- qwen1_5_1_8b_chat_qlora_custom_sft_e1.py
- qwen1_5_4b_chat_qlora_custom_sft_e1.py
- qwen1_5_72b_chat_qlora_custom_sft_e1.py
- qwen1_5_7b_chat_qlora_custom_sft_e1.py
- qwen_1_8b_chat_qlora_custom_sft_e1.py
- qwen_72b_qlora_custom_sft_e1.py
- qwen_7b_chat_qlora_custom_sft_e1.py
- starcoder_qlora_custom_sft_e1.py
- yi_34b_qlora_custom_sft_e1.py
- yi_6b_qlora_custom_sft_e1.py
- zephyr_7b_beta_qlora_custom_sft_e1.py
- deepseek_coder_6_7b_base_qlora_code_alpaca_e3.py
- deepseekcoder_6_7b_instruct_qlora_code_alpaca_e3.py
- deepseek_moe_16b_base_full_oasst1_e3.py
- deepseek_moe_16b_base_qlora_oasst1_e3.py
- deepseek_moe_16b_chat_full_oasst1_e3.py
- deepseek_moe_16b_chat_qlora_oasst1_e3.py
- deepseek_v2_chat_full_alpaca_e3.py
- deepseek_v2_lite_chat_full_alpaca_e3.py
- deepseek_v2_lite_chat_full_alpaca_e3_32k_varlen.py
- deepspeed_zero1.json
- deepspeed_zero2.json
- deepspeed_zero2_offload.json
- deepspeed_zero3.json
- deepspeed_zero3_offload.json
- internlm2_chat_1_8b_dpo_full.py
- internlm2_chat_1_8b_dpo_full_varlenattn.py
- internlm2_chat_1_8b_dpo_full_varlenattn_jsonl_dataset.py
- internlm2_chat_7b_dpo_qlora_varlenattn.py
- llama3_8b_instruct_dpo_qlora_varlenattn.py
- gemma_2b_full_alpaca_e3.py
- gemma_2b_qlora_alpaca_e3.py
- gemma_2b_it_full_alpaca_e3.py
- gemma_2b_it_qlora_alpaca_e3.py
- gemma_7b_full_alpaca_e3.py
- gemma_7b_qlora_alpaca_e3.py
- gemma_7b_it_full_alpaca_e3.py
- gemma_7b_it_qlora_alpaca_e3.py
- internlm2_1_8b_full_alpaca_e3.py
- internlm2_1_8b_qlora_alpaca_e3.py
- internlm2_20b_full_finetune_custom_dataset_e1.py
- internlm2_20b_qlora_alpaca_e3.py
- internlm2_20b_qlora_arxiv_gentitle_e3.py
- internlm2_20b_qlora_code_alpaca_e3.py
- internlm2_20b_qlora_colorist_e5.py
- internlm2_20b_qlora_lawyer_e3.py
- internlm2_20b_qlora_msagent_react_e3_gpu8.py
- internlm2_20b_qlora_oasst1_512_e3.py
- internlm2_20b_qlora_oasst1_e3.py
- internlm2_20b_qlora_sql_e3.py
- internlm2_5_chat_7b_full_finetune_custom_dataset_e1.py
- internlm2_5_chat_7b_qlora_alpaca_e3.py
- internlm2_5_chat_7b_qlora_oasst1_e3.py
- internlm2_7b_full_finetune_custom_dataset_e1.py
- internlm2_7b_full_finetune_custom_dataset_e1_sequence_parallel_4.py
- internlm2_7b_qlora_alpaca_e3.py
- internlm2_7b_qlora_arxiv_gentitle_e3.py
- internlm2_7b_qlora_code_alpaca_e3.py
- internlm2_7b_qlora_colorist_e5.py
- internlm2_7b_qlora_json_e3.py
- internlm2_7b_qlora_lawyer_e3.py
- internlm2_7b_qlora_msagent_react_e3_gpu8.py
- internlm2_7b_qlora_oasst1_512_e3.py
- internlm2_7b_qlora_oasst1_e3.py
- internlm2_7b_qlora_sql_e3.py
- internlm2_7b_w_internevo_dataset.py
- internlm2_7b_w_tokenized_dataset.py
- internlm2_7b_w_untokenized_dataset.py
- internlm2_chat_1_8b_full_alpaca_e3.py
- internlm2_chat_1_8b_qlora_alpaca_e3.py
- internlm2_chat_20b_full_finetune_custom_dataset_e1.py
- internlm2_chat_20b_qlora_alpaca_e3.py
- internlm2_chat_20b_qlora_code_alpaca_e3.py
- internlm2_chat_20b_qlora_lawyer_e3.py
- internlm2_chat_20b_qlora_oasst1_512_e3.py
- internlm2_chat_20b_qlora_oasst1_e3.py
- internlm2_chat_7b_full_finetune_custom_dataset_e1.py
- internlm2_chat_7b_qlora_alpaca_e3.py
- internlm2_chat_7b_qlora_code_alpaca_e3.py
- internlm2_chat_7b_qlora_lawyer_e3.py
- internlm2_chat_7b_qlora_oasst1_512_e3.py
- internlm2_chat_7b_qlora_oasst1_e3.py
- internlm_20b_qlora_alpaca_e3.py
- internlm_20b_qlora_alpaca_enzh_e3.py
- internlm_20b_qlora_alpaca_enzh_oasst1_e3.py
- internlm_20b_qlora_alpaca_zh_e3.py
- internlm_20b_qlora_arxiv_gentitle_e3.py
- internlm_20b_qlora_code_alpaca_e3.py
- internlm_20b_qlora_colorist_e5.py
- internlm_20b_qlora_lawyer_e3.py
- internlm_20b_qlora_msagent_react_e3_gpu8.py
- internlm_20b_qlora_oasst1_512_e3.py
- internlm_20b_qlora_oasst1_e3.py
- internlm_20b_qlora_open_platypus_e3.py
- internlm_20b_qlora_sql_e3.py
- internlm_7b_full_alpaca_e3.py
- internlm_7b_full_alpaca_enzh_e3.py
- internlm_7b_full_alpaca_enzh_oasst1_e3.py
- internlm_7b_full_alpaca_zh_e3.py
- internlm_7b_full_intern_repo_dataset_template.py
- internlm_7b_full_oasst1_e3.py
- internlm_7b_qlora_alpaca_e3.py
- internlm_7b_qlora_alpaca_enzh_e3.py
- internlm_7b_qlora_alpaca_enzh_oasst1_e3.py
- internlm_7b_qlora_alpaca_zh_e3.py
- internlm_7b_qlora_arxiv_gentitle_e3.py
- internlm_7b_qlora_code_alpaca_e3.py
- internlm_7b_qlora_colorist_e5.py
- internlm_7b_qlora_json_e3.py
- internlm_7b_qlora_lawyer_e3.py
- internlm_7b_qlora_medical_e1.py
- internlm_7b_qlora_moss_sft_all_e1.py
- internlm_7b_qlora_moss_sft_all_e2_gpu8.py
- internlm_7b_qlora_moss_sft_plugins_e1.py
- internlm_7b_qlora_msagent_react_e3_gpu8.py
- internlm_7b_qlora_oasst1_512_e3.py
- internlm_7b_qlora_oasst1_e3.py
- internlm_7b_qlora_oasst1_e3_hf.py
- internlm_7b_qlora_oasst1_mmlu_e3.py
- internlm_7b_qlora_open_platypus_e3.py
- internlm_7b_qlora_openorca_e1.py
- internlm_7b_qlora_sql_e3.py
- internlm_7b_qlora_tiny_codes_e1.py
- internlm_chat_20b_qlora_alpaca_e3.py
- internlm_chat_20b_qlora_alpaca_enzh_e3.py
- internlm_chat_20b_qlora_alpaca_enzh_oasst1_e3.py
- internlm_chat_20b_qlora_alpaca_zh_e3.py
- internlm_chat_20b_qlora_code_alpaca_e3.py
- internlm_chat_20b_qlora_lawyer_e3.py
- internlm_chat_20b_qlora_oasst1_512_e3.py
- internlm_chat_20b_qlora_oasst1_e3.py
- internlm_chat_20b_qlora_open_platypus_e3.py
- internlm_chat_7b_qlora_alpaca_e3.py
- internlm_chat_7b_qlora_alpaca_enzh_e3.py
- internlm_chat_7b_qlora_alpaca_enzh_oasst1_e3.py
- internlm_chat_7b_qlora_alpaca_zh_e3.py
- internlm_chat_7b_qlora_arxiv_gentitle_e3.py
- internlm_chat_7b_qlora_code_alpaca_e3.py
- internlm_chat_7b_qlora_colorist_e5.py
- internlm_chat_7b_qlora_lawyer_e3.py
- internlm_chat_7b_qlora_medical_e1.py
- internlm_chat_7b_qlora_oasst1_512_e3.py
- internlm_chat_7b_qlora_oasst1_e3.py
- internlm_chat_7b_qlora_open_platypus_e3.py
- internlm_chat_7b_qlora_openorca_e1.py
- internlm_chat_7b_qlora_sql_e3.py
- internlm_chat_7b_qlora_tiny_codes_e1.py
- convert_to_official.py
- internvl_v1_5_internlm2_26b_finetune.py
- internvl_v1_5_internlm2_26b_lora_finetune.py
- internvl_v1_5_internlm2_26b_qlora_finetune.py
- internvl_v1_5_internlm2_2b_finetune.py
- internvl_v1_5_internlm2_2b_lora_finetune.py
- internvl_v1_5_internlm2_2b_qlora_finetune.py
- internvl_v1_5_phi3_4b_finetune.py
- internvl_v1_5_phi3_4b_lora_finetune.py
- internvl_v1_5_phi3_4b_qlora_finetune.py
- internvl_v2_internlm2_26b_finetune.py
- internvl_v2_internlm2_26b_lora_finetune.py
- internvl_v2_internlm2_26b_qlora_finetune.py
- internvl_v2_internlm2_2b_finetune.py
- internvl_v2_internlm2_2b_lora_finetune.py
- internvl_v2_internlm2_2b_qlora_finetune.py
- internvl_v2_internlm2_5_8b_finetune.py
- internvl_v2_internlm2_5_8b_lora_finetune.py
- internvl_v2_internlm2_5_8b_qlora_finetune.py
- internvl_v2_phi3_4b_finetune.py
- internvl_v2_phi3_4b_lora_finetune.py
- internvl_v2_phi3_4b_qlora_finetune.py
- llama2_70b_full_wizardlm_e1.py
- llama2_70b_int8_lora_open_platypus_e1.py
- llama2_70b_int8_lora_open_platypus_e1_hf.py
- llama2_70b_qlora_open_platypus_e1.py
- llama2_70b_qlora_open_platypus_e1_hf.py
- llama2_7b_full_pgbooks_400iters_sp1.py
- llama2_7b_full_pgbooks_400iters_sp4.py
- llama2_7b_full_wizardlm_e1.py
- llama2_7b_qlora_alpaca_e3.py
- llama2_7b_qlora_alpaca_enzh_e3.py
- llama2_7b_qlora_alpaca_enzh_oasst1_e3.py
- llama2_7b_qlora_alpaca_zh_e3.py
- llama2_7b_qlora_arxiv_gentitle_e3.py
- llama2_7b_qlora_code_alpaca_e3.py
- llama2_7b_qlora_colorist_e5.py
- llama2_7b_qlora_lawyer_e3.py
- llama2_7b_qlora_medical_e1.py
- llama2_7b_qlora_moss_sft_all_e1.py
- llama2_7b_qlora_moss_sft_all_e2_gpu8.py
- llama2_7b_qlora_moss_sft_plugins_e1.py
- llama2_7b_qlora_msagent_react_e3_gpu8.py
- llama2_7b_qlora_oasst1_512_e3.py
- llama2_7b_qlora_oasst1_e3.py
- llama2_7b_qlora_open_platypus_e3.py
- llama2_7b_qlora_openorca_e1.py
- llama2_7b_qlora_sql_e3.py
- llama2_7b_qlora_tiny_codes_e1.py
- llama2_7b_chat_qlora_alpaca_e3.py
- llama2_7b_chat_qlora_alpaca_enzh_e3.py
- llama2_7b_chat_qlora_alpaca_enzh_oasst1_e3.py
- llama2_7b_chat_qlora_alpaca_zh_e3.py
- llama2_7b_chat_qlora_arxiv_gentitle_e3.py
- llama2_7b_chat_qlora_code_alpaca_e3.py
- llama2_7b_chat_qlora_colorist_e5.py
- llama2_7b_chat_qlora_lawyer_e3.py
- llama2_7b_chat_qlora_medical_e1.py
- llama2_7b_chat_qlora_oasst1_512_e3.py
- llama2_7b_chat_qlora_oasst1_e3.py
- llama2_7b_chat_qlora_open_platypus_e3.py
- llama2_7b_chat_qlora_openorca_e1.py
- llama2_7b_chat_qlora_sql_e3.py
- llama2_7b_chat_qlora_tiny_codes_e1.py
- llama3_70b_instruct_qlora_alpaca_e3_2k_gpu8.py
- llama3_8b_full_alpaca_e3.py
- llama3_8b_instruct_full_alpaca_e3.py
- llama3_8b_instruct_qlora_alpaca_e3.py
- llama_7b_qlora_alpaca_e3.py
- llama_7b_qlora_alpaca_enzh_e3.py
- llama_7b_qlora_alpaca_enzh_oasst1_e3.py
- llama_7b_qlora_alpaca_zh_e3.py
- llama_7b_qlora_arxiv_gentitle_e3.py
- llama_7b_qlora_code_alpaca_e3.py
- llama_7b_qlora_colorist_e5.py
- llama_7b_qlora_lawyer_e3.py
- llama_7b_qlora_medical_e1.py
- llama_7b_qlora_moss_sft_all_e1.py
- llama_7b_qlora_moss_sft_all_e2_gpu8.py
- llama_7b_qlora_moss_sft_plugins_e1.py
- llama_7b_qlora_oasst1_512_e3.py
- llama_7b_qlora_oasst1_e3.py
- llama_7b_qlora_open_platypus_e3.py
- llama_7b_qlora_openorca_e1.py
- llama_7b_qlora_sql_e3.py
- llama_7b_qlora_tiny_codes_e1.py
- llama2_70b_full_alpaca_enzh_128k_sp8.py
- llama2_70b_full_alpaca_enzh_256k_sp16.py
- llama2_70b_full_alpaca_enzh_32k_sp4.py
- llama2_70b_full_alpaca_enzh_8k_sp1.py
- llama2_7b_full_alpaca_enzh_128k_sp8.py
- llama2_7b_full_alpaca_enzh_1M_sp16.py
- llama2_7b_full_alpaca_enzh_256k_sp8.py
- llama2_7b_full_alpaca_enzh_32k_sp1.py
- llama2_7b_full_alpaca_enzh_8k_sp1.py
- yi_34b_200k_full_alpaca_enzh_128k_sp8.py
- yi_34b_200k_full_alpaca_enzh_256k_sp8.py
- yi_34b_200k_full_alpaca_enzh_32k_sp2.py
- yi_34b_200k_full_alpaca_enzh_8k_sp1.py
- llava_internlm2_chat_1_8b_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_internlm2_chat_1_8b_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_internlm2_chat_20b_clip_vit_large_p14_336_e1_gpu8_finetune.py
- llava_internlm2_chat_20b_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_internlm2_chat_20b_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_internlm2_chat_7b_clip_vit_large_p14_336_e1_gpu8_finetune.py
- llava_internlm2_chat_7b_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_internlm2_chat_7b_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_internlm_chat_7b_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_internlm_chat_7b_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_llama3_70b_instruct_quant_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_llama3_8b_instruct_full_clip_vit_large_p14_336_e1_gpu8_finetune.py
- llava_llama3_8b_instruct_full_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_llama3_8b_instruct_full_clip_vit_large_p14_336_lora_e1_gpu8_internvl_finetune.py
- llava_llama3_8b_instruct_qlora_clip_vit_large_p14_336_e1_gpu1_finetune.py
- llava_llama3_8b_instruct_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_llama3_8b_instruct_clip_vit_large_p14_336_e1_gpu8_sharegpt4v_pretrain.py
- llava_llama3_8b_instruct_quant_clip_vit_large_p14_336_e1_gpu1_pretrain.py
- convert_xtuner_weights_to_hf.py
- convert_xtuner_weights_to_llava.py
- llava_v15_13b_finetune.py
- llava_v15_13b_finetune_lora.py
- llava_v15_13b_pretrain.py
- llava_v15_7b_finetune.py
- llava_v15_7b_finetune_lora.py
- llava_v15_7b_pretrain.py
- llava_phi3_mini_4k_instruct_full_clip_vit_large_p14_336_e1_gpu8_finetune.py
- llava_phi3_mini_4k_instruct_full_clip_vit_large_p14_336_full_e2_gpu8_internvl_finetune.py
- llava_phi3_mini_4k_instruct_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_phi3_mini_4k_instruct_clip_vit_large_p14_336_e1_gpu8_sharegpt4v_pretrain.py
- convert_phi_to_llama.py
- convert_xtuner_weights_to_hf.py
- convert_xtuner_weights_to_llava.py
- llava_vicuna_13b_v15_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_vicuna_13b_v15_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- llava_vicuna_7b_v15_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune.py
- llava_vicuna_7b_v15_qlora_clip_vit_large_p14_336_lora_e1_gpu8_finetune_refcoco.py
- llava_vicuna_7b_v15_clip_vit_large_p14_336_e1_gpu8_pretrain.py
- minicpm_1b_dpo_qlora.py
- minicpm_1b_full_alpaca_zh_e3.py
- minicpm_1b_lora_alpaca_zh_e3.py
- minicpm_1b_qlora_alpaca_enzh_e3.py
- minicpm_1b_qlora_alpaca_zh_e3.py
- minicpm_2b_dpo_qlora.py
- minicpm_2b_full_alpaca_zh_e3.py
- minicpm_2b_lora_alpaca_zh_e3.py
- minicpm_2b_qlora_alpaca_enzh_e3.py
- minicpm_2b_qlora_alpaca_zh_e3.py
- mistral_7b_full_finetune_custom_dataset_e1.py
- mistral_7b_qlora_skypile_pretrain_e1.py
- mistral_7b_w_tokenized_dataset.py
- mistral_7b_w_untokenized_dataset.py
- mixtral_8x7b_full_oasst1_e3.py
- mixtral_8x7b_qlora_oasst1_e3.py
- mixtral_8x7b_instruct_full_oasst1_e3.py
- mixtral_8x7b_instruct_qlora_oasst1_e3.py
- internlm2_chat_1_8b_orpo_full.py
- internlm2_chat_1_8b_orpo_full_varlenattn.py
- internlm2_chat_1_8b_orpo_full_varlenattn_jsonl_dataset.py
- internlm2_chat_7b_orpo_qlora_varlenattn_ultrafeedback_e5.py
- llama3_8b_instruct_orpo_qlora_varlenattn_ultrafeedback_e5.py
- phi3_mini_128k_instruct_full_alpaca_e3.py
- phi3_mini_128k_instruct_qlora_alpaca_e3.py
- phi3_mini_4k_instruct_full_alpaca_e3.py
- phi3_mini_4k_instruct_qlora_alpaca_e3.py
- qwen_1_8b_qlora_alpaca_e3.py
- qwen_1_8b_qlora_alpaca_enzh_e3.py
- qwen_1_8b_qlora_alpaca_enzh_oasst1_e3.py
- qwen_1_8b_qlora_alpaca_zh_e3.py
- qwen_1_8b_qlora_code_alpaca_e3.py
- qwen_1_8b_chat_qlora_alpaca_e3.py
- qwen_1_8b_chat_qlora_alpaca_enzh_e3.py
- qwen_1_8b_chat_qlora_alpaca_enzh_oasst1_e3.py
- qwen_1_8b_chat_qlora_alpaca_zh_e3.py
- qwen_1_8b_chat_qlora_code_alpaca_e3.py
- qwen_72b_qlora_alpaca_e3.py
- qwen_72b_qlora_alpaca_enzh_e3.py
- qwen_72b_qlora_alpaca_enzh_oasst1_e3.py
- qwen_72b_qlora_alpaca_zh_e3.py
- qwen_72b_qlora_code_alpaca_e3.py
- qwen_7b_qlora_alpaca_e3.py
- qwen_7b_qlora_alpaca_enzh_e3.py
- qwen_7b_qlora_alpaca_enzh_oasst1_e3.py
- qwen_7b_qlora_alpaca_zh_e3.py
- qwen_7b_qlora_arxiv_gentitle_e3.py
- qwen_7b_qlora_code_alpaca_e3.py
- qwen_7b_qlora_colorist_e5.py
- qwen_7b_qlora_lawyer_e3.py
- qwen_7b_qlora_medical_e1.py
- qwen_7b_qlora_moss_sft_all_e1.py
- qwen_7b_qlora_moss_sft_all_e2_gpu8.py
- qwen_7b_qlora_moss_sft_plugins_e1.py
- qwen_7b_qlora_oasst1_512_e3.py
- qwen_7b_qlora_oasst1_e3.py
- qwen_7b_qlora_open_platypus_e3.py
- qwen_7b_qlora_openorca_e1.py
- qwen_7b_qlora_sql_e3.py
- qwen_7b_qlora_tiny_codes_e1.py
- qwen_7b_chat_qlora_alpaca_e3.py
- qwen_7b_chat_qlora_alpaca_enzh_e3.py
- qwen_7b_chat_qlora_alpaca_enzh_oasst1_e3.py
- qwen_7b_chat_qlora_alpaca_zh_e3.py
- qwen_7b_chat_qlora_arxiv_gentitle_e3.py
- qwen_7b_chat_qlora_code_alpaca_e3.py
- qwen_7b_chat_qlora_colorist_e5.py
- qwen_7b_chat_qlora_lawyer_e3.py
- qwen_7b_chat_qlora_medical_e1.py
- qwen_7b_chat_qlora_oasst1_512_e3.py
- qwen_7b_chat_qlora_oasst1_e3.py
- qwen_7b_chat_qlora_open_platypus_e3.py
- qwen_7b_chat_qlora_openorca_e1.py
- qwen_7b_chat_qlora_sql_e3.py
- qwen_7b_chat_qlora_tiny_codes_e1.py
- qwen1_5_0_5b_full_alpaca_e3.py
- qwen1_5_0_5b_qlora_alpaca_e3.py
- qwen1_5_0_5b_chat_full_alpaca_e3.py
- qwen1_5_0_5b_chat_qlora_alpaca_e3.py
- qwen1_5_110b_full_alpaca_e3.py
- qwen1_5_110b_qlora_alpaca_e3.py
- qwen1_5_110b_chat_full_alpaca_e3.py
- qwen1_5_110b_chat_qlora_alpaca_e3.py
- qwen1_5_110b_chat_qlora_alpaca_e3_16k_2gpus.py
- qwen1_5_14b_full_alpaca_e3.py
- qwen1_5_14b_qlora_alpaca_e3.py
- qwen1_5_14b_chat_full_alpaca_e3.py
- qwen1_5_14b_chat_qlora_alpaca_e3.py
- qwen1_5_1_8b_full_alpaca_e3.py
- qwen1_5_1_8b_qlora_alpaca_e3.py
- qwen1_5_1_8b_chat_full_alpaca_e3.py
- qwen1_5_1_8b_chat_qlora_alpaca_e3.py
- qwen1_5_4b_full_alpaca_e3.py
- qwen1_5_4b_qlora_alpaca_e3.py
- qwen1_5_4b_chat_full_alpaca_e3.py
- qwen1_5_4b_chat_qlora_alpaca_e3.py
- qwen1_5_72b_full_alpaca_e3.py
- qwen1_5_72b_qlora_alpaca_e3.py
- qwen1_5_72b_chat_full_alpaca_e3.py
- qwen1_5_72b_chat_qlora_alpaca_e3.py
- qwen1_5_7b_full_alpaca_e3.py
- qwen1_5_7b_qlora_alpaca_e3.py
- qwen1_5_7b_chat_full_alpaca_e3.py
- qwen1_5_7b_chat_qlora_alpaca_e3.py
- qwen1_5_moe_a2_7_b_chat_full_alpaca_e3.py
- internlm2_chat_1_8b_reward_full_ultrafeedback.py
- internlm2_chat_1_8b_reward_full_varlenattn_jsonl_dataset.py
- internlm2_chat_1_8b_reward_full_varlenattn_ultrafeedback.py
- internlm2_chat_1_8b_reward_qlora_varlenattn_ultrafeedback.py
- llama3_8b_instruct_reward_full_varlenattn_ultrafeedback.py
- starcoder_qlora_stack_exchange_example.py
- yi_34b_qlora_alpaca_enzh_e3.py
- yi_6b_qlora_alpaca_enzh_e3.py
- zephyr_7b_beta_qlora_alpaca_e3.py
- __init__.py
- __init__.py
- default_collate_fn.py
- mmlu_collate_fn.py
- preference_collate_fn.py
- __init__.py
- alpaca_map_fn.py
- alpaca_zh_map_fn.py
- arxiv_map_fn.py
- code_alpaca_map_fn.py
- colors_map_fn.py
- crime_kg_assitant_map_fn.py
- default_map_fn.py
- law_reference_map_fn.py
- llava_map_fn.py
- medical_map_fn.py
- msagent_map_fn.py
- oasst1_map_fn.py
- openai_map_fn.py
- openorca_map_fn.py
- pretrain_map_fn.py
- sql_map_fn.py
- stack_exchange_map_fn.py
- tiny_codes_map_fn.py
- wizardlm_map_fn.py
- __init__.py
- template_map_fn.py
- __init__.py
- intern_repo.py
- length_grouped.py
- __init__.py
- concat_dataset.py
- huggingface.py
- intern_repo.py
- internvl_dataset.py
- json_dataset.py
- llava.py
- modelscope.py
- moss_sft.py
- preference_dataset.py
- refcoco_json.py
- utils.py
- __init__.py
- deepspeed.py
- __init__.py
- dataset_info_hook.py
- evaluate_chat_hook.py
- hf_checkpoint_hook.py
- throughput_hook.py
- varlen_attn_args_to_messagehub_hook.py
- __init__.py
- loops.py
- __init__.py
- __init__.py
- mmlu_metric.py
- reward_metric.py
- __init__.py
- __init__.py
- layer_norm.py
- rms_norm.py
- rotary.py
- __init__.py
- attention.py
- baichuan.py
- cohere.py
- deepseek_v2.py
- deepseek_v3.py
- internlm.py
- internlm2.py
- llama.py
- mistral.py
- phi3.py
- qwen2.py
- utils.py
- yi.py
- __init__.py
- configuration_projector.py
- modeling_projector.py
- __init__.py
- __init__.py
- configuration_deepseek.py
- modeling_deepseek.py
- tokenization_deepseek_fast.py
- __init__.py
- configuration_deepseek.py
- modeling_deepseek.py
- tokenization_deepseek_fast.py
- __init__.py
- configuration_mixtral.py
- modeling_mixtral.py
- __init__.py
- __init__.py
- dpo.py
- internvl.py
- llava.py
- orpo.py
- reward.py
- sft.py
- utils.py
- __init__.py
- attention.py
- comm.py
- data_collate.py
- reduce_loss.py
- sampler.py
- setup_distributed.py
- __init__.py
- arxiv.py
- convert_refcoco.py
- __init__.py
- configuration_internlm2.py
- modeling_internlm2.py
- merge.py
- pth_to_hf.py
- split.py
- __init__.py
- api.py
- calculate.py
- search.py
- solve.py
- chat.py
- check_custom_dataset.py
- copy_cfg.py
- eval_refcoco.py
- get_data_order.py
- list_cfg.py
- list_dataset_format.py
- log_dataset.py
- mmbench.py
- process_untokenized_datasets.py
- process_untokenized_datasets_legacy.py
- process_untokenized_llava_data.py
- test.py
- tokenize_ftdp_datasets.py
- train.py
- utils.py
- __init__.py
- constants.py
- fileio.py
- handle_moe_load_and_save.py
- stop_criteria.py
- templates.py
- zero_to_any_dtype.py
- __init__.py
- entry_point.py
- registry.py
- version.py
- README.md
- train_example.json
- config.json
- configuration.json
- configuration_deepseek.py
- generation_config.json
- LICENSE
- modeling_deepseek.py
- modeling_deepseek_origin.py
- tokenizer.json
- tokenizer_config.json
- config.json
- configuration.json
- configuration_deepseek.py
- modeling_deepseek.py
- tokenizer.json
- tokenizer_config.json
- .gitignore
- LICENSE
- log.png
- README.md
- README_zh.md
- requirements.txt
# Use via CDN
jsDelivrjsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.
Command Glossary
Commands referenced in this DOCs, explained below.
conda activate
View Details ▼
conda activate
Activate a conda environment.
See also: `conda deactivate`.
conda activate myenv
Activate an existing environment named `myenv`:
conda activate {{path/to/myenv}}
Activate an existing environment located at custom path:
conda activate --stack myenv
Stack `myenv` environment on top of a previous environment making libraries/commands/variables from both accessible:
conda create
View Details ▼
conda create
Create new conda environments.
conda create {{[-y|--yes]}} {{[-n|--name]}} py39 python=3.9 "numpy>=1.11" scipy
Create a new environment named `py39`, install Python 3.9, NumPy v1.11 or above in it, and the latest stable version of SciPy. Say yes to all confirmations:
conda create {{[-n|--name]}} myenv --file {{file1.yml}} --file {{file2.yml}}
Create a new environment named `myenv` and install packages listed in files:
conda create {{[-p|--prefix]}} {{path/to/myenv}}
Create a new environment at a custom path (i.e. prefix):
nvcc
View Details ▼
nvcc
The NVIDIA CUDA Compiler Driver.
nvcc {{path/to/source.cu}} {{[-o|--output-file]}} {{path/to/executable}}
Compile a CUDA program:
nvcc {{path/to/source.cu}} {{[-o|--output-file]}} {{path/to/executable}} {{[-g|--debug]}} {{[-G|--device-debug]}}
Generate debug information:
nvcc {{path/to/source.cu}} {{[-o|--output-file]}} {{path/to/executable}} {{[-I|--include-path]}} {{path/to/includes}} {{[-L|--library-path]}} {{path/to/library}} {{[-l|--library]}} {{library_name}}
Include libraries from a different path:
nvidia-smi
View Details ▼
nvidia-smi
Aid the management and monitoring of NVIDIA GPU devices.
nvidia-smi
Display information on all available GPUs and processes using them:
nvidia-smi {{[-q|--query]}}
Display more detailed GPU information:
nvidia-smi dmon
Monitor overall GPU usage with 1-second update interval:
pip install
View Details ▼
pip install
Install Python packages.
pip install {{package1 package2 ...}}
Install one or more packages:
pip install {{package1 package2 ...}} {{[-U|--upgrade]}}
Upgrade all specified packages to the latest version, installing any that are not already present:
pip install {{package}}=={{version}}
Install a specific version of a package:
fallocate
View Details ▼
fallocate
Reserve or deallocate disk space to files.
The utility allocates space without zeroing.
fallocate {{[-l|--length]}} 700M {{path/to/file}}
Reserve a file taking up 700 MiB of disk space:
fallocate {{[-c|--collapse-range]}} {{[-l|--length]}} 200MB {{path/to/file}}
Shrink an already allocated file by 200 MB:
fallocate {{[-c|--collapse-range]}} {{[-o|--offset]}} 100M {{[-l|--length]}} 20M {{path/to/file}}
Shrink 20 MB of space after 100 MiB in a file:
lscpu
View Details ▼
lscpu
Display information about the CPU architecture.
lscpu
Display information about all CPUs:
lscpu {{[-e|--extended]}}
Display information in a table:
lscpu {{[-e|--extended]}} {{[-b|--online]}}
Display only information about online CPUs in a table:
mkswap
View Details ▼
mkswap
Set up a Linux swap area on a device or in a file.
Note: `path/to/file` can either point to a regular file or a swap partition.
sudo mkswap {{path/to/file}}
Set up a given swap area:
sudo mkswap {{[-c|--check]}} {{path/to/file}}
Check a partition for bad blocks before creating the swap area:
sudo mkswap {{[-L|--label]}} {{label}} {{/dev/sdXY}}
Specify a label for the partition (to allow `swapon` to use the label):
sbatch
View Details ▼
sbatch
Submit a batch job to the SLURM scheduler.
sbatch {{path/to/job.sh}}
Submit a batch job:
sbatch --job-name={{myjob}} {{path/to/job.sh}}
Submit a batch job with a custom name:
sbatch --time={{00:30:00}} {{path/to/job.sh}}
Submit a batch job with a time limit of 30 minutes:
swapon
View Details ▼
swapon
Enable devices and files for swapping.
Note: `path/to/file` can either point to a regular file or a swap partition.
swapon
Show swap information:
sudo swapon {{path/to/file}}
Enable a given swap area:
sudo swapon {{[-a|--all]}}
Enable all swap areas specified in `/etc/fstab` except those with the `noauto` option:
