- deep training framework based on transformers
- pip install -U deep_training
- 源码安装
pip install -U git+https://github.com/ssbuild/deep_training.git
- 源码重装
pip install -U git+https://github.com/ssbuild/deep_training.git --no-deps --force-reinstall
- 
2024-06-10 
- 
2024-02-15 - 0.2.11 support internlm2 https://github.com/ssbuild/internlm2_finetuning
 
- 
2023-12-02 - 0.2.10 update qwen model for 1.8b 7b 14b 72b
- 0.2.10.post0 fix qwen attention_mask
 
- 
2023-11-13 - 0.2.9 release
- 0.2.9.post0 support chatglm3-6b-32k
 
- 
2023-10-22 - 0.2.7
- support clip 完整训练 https://github.com/ssbuild/clip_finetuning
- support asr seq2seq 完整训练 https://github.com/ssbuild/asr_seq2seq_finetuning
- support asr ctc 完整训练 https://github.com/ssbuild/asr_ctc_finetuning
- support object detection 完整训练 https://github.com/ssbuild/detection_finetuning
- support semantic segmentation 完整训练 https://github.com/ssbuild/semantic_segmentation
- support chatglm3 完整训练 https://github.com/ssbuild/chatglm3_finetuning
 
- 0.2.7.post1
- support skywork 完整训练 https://github.com/ssbuild/skywork_finetuning
 
- 0.2.7.post2
- support bluelm 完整训练 https://github.com/ssbuild/bluelm_finetuning
 
- 0.2.7.post3
- support yi 完整训练 https://github.com/ssbuild/yi_finetuning
 
- 0.2.7.post4
- fix dataclass serialization in deepspeed
 
 
- 0.2.7
- 
2023-10-16 - 0.2.6 support muti-model
 
- 
2023-10-07 - 0.2.5
- support colossalai 训练 ,策略 ddp ,gemini,gemini_auto,zero2,zero2_cpu,3d
 
- 0.2.5.post2
- support accelerator 训练 , fix some bug in accelerator and hf trainer
 
- 0.2.5.post4
- fix trainer some bug
 
 
- 0.2.5
- 
2023-09-26 - 0.2.4
- support transformers trainer and qwen-7b 新版 和 qwen-14b , 旧版不再支持,旧版可以安装 deep_training <= 0.2.3
 
- 0.2.4.post3
- support ia3 finetuning
 
 
- 0.2.4
- 
2023-09-21 - 0.2.3
- support dpo 完整训练 dpo_finetuning
 
 
- 0.2.3
- 
2023-09-06 - 0.2.2
- 调整baichuan模块命名 adjust baichuan v2 完整训练 baichuan2_finetuning
 
- 0.2.2.post0
- fix baichuan ptv2
 
- 0.2.2.post1
- fix rwkv4 a bug
 
- 0.2.2.post4
- fix llama and baichuan mask bug
 
 
- 0.2.2
- 
2023-09-02 - 0.2.1
- fix llama model
 
 
- 0.2.1
- 
2023-08-23 - 0.2.0
- release lora内部调整
 
- 0.2.0.post1
- add xverse-13b chat and fix muti lora
 
 
- 0.2.0
- 
2023-08-16 - 0.1.21
- release 增加 5种 rope scale 方法 , fix chatglm2-6b-32k 推理 rope_ratio
 
- 0.1.21.post1
- fix moss rope
 
 
- 0.1.21
- 
2023-08-09 - 0.1.17
- update qwen model
 
- 0.1.17.post0
- update qwen config
 
 
- 0.1.17
- 
2023-08-08 - 0.1.15.rc2
- support XVERSE-13B 完整训练 xverse_finetuning
 
 
- 0.1.15.rc2
- 
2023-08-05 - 0.1.13
- support qwen(千问) 完整训练 qwen_finetuning
 
- 0.1.13.post2
- fix quantization bug
 
- 0.1.14
- release fix qwen stream
 
 
- 0.1.13
- 
2023-07-18 - 0.1.12
- support InternLm(书生) 完整训练 internlm_finetuning
- support baichuan v2 完整训练 baichuan2_finetuning
- fix adalora some bugs
- support rwkv world training
 
 
- 0.1.12
- 
2023-07-04 - 0.1.11 rc1
- support baichuan model 完整训练 baichuan_finetuning
- support chatglm2 model 完整训练 chatglm2_finetuning
 
- 0.1.11
- fix baichuan and chatglm2 some bugs
- support conv2d for lora
- support arrow parquet dataset
 
 
- 0.1.11 rc1
- 
2023-06-06 
- 
2023-06-06 - 0.1.10
- release add qlora and support more optimizer and scheduler
- support lora prompt for deepspeed training
- support rwkv4 完整训练 rwkv_finetuning
 
- 0.1.10.post0
- fix package setup for cpp and cu code for rwkv4
 
- 0.1.10.post1
- fix infer for rwkv4
 
 
- 0.1.10
- 
2023-05-24 - 0.1.8
- fix load weight in prompt_tuning,p_tuning,prefix_tuning,adaption_prompt
 
 
- 0.1.8
- 
2023-05-19 - 0.1.7
- fix 0.1.5 rl bugs
 
- 0.1.7.post1
- fix chatglm-6b-int4,chatglm-6b-int4 p-tuning-v2 training , fix ilql lightning import
- fix load weight in prompt_tuning,p_tuning,prefix_tuning,adaption_prompt
 
 
- 0.1.7
- 
2023-05-10 - 0.1.5
- fix lora v2 modules_to_save 自定义额外训练模块
- support reward ppo llm 完整训练 rlhf_llm
- support reward ppo chatglm 完整训练 rlhf_chatglm
- support reward ppo chatyuan 完整训练 rlhf_chatyuan
 
- 0.1.5.post2 release
- fix prompt modules_to_save 自定义额外训练模块
- support ilql 离线模式训练 ilql 完整训练 rlhf_llm
 
- 0.1.5.post4 release
- fix opt model hidden_size for ppo ilql
- fix ppotrainer ilqltrainer deepspeed save weight
- import AdmaW from transformers or but torch firstly
 
 
- 0.1.5
- 
2023-05-02 - 0.1.4
- support prompt_tuning,p_tuning,prefix_tuning,adaption_prompt
 
 
- 0.1.4
- 
2023-04-21 - 0.1.3rc0
- support moss chat模型 完整训练参考 moss_finetuning
- moss 量化int4 int8推理
 
- 0.1.3.post0
- 新版本基于lightning, pytorch-lightning 更名 lightning,分离numpy-io模块
 
 
- 0.1.3rc0
- 
2023-04-11 - 0.1.2
- 重构lora v2, 增加adalora
 
- 0.1.2.post0
- fix lova v1,lova v2 load_in_8bit
 
 
- 0.1.2
- 
2023-04-07 - deep_training 0.1.1
- update chatglm config
 
 
- deep_training 0.1.1
- 
2023-04-02 - release 0.1.0 and lightning >= 2
 
- 
2023-03-15 - 0.0.18
- support ChatGLM模型(稳定版>=0.0.18.post7) 完整训练参考 chatglm_finetuning
 
- fix deepspeed进程数据平衡
- 0.0.18.post9
- 增加流式输出接口stream_chat接口
 
- 0.0.20 ChatGLM lora
- 加载权重继续训练 , 修改数据数据编码 ,权重自适应
 
- 0.0.21.post0
- fix ChatGLM deepspeed stage 3 权重加载
 
 
- 0.0.18
- 
2023-03-09 - 增加LLaMA 模型(并行版) 完整训练参考 llama_finetuning
 
- 
2023-03-08 - 增加LLaMA 模型(非模型并行版) 完整训练参考 poetry_training
 
- 
2023-03-02 - 增加loRA 训练 , lion,lamb优化器 , 完整训练参考 chatyuan_finetuning
 
- 
2023-02-15 - 增加诗歌PaLM预训练模型
 
- 
2023-02-13 - 增加中文语法纠错模型gector, seq2seq语法纠错模型
 
- 
2023-02-09 - 增加诗歌t5decoder预训练, 诗歌laMDA预训练模型 , t5encoder 预训练模型
 
- 
2023-02-07 - 增加层次分解位置编码选项,让transformer可以处理超长文本
 
- 
2023-01-24 - 增加诗歌gpt2预训练,诗歌t5预训练,诗歌unilm预训练
 
- 
2023-01-20 - 增加对抗训练 FGM, FGSM_Local,FreeAT, PGD, FGSM,FreeAT_Local, 其中FreeAT推荐使用FreeAT_Local,FGSM 推荐使用 FGSM_Local
 
- 
2023-01-19 - 增加promptbertcse监督和非监督模型
 
- 
2023-01-16 - 增加diffcse 监督和非监督模型
 
- 
2023-01-13 - 增加ESimcse 模型
 
- 
2023-01-11 - 增加TSDAE句向量模型
 
- 
2023-01-09 - 增加infonce监督和非监督,simcse监督和非监督,SPN4RE关系模型抽取
 
- 
2023-01-06 - 增加onerel关系模型抽取,prgc关系模型抽取,pure实体模型提取
 
- 
2022-12-24 - 增加unilm模型蒸馏和事件抽取模型
 
- 
2022-12-16 - crf_cascad crf级联抽取实体
- span ner 可重叠多标签,非重叠多标签两种实现方式抽取实体
- mhs_ner 多头选择实体抽取模型
- w2ner 实体抽取模型
- tplinkerplus 实体抽取
- tpliner 关系抽取模型
- tplinkerplus 关系抽取模型
- mhslinker 多头选择关系抽取模型
 
- 
2022-11-17: - simcse-unilm 系列
- simcse-bert-wwm 系列
- tnews circle loss
- afqmc siamese net similar
 
- 
2022-11-15: - unilm autotitle seq2seq autotitle
- 普通分类,指针提取命名实体,crf提取命名实体
- prefixtuning 分类 , prefixtuning 分类 , prefixtuning 指针提取命名实体 , prefixtuning crf 提取命名实体
 
- 
2022-11-12: - gplinker (全局指针提取)
- casrel (A Novel Cascade Binary Tagging Framework for Relational Triple Extraction 参考 https://github.com/weizhepei/CasRel)
- spliner (指针提取关系 sigmoid pointer or simple pointer)
 
- 
2022-11-11: - cluener_pointer 中文命名实体提取 和 cluener crf 中文命名实体提取
- tnews 中文分类
 
- 
2022-11-06: - mlm,gpt2,t5等模型预训练任务
 
- 预训练:
- 数据参考 THUCNews新闻文本分类数据集的子集
- mlm预训练例子 bert roberta等一些列中文预训练
- lm预训练例子 gpt2等一些列中文预训练
- seq2seq 预训练例子 t5 small等一些列中文预训练
- unilm 预训练例子 unilm bert roberta 等一些列中文预训练  
 
- 中文分类:
- 例子 tnews 中文分类
 
- 命名实体提取:
- 参考数据 cluner
- cluener 全局指针提取
- cluener crf提取
- cluener crf prompt提取
- cluener mhs ner多头选择提取
- cluener span指针提取
- cluener crf 级联提取
- cluener tplinkerplus 提取
- pure 提取
- cluener w2ner 提取
 
- 关系提取
- 参考数据 duie和法研杯第一阶段数据
- gplinker 关系提取
- casrel 关系提取
- spliner 关系提取
- mhslinker 关系提取
- tplinker 关系提取
- tplinkerplus 关系提取
- onerel 关系抽取
- prgc 关系提取
- spn4re 关系提取
 
- 事件提取
- 参考数据 duee事件抽取 DuEE v1.0数据集
- gplinker 事件提取
 
-  prompt 系列:
- 例子 prefixprompt tnews中文分类
- 例子 prefixtuning tnews 中文分类
- 例子 prefixtuning cluener 命名实体全局指针提取
- 例子 prefixtuning cluener 命名实体crf提取
- 例子 prompt mlm 自行构建数据模板集,训练参考 pretrain/mlm_pretrain
- 例子 prompt lm 自行构建数据模板集,训练参考 pretrain/seq2seq_pretrain , pretrain/lm_pretrain
 
-  simcse 系列:
- simcse-unilm 系列 例子 unilm+simce 参考数据 THUCNews新闻文本分类数据集的子集
- simcse-bert-wwm 系列 例子 mlm+simcse 参考数据 THUCNews新闻文本分类数据集的子集
 
-  sentense embeding:
- circle loss 例子 tnews circle loss
- siamese net 例子 afqmc siamese net similar
 
   lamb,adma,adamw_hf,adam,adamw,adamw_torch,adamw_torch_fused,adamw_torch_xla,adamw_apex_fused,
   adafactor,adamw_anyprecision,sgd,adagrad,adamw_bnb_8bit,adamw_8bit,lion,lion_8bit,lion_32bit,
   paged_adamw_32bit,paged_adamw_8bit,paged_lion_32bit,paged_lion_8bit,
   lamb_fused_dp adagrad_cpu_dp adam_cpu_dp adam_fused_dp
  linear,WarmupCosine,CAWR,CAL,Step,ReduceLROnPlateau, cosine,cosine_with_restarts,polynomial,
  constant,constant_with_warmup,inverse_sqrt,reduce_lr_on_plateau
Create a model factory, lightweight and efficient training program and make it easier, training model easier to get started.
- pytorch-task-example
- chatmoss_finetuning
- chatglm_finetuning
- chatglm2_finetuning
- t5_finetuning
- llm_finetuning
- llm_rlhf
- chatglm_rlhf
- t5_rlhf
- rwkv_finetuning
- baichuan_finetuning
- internlm_finetuning
- qwen_finetuning
- xverse_finetuning
- auto_finetuning
- aigc_serving
纯粹而干净的代码
本仓库的代码依照 Apache-2.0 协议开源
QQ group:821096761