Recent Releases of textgen
textgen - 1.1.2
1.1.2版本
GPT模型支持了NEFTune给embedding加噪SFT训练方法,SFT中使用
--neft_alpha参数启用 NEFTune,例如--neft_alpha 5,示例代码:https://github.com/shibing624/textgen/blob/main/examples/gpt/trainingllamademo.py兼容了 ChatGLM-6B 1,2,3 / Baichuan 1,2 / LLaMA 1,2 / BLOOM / Mistral / QWen 等GPT模型LoRA微调训练和预测,可以用于多轮对话微调训练
Full Changelog: https://github.com/shibing624/textgen/compare/1.1.1...1.1.2
- Python
Published by shibing624 over 2 years ago
textgen - 1.1.1
v1.1.1版本
- 支持多卡推理,推理速度加倍,调库textgen做batch推理,多卡推理更方便、快速。
多卡数据并行,batch推理:
cd examples/gpt
CUDA_VISIBLE_DEVICES=0,1 torchrun --nproc_per_node 2 inference_multigpu_demo.py --model_type chatglm --base_model THUDM/chatglm-6b
- 优化ChatGLM-6B/Baichuan/LLaMA2/BLOOM的多轮对话SFT代码,逻辑合并到textgen/gpt下,统一处理多模型微调,加入prompt模板支持。
Full Changelog: https://github.com/shibing624/textgen/compare/1.1.0...1.1.1
- Python
Published by shibing624 over 2 years ago
textgen - v1.1.0
v1.1.0版本
- 发布基于ShareGPT4数据集微调的中英文Vicuna-13B模型shibing624/vicuna-baichuan-13b-chat,和对应的LoRA模型shibing624/vicuna-baichuan-13b-chat-lora;
- 支持多轮对话微调;多轮对话样本格式:examples/data/sharegptzh100_format.jsonl
Full Changelog: https://github.com/shibing624/textgen/compare/1.0.2...1.1.0
- Python
Published by shibing624 over 2 years ago
textgen - v1.0
v1.0版本
- 新增ChatGLM/LLaMA/Bloom模型的多轮对话微调训练,并发布医疗问诊LoRA模型shibing624/ziya-llama-13b-medical-lora。
训练 ChatGLM/LLaMA/Bloom 微调模型
- 支持自定义训练数据集和训练参数,数据集格式参考examples/data/zhcsctest.tsv或者shibing624/alpaca-zh
- 支持AdaLoRA、LoRA、PTuning、PrefixTuning等部分参数微调方法,也支持全参微调
- 支持多卡训练,支持混合精度训练
Full Changelog: https://github.com/shibing624/textgen/compare/0.2.7...1.0.0
- Python
Published by shibing624 over 2 years ago
textgen - v0.2.7
v0.2.7版本
- 新增ChatGLM/LLaMA/Bloom模型的SFT微调训练,并发布适用于通用对话和中文纠错的LoRA模型。
| Model | Arch | Introduction | Train Script | Predict Script | |:----------------------------------------------------------------------------------------------------------|:-----------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------| | shibing624/chatglm-6b-csc-zh-lora | ChatGLM-6B | 在27万中文拼写纠错数据shibing624/CSC上微调了一版ChatGLM-6B,纠错效果有提升,发布微调后的LoRA权重 | training script | predict script | | shibing624/chatglm-6b-belle-zh-lora | ChatGLM-6B | 在100万条中文ChatGPT指令Belle数据集BelleGroup/train1MCN上微调了一版ChatGLM-6B,问答效果有提升,发布微调后的LoRA权重 | training script | predict script | | shibing624/llama-13b-belle-zh-lora | LLaMA-13B | 在100万条中文ChatGPT指令Belle数据集BelleGroup/train1MCN上微调了一版Llama-13B,问答效果有提升,发布微调后的LoRA权重 | training script | predict script | | shibing624/chinese-alpaca-plus-7b-hf | LLaMA-7B | 中文LLaMA-Plus, Alpaca-Plus 7B版本,在LLaMA-7B上扩充了中文词表并继续预训练120G文本(通用领域),在4M指令数据集上微调后得到的中文Alpaca-plus模型 | training script | predict script | | shibing624/chinese-alpaca-plus-13b-hf | LLaMA-13B | 中文LLaMA-Plus, Alpaca-Plus 13B版本,在LLaMA-13B上扩充了中文词表并继续预训练120G文本(通用领域),在4.3M指令数据集上微调后得到的中文Alpaca-plus模型 | training script | predict script |
Full Changelog: https://github.com/shibing624/textgen/compare/0.2.5...0.2.7
- Python
Published by shibing624 over 2 years ago
textgen - 0.2.5
What's Changed
- pad labels to max length by @xingener in https://github.com/shibing624/textgen/pull/25
New Contributors
- @xingener made their first contribution in https://github.com/shibing624/textgen/pull/25
Full Changelog: https://github.com/shibing624/textgen/compare/0.2.0...0.2.5
- Python
Published by shibing624 almost 3 years ago
textgen - 0.1.7
What's Changed
- del the repeated torch.nn.DataParallel by @wiserxin in https://github.com/shibing624/textgen/pull/7
New Contributors
- @wiserxin made their first contribution in https://github.com/shibing624/textgen/pull/7
Full Changelog: https://github.com/shibing624/textgen/compare/0.1.5...0.1.7
- Python
Published by shibing624 about 3 years ago
textgen - 0.1.5 add songnet and copyt5
Full Changelog: https://github.com/shibing624/textgen/compare/0.1.1...0.1.5
- Python
Published by shibing624 over 3 years ago