shibing624/MedicalGPT
MedicalGPT: Training Your Own Medical GPT Model with ChatGPT Training Pipeline. 训练医疗大模型,实现了包括增量预训练(PT)、有监督微调(SFT)、RLHF、DPO、ORPO。
PythonApache-2.0
Pinned issues
Issues
- 1
- 1
关于llama3的权重转换
#378 opened by tszslovewanpu - 2
医学大模型全流程体验
#377 opened by YoshuaBengio - 2
- 2
ppo训练时出现问题:UserWarning: KL divergence is starting to become negative: -233.50
#374 opened by user2311717757 - 4
运行pretraining.py时报错:RuntimeError: CUDA error: device-side assert triggered
#376 opened by Wenting1227 - 1
vocab扩展后的模型合并问题
#373 opened by sungatetop - 1
有没有人能分享下自己微调后的模型id,我懒得弄,只想吃现成的
#372 opened by aqpmzngldh - 1
AMD 执行 run_pt.sh失败
#371 opened by liuyang6055 - 4
关于提前结束训练
#368 opened by tszslovewanpu - 6
orpo脚本NoneType问题
#363 opened by songyao199681 - 6
预训练后模型出现自问自答、输出未知序列、重复口吃现象
#351 opened by Peter-of-Astora - 2
dpo_training.py eal存在空的情况
#369 opened by 14686039 - 2
UserWarning: None of the inputs have requires_grad=True. Gradients will be None
#359 opened by cove1011 - 1
几步的训练怎么都是独立的,rm都没用sft的adapter
#365 opened by cqray1990 - 1
对chat模型进行二次预训练后,自问自答
#366 opened by wsl1014 - 1
训练reward_modeling.py
#364 opened by cqray1990 - 1
reward_modeling咨询
#361 opened by tuqingwen - 2
Regarding RLHF and DPO training data
#358 opened by Aniketto16 - 5
- 1
运行inference.py文件,报AttributeError: property 'eos_token' of 'ChatGLMTokenizer' object has no setter
#353 opened by liulint - 2
使用deepspeed 全参数sft后,inference 回答的都为空,有解决办法吗
#357 opened by Yian320 - 3
全参数SFT后无法正常infer
#331 opened by nuoma - 1
ValueError: operands could not be broadcast together with remapped shapes [original->remapped]: (3,2) and requested shape (1,2)
#356 opened by Riapy - 2
lora模型合并
#355 opened by sevenandseven - 0
扩充词表后能否直接进行SFT呢?
#352 opened by HaotianLiu123 - 5
assert tokenzier_vocab_size > model_vocab_size
#350 opened by sevenandseven - 1
- 1
llama进行rm训练的时候,出现问题ValueError: weight is on the meta device, we need a `value` to put in on cpu.
#347 opened by cove1011 - 1
关于Chatglm3的增量预训练
#348 opened by XueMoonLit - 1
- 1
ValueError: The model does not have a language model head, please use a model that has one.
#345 opened by cove1011 - 2
- 3
chatglm2合并sft_qlora后,推理出现自动续答
#339 opened by Lxhnnn - 1
ChatGLMForSequenceClassification rm步骤出错
#341 opened by cove1011 - 0
TypeError: ChatGLMForSequenceClassification.forward() got an unexpected keyword argument 'output_attentions'
#343 opened by cove1011 - 2
chatglm3训练在rm之后,进行lora模型权重合并到base model,出现问题:ValueError: chatglm does not support sequence classification
#344 opened by cove1011 - 1
单机多卡sft deepspeed zero3 训练一直卡在训练阶段
#330 opened by lainxx - 5
DPO阶段报错
#334 opened by small-white-zs - 1
Please provide an offload_folder
#335 opened by cove1011 - 3
请问这是什么oserror
#337 opened by cove1011 - 3
全量预训练baichuan-7b Out of memory
#338 opened by FFFFFzx - 1
dpo_training训练chatglm3-6b模型报错。
#340 opened by xiaochaich - 2
大佬好,请教几个模型训练问题
#336 opened by xxyp - 2
训练完之后保存的时候出现无法连接,requests.exceptions.ConnectionError: (MaxRetryError("HTTPSConnectionPool(host='huggingface.co', port=443
#333 opened by josangmi - 2
预训练报错,之前有正常跑成功过,现在报这个错误,有人遇到过吗?谢谢啦
#326 opened by zxx20231119 - 4
Question about hardware requirements - 关于硬件要求的问题
#325 opened by elieobeid7 - 1
问题 flashattention
#329 opened by wuguangshuo - 3
这个错误是因为训练的单条数据太长了吗?截断是不是修改配置就可以了?
#328 opened by zxx20231119 - 1
SFT微调报错
#327 opened by ZhuangXialie