issues
search
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。
Apache License 2.0
2.94k
stars
451
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
chatglm2合并sft_qlora后,推理出现自动续答
#339
Lxhnnn
closed
3 months ago
3
全量预训练baichuan-7b Out of memory
#338
FFFFFzx
closed
3 months ago
3
请问这是什么oserror
#337
cove1011
closed
3 months ago
3
大佬好,请教几个模型训练问题
#336
xxyp
closed
3 months ago
2
Please provide an offload_folder
#335
cove1011
closed
3 months ago
1
DPO阶段报错
#334
small-white-zs
closed
3 months ago
5
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
josangmi
closed
4 months ago
2
chatglm2-6b预训练合并权重后,推理加载,AttributeError: can't set attribute
#332
weedon666
closed
3 months ago
2
全参数SFT后无法正常infer
#331
nuoma
closed
4 months ago
3
单机多卡sft deepspeed zero3 训练一直卡在训练阶段
#330
lainxx
opened
4 months ago
1
问题 flashattention
#329
wuguangshuo
closed
5 months ago
1
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
zxx20231119
closed
5 months ago
3
SFT微调报错
#327
ZhuangXialie
closed
5 months ago
1
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
zxx20231119
closed
4 months ago
2
Question about hardware requirements - 关于硬件要求的问题
#325
elieobeid7
closed
4 months ago
4
全量微调,没有用lora,之后的参数怎么合并到原模型中
#324
WangZY1111
closed
4 months ago
1
11pul
#323
yuwangnexusera
closed
4 months ago
1
扩充词表后,预训练完成进行lora合并时出现size mismatch问题,请问这个问题该怎么解决
#322
lainxx
closed
3 months ago
1
Update transformers in requirements.txt
#321
dividez
closed
5 months ago
0
llama2,sft全参训练,fp32,loss若干step后降为0
#320
wangrx33
closed
5 months ago
0
2机器16卡出现的问题
#319
listwebit
closed
5 months ago
1
求助:2机16卡预训练出bug问题,请大佬帮忙看看
#318
listwebit
closed
3 months ago
8
增量预训练 deepspeed模式,小模型可以run,大一些的模型报错
#317
listwebit
closed
5 months ago
1
pt阶段,是否支持3D并行呢
#316
listwebit
closed
5 months ago
1
请问,pt阶段,基础模型比较大(Yi-67B),多机多卡用那种训练比较好呢?
#315
listwebit
opened
5 months ago
1
增量预训练时候,用大一些的模型(Yi-34B)在8*80G服务器上全量训练,无论-block_size 设置多少都报显存不足
#314
listwebit
closed
5 months ago
2
LOSS下降趋势呈阶梯状,有明显【断崖下降】痕迹,请问,有头绪么
#313
Thunderltx
closed
5 months ago
3
PPO的训练代码里,预处理函数把数据切成了单字,这是正常的吗?
#312
chiquitita-101
closed
1 month ago
4
合并多个模型后,合并模型处理请求会卡死
#311
growmuye
closed
5 months ago
7
预训练的模型效果比之原始模型差太多
#310
lljpwrs
closed
5 months ago
2
为什么指定gpu只能在0,1卡,换成2,3卡就卡住了?
#309
sxl1993
closed
5 months ago
1
模型inference输出不能复现
#308
wangrx33
closed
5 months ago
1
用bloom没问题,但是切换ChatGLM2后抛了这个异常
#307
yuwangnexusera
closed
5 months ago
1
这是为什么啊?!用指令SFT后,模型的回答都变得短了,短就算了,还不对, 新的内容没学习进去
#306
Thunderltx
closed
5 months ago
4
增量训练的时候,执行run_pt.sh 报错
#305
listwebit
closed
5 months ago
1
请问在进行增量预训练的时候,加载的模型必须是base模型吗,可以是微调过的chat模型吗?
#304
listwebit
closed
5 months ago
1
请问可以支持yuan-2.0的51B模型支持增量预训练(持续预训练)吗,请大佬回复
#303
listwebit
closed
4 months ago
1
使用 llama2-7b的模型调用 pretraining.py进行预训练之后,在使用 inference.py进行 inference 时--template_name应该填什么?
#302
lljpwrs
closed
5 months ago
1
进行预训练的“model_name_or_path”参数也是HF格式的吗
#301
Peter-of-Astora
closed
5 months ago
3
大佬,reward_model的训练是不是也不支持chatglm。
#300
LanShanPi
closed
5 months ago
3
colab乱码,求大佬解答
#299
small-white-zs
closed
6 months ago
6
跑DPO的时候出现下面的问题。
#298
LanShanPi
closed
5 months ago
4
大佬,使用多卡3090跑baichuan2-13b时,感觉模型好像没有分布到各个显卡上,显存一下就满了oom了。怎么解决?
#297
tuqingwen
closed
5 months ago
1
yi-6B sft loss为0
#296
nuoma
closed
6 months ago
3
运行Demo出现问题:AttributeError: 'ChatGLMTokenizer' object has no attribute 'sp_tokenizer'. Did you mean: '_tokenize'?
#295
Peter-of-Astora
closed
5 months ago
7
运行python supervised_finetuning.py,发生KeyError: 'conversation'
#294
ospreyclaw
closed
5 months ago
3
请教DPO多轮对话的问题
#293
chloefresh
opened
6 months ago
3
推理不适用指定的卡
#292
nuoma
closed
6 months ago
2
在单机多卡监督微调时使用的策略是DP还是DDP?
#291
CNUIGB
opened
6 months ago
1
合并词表后运行pt报错
#290
tuqingwen
closed
5 months ago
2
Previous
Next