issues
search
hiyouga
/
ChatGLM-Efficient-Tuning
Fine-tuning ChatGLM-6B with PEFT | 基于 PEFT 的高效 ChatGLM 微调
Apache License 2.0
3.66k
stars
471
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
微调后不起效果，是我哪步有问题吗
#406
RemSynch
closed
1 year ago
1
CUDA error: an illegal memory access was encountered
#405
suckseed5
closed
1 year ago
1
epoch的选择&数据集构造
#404
1028686314
closed
1 year ago
1
数据集oaast_rm_zh问题
#403
PluginsKers
closed
1 year ago
0
用默认参数微调chatglm2之后对话能力大幅度下降
#402
caoxueqian19901214
closed
1 year ago
1
step和epoch一样
#401
qingjiaozyn
closed
1 year ago
1
lora_target 的可用值有哪些？
#400
godcrying
closed
1 year ago
2
不理解为什么这个模型是paddingleft. 编码之后开始全是一堆-100的token, 感觉很难llm收敛.
#399
zhangbo2008
closed
1 year ago
0
仓库过大，git下载慢
#398
JackKuo666
closed
1 year ago
2
跑sft阶段出现这个问题，环境应该是ok的
#397
wangweihua11
closed
1 year ago
2
2400条数据,10个epoch,pre_seq_len=128,lora训练,为什么推理时显示trainable params: 0 || all params: 6243584000 || trainable%: 0.0000,是数据量不够吗,下边贴出训练参数
#396
zeze813
closed
1 year ago
1
PPO的log问题修复
#395
chosenone75
closed
1 year ago
3
请问下在4-bit量化模式Lora微调最小的显存要求
#394
NJU201220184
closed
1 year ago
1
Learning Scheduler Issue
#393
XingzhiZhou
closed
1 year ago
2
4*V100 32g配置，满足zero3全量微调吗？
#392
gongjunjin
closed
1 year ago
4
collator.py的第126行与preprocess.py的preprocess_supervised_dataset是不是存在冲突？input_ids拼接了2次labels
#391
jennfer0808
closed
1 year ago
2
ValueError: Cannot merge LORA layers when the model is loaded in 8-bit mode
#390
jeriff
closed
1 year ago
0
显存占用问题
#389
lileilai
closed
1 year ago
0
lora微调之后导出的完整模型文件应该要如何调用 我使用transformers的接口加载模型报错
#388
ecent2000
closed
1 year ago
2
请问lora训练是只会训练注意力层，在注意力层加适配器吗？而不训练前馈层。然后freeze是训练前馈层
#387
DRZJ1
closed
1 year ago
1
add related projects and wiki reference
#386
codemayq
closed
1 year ago
0
when `per_device_eval_batch_size` > 1 and launch by deepspeed, RuntimeError: Tensors must be contiguous
#385
jiahuanluo
closed
1 year ago
5
chatglm2进行lora微调，怎么执行断点续训呢？
#384
Qiang-HU
closed
1 year ago
2
ptuning后，模型参数不匹配
#383
HLF217
closed
1 year ago
0
为什么加载sft权重计算的指标和sft+ppo微调的权重计算的指标是一样的？
#382
JingyuHuAtVoxelCloud
closed
1 year ago
4
[BUG] PPO训练报错 RuntimeError: Expected all tensors to be on the same device
#381
qinqinqaq
closed
1 year ago
1
关于PPO中ref_model的问题
#380
JoyYChen
closed
1 year ago
9
为什么glm2训练需要较大的学习率才有效
#379
JingBob
closed
1 year ago
4
多卡微调错误：Exception: Could not find the transformer layer class to wrap in the model.
#378
lrx1213
closed
1 year ago
0
lora后模型的泛化能力如何增强呢？
#377
mawenju203
closed
1 year ago
2
请问如何制作自己的数据集？
#376
andyzhou1982
closed
1 year ago
4
请问用Lora微调2万条医疗问答，为什么Loss只能降低到2.7，效果不佳
#375
tongjialiang
closed
1 year ago
5
datasets.builder.DatasetGenerationError: An error occurred while generating the dataset
#374
ArtificialZeng
closed
1 year ago
0
请问int8版本的ppo怎么运行
#373
couldn
closed
1 year ago
2
关于sft之后，奖励模型和rlhf
#372
lzh1998-jansen
closed
1 year ago
3
单机4卡3090训练lora报错
#371
1006076811
closed
1 year ago
3
RuntimeError: expected scalar type Half but found Float
#370
jazzlee008
closed
1 year ago
3
使用train_sft_with_dev_set.sh进行 lora的微调 如何平衡 'loss' 和 'eval_loss'
#369
mawenju203
closed
1 year ago
2
LoRA微调已经训练好了，关于Evaluate和Predict的几个疑问
#368
FreeYiran
closed
1 year ago
1
多机多卡训练是否有相关资料呢？
#367
yifan-chen-2020
closed
1 year ago
2
建议对 data/*.json 文件使用 lfs 管理
#366
Haskely
closed
1 year ago
3
有关ChatGLM-Efficient-Tuning + lora微调出现过拟合的讨论
#365
mawenju203
closed
1 year ago
3
请问用example/里的quantized_finetune_with_local_model微调后怎么正确加载output并启动demo
#364
R0nnyz
closed
1 year ago
1
如何设置输入长度
#363
OpenHuShen
closed
1 year ago
5
训练不小心中断后，如何设置进行断点续训？
#362
FreeYiran
closed
1 year ago
9
在服务器上执行 src/train_web.py 报错
#361
FreeYiran
closed
1 year ago
4
模型训练的问题？ 模型出现过拟合
#360
mawenju203
closed
1 year ago
2
微信群满啦，麻烦更新下二维码
#359
hongshuo-wang
closed
1 year ago
1
在web_demo_finetuned.py运行ChatGLM（不是ChatGLM2）时报ValueError: not enough values to unpack (expected 3, got 2)
#358
yug030
closed
1 year ago
0
QLoRA报错expected scalar type Half but found Float
#357
wangweikang1229
closed
1 year ago
1
Next