Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BUG/Help] 求大佬看下,微调多轮对话,预测时出现130000 is not in list问题 #432

Closed
1 task done
q497629642 opened this issue Apr 7, 2023 · 38 comments
Closed
1 task done

Comments

@q497629642
Copy link

Is there an existing issue for this?

  • I have searched the existing issues

Current Behavior

screenshot-20230407-113029

Expected Behavior

No response

Steps To Reproduce

使用AutoTokenizer、AutoModel加载微调好的多轮对话checkpoint,使用stream_chat预测

Environment

- OS:
- Python:
- Transformers:
- PyTorch:
- CUDA Support (`python -c "import torch; print(torch.cuda.is_available())"`) :

Anything else?

No response

@Leawnn
Copy link

Leawnn commented Apr 7, 2023

应该是大佬们昨晚上在hf上更新了新模型和代码的原因,重新下载拉取试试。我也是碰到这个问题

@q497629642
Copy link
Author

我试试

@q497629642
Copy link
Author

应该是大佬们昨晚上在hf上更新了新模型和代码的原因,重新下载拉取试试。我也是碰到这个问题

试了,还是报那个问题

@zwkkk
Copy link

zwkkk commented Apr 7, 2023

遇到这个问题+1

@duzx16
Copy link
Member

duzx16 commented Apr 7, 2023

你要更新你的checkpoint目录里面的ice_text.modeltokenization_chatglm.py

@tutuxxx
Copy link

tutuxxx commented Apr 7, 2023

你要更新你的checkpoint目录里面的ice_text.modeltokenization_chatglm.py

请问怎么更新能说具体点吗?昨晚是更新了代码,重新下载了模型,然后花了4个多小时训练的,有办法补救吗

@czb7046
Copy link

czb7046 commented Apr 7, 2023

150001 not in the list, reinstall everthing still not work.

@q497629642
Copy link
Author

你要更新你的checkpoint目录里面的ice_text.modeltokenization_chatglm.py

更新了,还是报错

@duzx16
Copy link
Member

duzx16 commented Apr 7, 2023

你要更新你的checkpoint目录里面的ice_text.modeltokenization_chatglm.py

请问怎么更新能说具体点吗?昨晚是更新了代码,重新下载了模型,然后花了4个多小时训练的,有办法补救吗

把加载模型的目录下的这两个文件替换了,从 https://huggingface.co/THUDM/chatglm-6b 下载

@zwkkk
Copy link

zwkkk commented Apr 7, 2023

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004]
inputs 类型#上衣材质#牛仔布颜色#白色风格#简约图案#刺绣衣样式#外套衣款式#破洞
label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004]
labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。
全更新了,这tokenizer之后貌似啥mask也没有

@bsmarcebte733
Copy link

你要更新你的checkpoint目录里面的ice_text.modeltokenization_chatglm.py

请问怎么更新能说具体点吗?昨晚是更新了代码,重新下载了模型,然后花了4个多小时训练的,有办法补救吗

把加载模型的目录下的这两个文件替换了,从 https://huggingface.co/THUDM/chatglm-6b 下载

依然无法解决

@q497629642
Copy link
Author

你要更新你的checkpoint目录里面的ice_text.modeltokenization_chatglm.py

请问怎么更新能说具体点吗?昨晚是更新了代码,重新下载了模型,然后花了4个多小时训练的,有办法补救吗

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004] inputs 类型#上衣_材质#牛仔布_颜色#白色_风格#简约_图案#刺绣_衣样式#外套_衣款式#破洞 label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004] labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。 全更新了,这tokenizer之后貌似啥mask也没有

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004] inputs 类型#上衣_材质#牛仔布_颜色#白色_风格#简约_图案#刺绣_衣样式#外套_衣款式#破洞 label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004] labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。 全更新了,这tokenizer之后貌似啥mask也没有

估计就是分词还是有问题

@duzx16
Copy link
Member

duzx16 commented Apr 7, 2023

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004] inputs 类型#上衣_材质#牛仔布_颜色#白色_风格#简约_图案#刺绣_衣样式#外套_衣款式#破洞 label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004] labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。 全更新了,这tokenizer之后貌似啥mask也没有

不好意思,还需要更新一下 tokenizer_config.json。150001就是gmask,decode的时候本来就是不显示的。更新之后应该是130001,去掉了前20000个不用的image token

@duzx16 duzx16 pinned this issue Apr 7, 2023
@Leawnn
Copy link

Leawnn commented Apr 7, 2023

我可以了,把那几个文件更新之后,需要tokenizer 也要换成自己的chepoint路径(还要把.cache/huggingface/modules/transformers_modules/checkpoint-1000/缓存删掉)
tokenizer = AutoTokenizer.from_pretrained("./output/XXX-chatglm-6b-pt-8-1e-2/checkpoint-1000", trust_remote_code=True)
model = AutoModel.from_pretrained("./output/XXX-chatglm-6b-pt-8-1e-2/checkpoint-1000", trust_remote_code=True).half().cuda()

@zwkkk
Copy link

zwkkk commented Apr 7, 2023

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004] inputs 类型#上衣_材质#牛仔布_颜色#白色_风格#简约_图案#刺绣_衣样式#外套_衣款式#破洞 label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004] labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。 全更新了,这tokenizer之后貌似啥mask也没有

不好意思,还需要更新一下 tokenizer_config.json。150001就是gmask,decode的时候本来就是不显示的。更新之后应该是130001,去掉了前20000个不用的image token

ok了,是这个问题

@q497629642
Copy link
Author

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004] inputs 类型#上衣_材质#牛仔布_颜色#白色_风格#简约_图案#刺绣_衣样式#外套_衣款式#破洞 label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004] labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。 全更新了,这tokenizer之后貌似啥mask也没有

不好意思,还需要更新一下 tokenizer_config.json。150001就是gmask,decode的时候本来就是不显示的。更新之后应该是130001,去掉了前20000个不用的image token

感谢,可以了

@XiaTiaoQAQ
Copy link

可以解决,谢谢!

@fiona-aa
Copy link

fiona-aa commented Apr 7, 2023

重新下载了模型和config文件,仍然报错

@fiona-aa
Copy link

fiona-aa commented Apr 7, 2023

[5, 65421, 61, 67329, 32, 98339, 61, 72043, 32, 65347, 61, 70872, 32, 69768, 61, 68944, 32, 67329, 64103, 61, 96914, 0, 0, 5, 87052, 96914, 81471, 64562, 65759, 64493, 64988, 6, 65840, 65388, 74531, 63825, 75786, 64009, 63823, 65626, 63882, 64619, 65388, 6, 64480, 65604, 85646, 110945, 10, 64089, 65966, 87052, 67329, 65544, 6, 71964, 70533, 64417, 63862, 89978, 63991, 63823, 77284, 88473, 64219, 63848, 112012, 6, 71231, 65099, 71252, 66800, 85768, 64566, 64338, 100323, 75469, 63823, 117317, 64218, 64257, 64051, 74197, 6, 63893, 0]
默认的AdvertiseGen数据集跑的

@shibing624
Copy link

下载最新权重文件,运行evaluate.sh 后依然报错,应该是150001未变为130001导致的。

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004]

@duzx16
Copy link
Member

duzx16 commented Apr 7, 2023

[5, 65421, 61, 67329, 32, 98339, 61, 72043, 32, 65347, 61, 70872, 32, 69768, 61, 68944, 32, 67329, 64103, 61, 96914, 0, 0, 5, 87052, 96914, 81471, 64562, 65759, 64493, 64988, 6, 65840, 65388, 74531, 63825, 75786, 64009, 63823, 65626, 63882, 64619, 65388, 6, 64480, 65604, 85646, 110945, 10, 64089, 65966, 87052, 67329, 65544, 6, 71964, 70533, 64417, 63862, 89978, 63991, 63823, 77284, 88473, 64219, 63848, 112012, 6, 71231, 65099, 71252, 66800, 85768, 64566, 64338, 100323, 75469, 63823, 117317, 64218, 64257, 64051, 74197, 6, 63893, 0] 默认的AdvertiseGen数据集跑的

更新ice_text.model

@duzx16
Copy link
Member

duzx16 commented Apr 7, 2023

下载最新权重文件,运行evaluate.sh 后依然报错,应该是150001未变为130001导致的。

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004]

更新ice_text.model, tokenization_chatglm.pytokenizer_config.json

@mvllwong
Copy link

mvllwong commented Apr 7, 2023

哎,还是想吐槽一句。hf的代码更新了,本地代码的复制代码处理方式都不会同步更新到,sentencepiece会跑不起来

@bigbrother666sh
Copy link

不好意思,我是加载的 int4qe模型,也出现了 150001 is not in list的报错,我看了大家上面的讨论,但我没有在chatglm-6b-int4-qe文件夹中找到ice_text.model, tokenization_chatglm.py 和 tokenizer_config.json这三个文件啊?
是修改哪里 (我一个小时前刚刚删除了这个文件夹重新从hf上进行了下载……)

@LiuChen19960902
Copy link

下载最新权重文件,运行evaluate.sh 后依然报错,应该是150001未变为130001导致的。
input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004]

更新ice_text.model, tokenization_chatglm.pytokenizer_config.json

我今天重新下了一次所有的文件,并且重新pull了一次代码,但是微调还是130001 is not in list,/(ㄒoㄒ)/~~

@duzx16
Copy link
Member

duzx16 commented Apr 8, 2023

不好意思,我是加载的 int4qe模型,也出现了 150001 is not in list的报错,我看了大家上面的讨论,但我没有在chatglm-6b-int4-qe文件夹中找到ice_text.model, tokenization_chatglm.py 和 tokenizer_config.json这三个文件啊? 是修改哪里 (我一个小时前刚刚删除了这个文件夹重新从hf上进行了下载……)

https://huggingface.co/THUDM/chatglm-6b-int4-qe/tree/main

@xiaoyaolangzhi
Copy link

出现130000有两种情况:
第一种情况是推理预训练模型时出现的,这种情况是预训练模型的相关文件没有更新导致的,下载最新的ce_text.model, tokenization_chatglm.py 和 tokenizer_config.json替换掉。
第二种情况是推理微调模型时出现的,这种情况是生成的微调模型的相关文件是旧版本导致的,复制预训练模型下的ce_text.model, tokenization_chatglm.py 和 tokenizer_config.json替换掉output下对应文件。
不知道为什么微调时一直生成旧版本的文件,删除了.cache/huggingface/modules/transformers_modules下的缓存无效

@shibing624
Copy link

@xiaoyaolangzhi 感谢,用第二种情况cp了之后能正常推理了。

商品广告文案能正常预测了,但‘介绍下北京’之类的正常的对话能力严重退化,P-tuning方法还是会造成模型能力变窄甚至退化,我用LoRA方法不会影响原始模型预测能力。

@duzx16
Copy link
Member

duzx16 commented Apr 8, 2023

出现130000有两种情况: 第一种情况是推理预训练模型时出现的,这种情况是预训练模型的相关文件没有更新导致的,下载最新的ce_text.model, tokenization_chatglm.py 和 tokenizer_config.json替换掉。 第二种情况是推理微调模型时出现的,这种情况是生成的微调模型的相关文件是旧版本导致的,复制预训练模型下的ce_text.model, tokenization_chatglm.py 和 tokenizer_config.json替换掉output下对应文件。 不知道为什么微调时一直生成旧版本的文件,删除了.cache/huggingface/modules/transformers_modules下的缓存无效

@xiaoyaolangzhi 之前的tokenizer在保存config的时候有一个bug,已经修复了,多谢指。之前保存的checkpoint还需要手动更新一下。

@AItechnology
Copy link

上面提到的checkpoint怎么手动更新呢?

@duzx16
Copy link
Member

duzx16 commented Apr 14, 2023

#596 之前有一个bug就是数据里如果有 [MASK] 会被 tokenizer 当作特殊token,导致 [gMASK] 没有加上。目前已经修复了

@Skywalker-Harrison
Copy link

出现130000有两种情况: 第一种情况是推理预训练模型时出现的,这种情况是预训练模型的相关文件没有更新导致的,下载最新的ce_text.model, tokenization_chatglm.py 和 tokenizer_config.json替换掉。 第二种情况是推理微调模型时出现的,这种情况是生成的微调模型的相关文件是旧版本导致的,复制预训练模型下的ce_text.model, tokenization_chatglm.py 和 tokenizer_config.json替换掉output下对应文件。 不知道为什么微调时一直生成旧版本的文件,删除了.cache/huggingface/modules/transformers_modules下的缓存无效

在预测的时候出错了,下载了最新的文件,也没有效果。output文件夹下面没有找到tokenization_chatglm.py和tokenizer_config.json. 这个只有model文件夹有

@seasnakes
Copy link

input_ids [20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20003, 20005, 85421, 20061, 95898, 20032, 88554, 20061, 97257, 84555, 20032, 85107, 20061, 86268, 20032, 85347, 20061, 91689, 20032, 89768, 20061, 105428, 20032, 85173, 93942, 20061, 90984, 20032, 85173, 90936, 20061, 84703, 85509, 150001, 150004] inputs 类型#上衣_材质#牛仔布_颜色#白色_风格#简约_图案#刺绣_衣样式#外套_衣款式#破洞 label_ids [20005, 91689, 86561, 87061, 97257, 90984, 20006, 92194, 85173, 84290, 84622, 101549, 83823, 85173, 84290, 103343, 83832, 83912, 85209, 84703, 85509, 84051, 20006, 89418, 98598, 107019, 20006, 84257, 91319, 86069, 94197, 83823, 85173, 92265, 84880, 84131, 83832, 93416, 105428, 86261, 20006, 85594, 107834, 20006, 93412, 125145, 85388, 83823, 150001, 150004] labels 简约而不简单的牛仔外套,白色的衣身十分百搭。衣身多处有做旧破洞设计,打破单调乏味,增加一丝造型看点。衣身后背处有趣味刺绣装饰,丰富层次感,彰显别样时尚。 全更新了,这tokenizer之后貌似啥mask也没有

不好意思,还需要更新一下 tokenizer_config.json。150001就是gmask,decode的时候本来就是不显示的。更新之后应该是130001,去掉了前20000个不用的image token

怎么更改 是去tokenizer_config.json 里 "gmask_token": "[gMASK]", [gMASK] 改成130001吗?

@shibing624
Copy link

下载最新权重文件 https://huggingface.co/THUDM/chatglm-6b ,这个问题已经修复了。

@ttw1018
Copy link

ttw1018 commented Apr 27, 2023

remove add_special_tokens=False

@dkqkxx
Copy link

dkqkxx commented May 29, 2023

如果使用pipeline,可能需要修改 ../python3.9/site-packages/transformers/pipelines/text_generation.py:203

def preprocess(self, prompt_text, prefix="", handle_long_generation=None, **generate_kwargs):
    inputs = self.tokenizer(
        prefix + prompt_text, padding=False, add_special_tokens=True, return_tensors=self.framework
    )

@shnyyds
Copy link

shnyyds commented Aug 16, 2023

chatglm2-6b也报错

@jxaizj
Copy link

jxaizj commented Aug 17, 2023 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests