-
### Checklist
- [X] 1. I have searched related issues but cannot get the expected help.
- [X] 2. The bug has not been fixed in the latest version.
### Describe the bug
我按照教程运行example/vl/中的extract_x…
-
我看issue里好多问题都是需要更新最新版sat和代码,sat更新到0.4.4以上就报错,0.3.7可以正常运行,但是用0.3.7的代码看issue里就会报其他问题错误,请问如何解决?
具体报错信息:
For torch.distributed users or loading model parallel models, set environment variables RANK, W…
-
/root/autodl-tmp/XrayGLM/chatglm-6b tokenizer! Now you can get tokenizer() everywhere
其中ChatGLM-6B中的tokenizer()需要全部的权重吗?
-
跑web_demo.py时出现client error:
botocore.exceptions.ClientError: An error occurred (403) when calling the HeadObject operation: Forbidden
完整stack trace如下:
Failed to download or check, if you already…
-
-
运行环境 :
- CUDA11.7
- python:3.10
- torch 2.0
- 其余的按照requirments.txt安装
命令行输入`python webdemo.py`出现错误以下是错误的输出信息
`[2023-06-27 15:16:09,645] [INFO] [real_accelerator.py:110:get_accelerator] Settin…
-
您好,非常感谢您的项目,由于我自己的图片不需要224尺寸,128或96的都可以满足。请问我用小于224的图片能进行finetune吗?需要修改相关代码吗?谢谢!
-
![image](https://github.com/THUDM/VisualGLM-6B/assets/133647165/ce28f608-e44e-4b6d-ae3e-a7ebc8ddcfe8)
启动项目时下载sat模型,去了清华源下载,但是看其他issues说这个地址是旧的,我自己下载后启动包pad_token_id问题,与#188问题相同,请问怎么解决
-
### Is your feature request related to a problem? Please describe.
通过微调模型,支持视觉信息输入,类似于GPT-4。
目前已经在LLaMA语言模型上进行尝试,且结果不错。
参考:https://github.com/feizc/Visual-LLaMA
在迁移到chatglm的时候遇到问题,是否有计划让cha…
-
[2023-06-01 08:10:25,288] [INFO] [RANK 0] > number of parameters on model parallel rank 0: 7811237376
[2023-06-01 08:10:25,761] [INFO] [RANK 0] global rank 0 is loading checkpoint /root/.sat_models/…