OpenBMB / MiniCPM-V

MiniCPM-V 2.6: A GPT-4V Level MLLM for Single Image, Multi Image and Video on Your Phone
Apache License 2.0
11.82k stars 831 forks source link

[BUG] <title>ollama官方文档进行的编译部署,无法与模型对话 #453

Open Evelynn-n opened 1 month ago

Evelynn-n commented 1 month ago

是否已有关于该错误的issue或讨论? | Is there an existing issue / discussion for this?

该问题是否在FAQ中有解答? | Is there an existing answer for this in FAQ?

当前行为 | Current Behavior

当我进行对话的时候会出现Error: an unknown error was encountered while running the model

(torch) orangepi@orangepi5b:~/project/ollama$ ./ollama run minicpm2.6

你好 Error: an unknown error was encountered while running the model (torch) orangepi@orangepi5b:~/project/ollama$

image image

期望行为 | Expected Behavior

应该有正常的对话

复现方法 | Steps To Reproduce

No response

运行环境 | Environment

- OS:Ubuntu22.04
- Python:3.11
- Transformers:4.44.0
- PyTorch:2.4.0
- CUDA (`python -c 'import torch; print(torch.version.cuda)'`):No

备注 | Anything else?

No response

HSIAOKUOWEI commented 1 month ago

我也是,需要使用他們提供的ollama版本,然後重新編譯他們提供的ollama,才能運行他們的miniCPM,非常的複雜

tc-mb commented 1 month ago

Sorry, you may need to compile the code from our fork and enable the ollama service in the next few weeks to use it normally. We will submit a PR to ollama as soon as possible, but before that, you can use this readme for trial use. https://github.com/OpenBMB/ollama/blob/minicpm-v2.6/examples/minicpm-v2.6/README.md

Evelynn-n commented 4 weeks ago

就是根据那个飞书上的步骤,用的v2.6的fork编译的,但是运行不了

image
zhuchenxi commented 4 weeks ago

same question

mycodemylife commented 3 weeks ago

同样有这个问题,按照教程一路走下来的。