wang yaqi

> 操作系统：CentOS Linux release 7.9.2009 (Core) python版本：Python 3.11.5 项目版本：v0.2.10 向量化使用的：m3e-base 向量数据库使用的：faiss 模型部署模式： CPU 配置：32c、128g内存、500G SSD 启动方式：nohup python startup.py -a > out.log 2>&1 & > > **问题描述 / Problem Description** 我测试主机上单独部署chatglm3-6B、qwen-14B、gemma-7B都可以正常应答。...

> > _🤖_ > > 嗨，@daocaorenzzl！很高兴再次见到你。希望你最近一切都好！ > > 根据您的描述，这个问题可能是由于模型返回的结果不符合预期，导致了`TypeError: object of type 'NoneType' has no len()`这个错误。在Langchain-Chatchat的代码中，如果模型返回的结果是`None`，那么在尝试获取其长度时就会出现这个错误。 > > 在Langchain-Chatchat的相关问题中，我找到了一些可能的解决方案： > > > > 1. 确保GPU内存没有满。如果GPU内存已满，考虑将嵌入模型移到CPU上。参考问题：[minimax-api agent-chat对话报错ERROR: TypeError: Caught exception: object...

> 升级fschat到0.36.1 transformer最新 您好，谢谢回复。fschat最新是0.2.36 没有找到0.36.1这个版本， transformer最新是指huggingface的transformers最新吗

> It is indeed what you said. When I use pretrain coco.pth to convert into onnx then run my code, there is no problem at all, but the custom training...

> To clarify, Torch model given correct predictions, but conversion to ONNX makes them garbage? Yes!! Using a custom training model to make predictions through the following code can get...

> Thanks for the info, we'll dig into it > > @BloodAxe , @Louis-Dupont Thank you very much for your attention to this issue, looking forward to hearing from you!

> Quick update on this one, I finally got the correct result by **NOT** normalizing the image to [0, 1] for my custom model. On the contrary, by not normalizing...

> > > Quick update on this one, I finally got the correct result by **NOT** normalizing the image to [0, 1] for my custom model. On the contrary, by...

> Hello @Yakiw > > > Issue: response validation, token counting logic, model specification, missing tokenization, placeholder for the API URL, and insufficient error handling. > > > Enhancements: Include...

hello @neehhaa06 Thank you very much. I didn’t find an API in flowise that can calculate the consumed tokens.