YuzaChongyi

Results 52 comments of YuzaChongyi

Live Streaming - Avoid regurgitating previously spoken details

Could you provide an example to illustrate what specific type of repetition you are referring to?

[BUG] flash_attn is needed but not specified in requirements

Thanks! We will look into the possibility of removing the dependency on flash attention.

[BUG] <title>模型只允许一句问题，没有历史消息/上下文

这里如果是使用 model.chat 接口的话，需要在传入的 msgs 里包含历史对话信息，streaming 模式在一定上下文长度内会复用 kv-cache，相当于已经记录了上下文

[BUG] `.to` is not supported for `4-bit` or `8-bit` bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct `dtype`.

你好，可以根据这个 [README](https://huggingface.co/openbmb/MiniCPM-o-2_6-int4#usage-of-minicpm-o-2_6-int4) 步骤安装 AutoGPTQ 使用 int4 量化推理 You can follow the steps in this [README](https://huggingface.co/openbmb/MiniCPM-o-2_6-int4#usage-of-minicpm-o-2_6-int4) to install AutoGPTQ and perform int4 quantized inference.

PVE 9.0 compatability

The model's download path must not contain any `.`. The `.` character will lead to failures during the dynamic importing process at runtime. For example, you can rename the folder/directory...

[BUG] <title>模型文件中resampler.py的代码错误

你好，可以更新一下 huggingface 仓库的最新代码

Use torch.compile with Minicpm-o

Thank you for your usage, torch.compile does have some room for compilation optimization and acceleration. We haven't actually tested it, so you could give it a try. We’d really appreciate...

[BUG] <title>使用官方示例代码报错

应该是文件下载的不全，可以尝试重新清除缓存后重新下载模型文件，这部分推理代码在 transformers=4.55.0 经过完整测试的

torch.distributed.elastic.multiprocessing.errors.ChildFailedError（exitcode: -9）反复在同一个进度出现

请问是用的哪一份微调代码进行训练的呢？ 目前的信息无法定位，如果稳定在同一个阶段出现比较像是数据问题

AssertionError: please make sure vector_quantize_pytorch and vocos are installed.

Please make sure these dependencies are installed，or you can run `pip install vector_quantize_pytorch vocos`