Mao Shizhuo

in your script (named llama.py): **it reads "from torch_int.nn.linear import W8A8BFP32OFP32LinearWithSFactor, W8A8BFP32OFP32Linear"** **But there is no such class or function named "W8A8BFP32OFP32LinearWithSFactor" in linear.py in torch-int/nn/linear.py at all!!** ![Uploading image.png…]()

[Feature] Cannot find a huggingafce source code to reload model找不到加载hf模型的源代码

1

### Describe the feature 我需要加载自定义的INT8量化模型以开展数学推理能力的评测 我现在只需要把加载hf模型的源代码 model = Qwen2ForCausalLM.from_pretrained('/home/maoshizhuo/2025/deepseek-Qwen-1.5B', torch_dtype=torch.bfloat16, device_map='auto') 改成 model = Int8Qwen2ForCausalLM.from_pretrained('/home/maoshizhuo/smoothquant/int8_models/deepseek-Qwen-1.5B-smoothquant_ds_Qwen2_1.5B_2048', torch_dtype=torch.float16, device_map='auto') load_parameters(model,'./models/model_params_ds_qwen2_1.5B.pth') 就可以开始推理了。 但我找遍了opencompass/models下的模型，并没有模型使用huggingface的transformers库加载的，但是在使用手册里面是有这个的，并且--hf-path就是可以加载hf模型，但我找不到哪里有model = Qwen2ForCausalLM.from_pretrained或者AutoModelForCausalLM? ### Will you implement it? - [ ]...

[Bug] Take too much time on MATH-500 dataset evaluation

4

### Prerequisite - [x] I have searched [Issues](https://github.com/open-compass/opencompass/issues/) and [Discussions](https://github.com/open-compass/opencompass/discussions) but cannot get the expected help. - [x] The bug has not been fixed in the [latest version](https://github.com/open-compass/opencompass). ### Type...

### Prerequisite - [x] I have searched [Issues](https://github.com/open-compass/opencompass/issues/) and [Discussions](https://github.com/open-compass/opencompass/discussions) but cannot get the expected help. - [x] The bug has not been fixed in the [latest version](https://github.com/open-compass/opencompass). ### Type...