psp_dada

When I perform batch inference, the output never stops normally and continues to generate until it reaches the length limit. How can I solve this problem? An example of the...

Specifying the version of `flash-attn` to install can resolve this issue. It seems that the latest version of `flash-attn` is incompatible with `torch==2.0.1` specified in the `pyproject.toml`. I was able...

I also found that llama3.1 8B performs very poorly when used to reason through math problems (from the MATH dataset). Even with the most basic vanilla CoT (Chain of Thought)...

> Hi, thanks for pointing out the problem. I checked the the original AMBER dataset, they didn't provide such suffix like "Please answer yes or no.". For consistency, we did...

I encountered the warning: "Expanding inputs for image tokens in LLaVa should be done in processing. Please add `patch_size` and `vision_feature_select_strategy` to the model's processing config or set directly with...

> `vision_feature_select_strategy = model.config.vision_feature_select_strategy` and `patch_size = model.config.vision_config.patch_size` Thank you for your response, but there is an issue. I use transformers 4.46.0.dev0 using `pip install --upgrade git+https://github.com/huggingface/transformers.git`, which means this...

I ues the following code to decode the output from the model.generate: ``` new_sentence = decoded_output.strip('').strip('').strip()[len(prompt) + 1:].strip('').strip() ``` An interesting phenomenon is that the original output was: ``` The...

> @pspdada I just opened a PR for that (#34332). It is a totally unrelated issue and should be solved soon Thank you for your attention, and I wish for...

> @pspdada I just opened a PR for that (#34332). It is a totally unrelated issue and should be solved soon I've discovered a new issue after the merge of...

> @pspdada that is the expected logic for the new processing code, as we expand the text with as many "image" tokens as there will be image embeddings. When decoding...