cpwan
Hi there, I suggest adding pre-trained models for document visual question answering (vqa). **Motivation** Document VQA is an important task in OCR. It recognizes texts region and finds their relationship....
any updates on this issue?
here is where the random prefix is generated. still trying to figure out how to change that... https://github.com/huggingface/frp/blob/b10131fda49c673c7aeaf604377051521cdc2334/server/control.go#L465
Same problem with `neuralmagic/DeepSeek-R1-Distill-Qwen-32B-quantized.w4a16`. Already 96 GB of memory. model: "neuralmagic/DeepSeek-R1-Distill-Qwen-32B-quantized.w4a16" tensor-parallel-size: 2 max-model-len: 32768 max-num-seqs: 4
Similarly, for 8B model. 1. Cannot relate the huggingface reported json with the benchmark page. 2. Discrepancy between my own runs with the huggingface reported one. [Benchmark](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=meta&pinned=meta-llama%2FMeta-Llama-3.1-8B-Instruct_bfloat16_df34336b42332c6d360959e259cd6271c6a09fd4_True) on mmlu pro:...
Not sure, the maintainer does not respond at all.
@crazywoola @JohnJyong Ready for review ## Multi-page ### pdf:  ### docx **the docx format may not give page number info, so it is always 0.**  ## Single page...
It seems that the page number was removed in https://github.com/langgenius/dify/issues/8211 . Wonder if there is any chance to bring it back? @JohnJyong Like, don't need to include the `page` in...
@dosu This issue should not be closed... the request of page number attribute keeps arising from the issues. https://github.com/langgenius/dify/issues/11891