While following the instructions in the repository, I noticed that the section below appears to refer to a download link, but no hyperlink is actually provided:
"First, download the preprocessed data source (curated from CAMEL-AI and Wiki, generated by Llama3-70B-Instruct) from this link. Unzip the data to data/tune/StruXGPT, which should be:"
Could you kindly update the instructions with the correct hyperlink for the data source?
Additionally, I would like to inquire about the following:
Are the models mentioned in the documentation (e.g., ./third_party/LLaMA-Factory/output/LongBench-SFT-struct-llama2, ./third_party/LLaMA-Factory/output/MMedBench-SFT-struct-llama2, ./weights/StruXGPT-v1-Qwen1_5-7B, and ./weights/StruXGPT-v2-Llama2-7B) going to be provided as part of the project, or do users need to source them independently?
Will the training code for these models be made available, or is it intended to remain private?
Thank you for your hard work and this excellent project. I look forward to your clarification and updates!
Hi, thanks for your attention. The data and pre-trained weights will be released after our company's review process. The training code has already been released in this repository.
Dear Maintainers,
I hope this message finds you well.
While following the instructions in the repository, I noticed that the section below appears to refer to a download link, but no hyperlink is actually provided:
"First, download the preprocessed data source (curated from CAMEL-AI and Wiki, generated by Llama3-70B-Instruct) from this link. Unzip the data to data/tune/StruXGPT, which should be:"
Could you kindly update the instructions with the correct hyperlink for the data source?
Additionally, I would like to inquire about the following:
Are the models mentioned in the documentation (e.g., ./third_party/LLaMA-Factory/output/LongBench-SFT-struct-llama2, ./third_party/LLaMA-Factory/output/MMedBench-SFT-struct-llama2, ./weights/StruXGPT-v1-Qwen1_5-7B, and ./weights/StruXGPT-v2-Llama2-7B) going to be provided as part of the project, or do users need to source them independently? Will the training code for these models be made available, or is it intended to remain private? Thank you for your hard work and this excellent project. I look forward to your clarification and updates!