使用 GPUStack 部署时下载 CSGHub 模型文件报错
使用huggingface sdk 调用https://hub.opencsg.com/hf/api/models/Qwen/Qwen3-235B-A22B/revision/main?blobs=true接口没有返回文件大小,导致 gpustack 下载文件报错
@KinglyWayne pls assign someone to work in on. Thanks
@reboot-cloud how did you download the model? it works for me with following script
(base) xiangzhen@james-host Downloads % cat download.py from huggingface_hub import snapshot_download snapshot_download(repo_id="Qwen/Qwen3-235B-A22B",endpoint="https://hub.opencsg.com/hf")
(base) xiangzhen@james-host Downloads % pip list|grep transformers transformers 4.45.2
(base) xiangzhen@james-host Downloads % pip list|grep huggingface huggingface-hub 0.26.1
just have a quck check, it seems csghub api missing some fields, like size pointerSize https://huggingface.co/api/models/Qwen/Qwen3-235B-A22B/revision/main?blobs=true
@pulltheflower please have a look
@Rader Got it.
@reboot-cloud Thank you for your feedback — it's very valuable. After investigation, we found that the issue was caused by our lack of support for the blobs parameter. We've made the necessary changes, and this problem will be fixed in our next release.
@reboot-cloud Our SaaS has been updated, could you please try downloading again and see if it still gets this error. Thanks again for your report!