luogen1996/LLaVA-HR issues and pull requests

#16 - Got loss 0 using ZeRO2 in sft stage

Issue - State: closed - Opened by Eddy-W 8 months ago

#15 - Error Loading Parameters for Slow Vision Tower During Fine-Tuning on Custom Dataset

Issue - State: open - Opened by buckerF 9 months ago

#14 - Any plan to release your stage1 pretraining checkpoint?

Issue - State: closed - Opened by zjr2000 9 months ago

#13 - 【CUDA Version】

Issue - State: open - Opened by Luo-Z13 10 months ago

#12 - unfreeze vision encoder on SFT

Issue - State: closed - Opened by syp2ysy 10 months ago

#11 - locally downloaded model files doesn't seem to be supported by the code, as it always requires online downloading

Issue - State: open - Opened by chuangzhidan 10 months ago - 1 comment

#10 - 4bit support?

Issue - State: open - Opened by chigkim 10 months ago

#9 - demo doesn't work : Error while finding module specification for 'llava.serve.cli' (ModuleNotFoundError: No module named 'llava')

Issue - State: open - Opened by chuangzhidan 10 months ago - 2 comments

#8 - requests.exceptions.ConnectTimeout: (MaxRetryError("HTTPSConnectionPool(host='huggingface.co', port=443): Max retries exceeded with url: /timm/convnext_xxlarge.clip_laion2b_soup_ft_in1k/resolve/main/pytorch_model.bin (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f48e9d64670>, 'Connection to huggingface.co timed out. (connect timeout=10)'))"), '(Request ID: 52084a1b-5e0c-4d52-bcfb-41bf34c52df6)')

Issue - State: closed - Opened by gapjialin 11 months ago - 15 comments

#7 - How to load convnext model using local bin file

Issue - State: open - Opened by yuecao0119 11 months ago - 7 comments

#6 - Understanding how MR-Adapter works

Issue - State: open - Opened by SoroushMehraban 11 months ago

#5 - Can we finetune LLaVA-HR with the scripts provided by the original LLaVA 1.5?

Issue - State: open - Opened by zengxingchen 11 months ago - 3 comments

#4 - the training parameters of your single branch convnext encoder

Issue - State: open - Opened by yuecao0119 11 months ago - 3 comments

#3 - Got loss 0 after traing with larger input size in stage 2 (sft)

Issue - State: open - Opened by luohao123 11 months ago - 4 comments

#2 - Update README.md

Pull Request - State: closed - Opened by eltociear 11 months ago

#1 - Ablation study on using just single path encoder?

Issue - State: open - Opened by lucasjinreal 11 months ago - 10 comments

Ecosyste.ms: Issues

GitHub / luogen1996/LLaVA-HR issues and pull requests