mbzuai-oryx/LLaVA-pp issues and pull requests

#35 - How can I continue training on the pre-trained llava-llama3 model instead of training llama3 directly with my own data?

Issue - State: open - Opened by ganliqiang 29 days ago - 1 comment

#34 - Phi 3 Mini 128K leads to Tokenization Mismatch

Issue - State: open - Opened by ritwickchaudhry 3 months ago - 2 comments

#33 - INtegrate gemma 9B with siglip 512 px patch

Issue - State: open - Opened by Jayantverma2 4 months ago

#32 - Fixed issue with merging pretrain with Phi-3

Pull Request - State: open - Opened by alanrbtx 4 months ago

#31 - ValueError: Trying to set a tensor of shape torch.Size([128257, 4096]) in "weight" (which has shape torch.Size([128256, 4096])), this look incorrect.

Issue - State: open - Opened by basteran 6 months ago - 1 comment

#30 - How much memory are needed for fine-tuning llava-llama3

Issue - State: open - Opened by simplelifetime 6 months ago

#29 - Finetuning with lora output never ends.

Issue - State: open - Opened by gyupro 6 months ago - 5 comments

#28 - Total Parameters: 0

Issue - State: closed - Opened by believewhat 6 months ago

#27 - convert_to_hf

Issue - State: open - Opened by Jayantverma2 6 months ago

#26 - Saved weights during tuning with LoRA

Issue - State: open - Opened by fangruizhu 6 months ago

#25 - 【Error when merging LoRA weights】

Issue - State: open - Opened by Luo-Z13 6 months ago - 3 comments

#24 - finetune error about model size

Issue - State: closed - Opened by Skylight-Lark 6 months ago - 5 comments

#23 - inference error

Issue - State: closed - Opened by tsw123678 6 months ago - 1 comment

#22 - 'PreTrainedTokenizerFast' object has no attribute 'legacy'

Issue - State: closed - Opened by tsw123678 6 months ago - 3 comments

#21 - could you please support deepseek v2？

Issue - State: open - Opened by thesby 6 months ago
Labels: enhancement

#20 - what is s2 in vision load tower and how it effects training

Issue - State: open - Opened by Jayantverma2 6 months ago - 1 comment

#19 - Training Issue

Issue - State: open - Opened by DevonPeroutky 6 months ago - 1 comment

#18 - Can use Lora+base model. but for merging Lora+base is error

Issue - State: open - Opened by hellangleZ 6 months ago - 4 comments

#17 - Tokenization mismatch in Phi-3 when finetune process

Issue - State: closed - Opened by hellangleZ 6 months ago - 5 comments

#16 - Update README.md

Pull Request - State: open - Opened by jbn 6 months ago

#15 - Llava 1.6?

Issue - State: open - Opened by ElliottDyson 6 months ago - 4 comments
Labels: enhancement

#14 - Tokenization mismatch in LLaVA-LLaMA-3

Issue - State: closed - Opened by Luo-Z13 6 months ago - 2 comments

#13 - Exactly same as our study! Interesting!

Issue - State: closed - Opened by mu-cai 7 months ago - 2 comments

#12 - Installation Issue

Issue - State: open - Opened by orrzohar 7 months ago - 2 comments

#11 - zero3.json is missing

Issue - State: closed - Opened by matbee-eth 7 months ago - 1 comment

#10 - Can't load the HF model

Issue - State: closed - Opened by vfragoso 7 months ago - 1 comment

#9 - S2 finetuning

Issue - State: closed - Opened by xmu-xiaoma666 7 months ago - 1 comment

#8 - Train with llava-llama3

Issue - State: closed - Opened by hellangleZ 7 months ago - 9 comments

#7 - Using phi-3 and LLava but some fields of phi3 network not support

Issue - State: closed - Opened by hellangleZ 7 months ago - 21 comments

#6 - inference score with ScienceQA

Issue - State: closed - Opened by xmu-xiaoma666 7 months ago

Ecosyste.ms: Issues

GitHub / mbzuai-oryx/LLaVA-pp issues and pull requests