423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-16 20:00:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

add multimodal LLM BLIP-2 and InstructBLIP

Browse Source
This commit is contained in:
BUAADreamer
2024-04-23 19:22:42 +08:00
parent 4f3d558f67
commit e1afbea68f
3 changed files with 4 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
examples/mllm/sft_blip2.sh
View File

@@ -14,7 +14,7 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--overwrite_output_dir \
--cutoff_len 1024 \
--preprocessing_num_workers 16 \
--per_device_train_batch_size 1 \
--per_device_train_batch_size 4 \
--per_device_eval_batch_size 1 \
--gradient_accumulation_steps 8 \
--lr_scheduler_type cosine \
@@ -30,5 +30,4 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--val_size 0.1 \
--plot_loss \
--quantization_bit 8 \
--image_path /home/LAB/fengzc/LLM/checkpoints/liuhaotian/LLaVA-Instruct-150K/images/coco/train2017
--image_path /home/LAB/fengzc/LLM/checkpoints/liuhaotian/LLaVA-Instruct-150K/images/coco/train2017