deepgeek/MiniCPM-o
2
0
Fork 0
mirror of https://github.com/OpenBMB/MiniCPM-V.git synced 2026-02-05 18:29:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

update for multi images sft

Browse Source
This commit is contained in:
qianyu chen
2024-08-13 13:57:48 +08:00
committed by GitHub
parent 61e942ec7c
commit 7842ec1228
5 changed files with 143 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
finetune/finetune_ds.sh
View File

@@ -13,7 +13,7 @@ MODEL="openbmb/MiniCPM-V-2_6"
DATA="path/to/trainging_data"
EVAL_DATA="path/to/test_data"
LLM_TYPE="qwen2" # if use openbmb/MiniCPM-V-2, please set LLM_TYPE=minicpm, if use openbmb/MiniCPM-Llama3-V-2_5, please set LLM_TYPE="llama3"
MODEL_MAX_Length=4096 # if use openbmb/MiniCPM-V-2 or openbmb/MiniCPM-Llama3-V-2_5, please set MODEL_MAX_Length=2048
DISTRIBUTED_ARGS="
@@ -39,7 +39,7 @@ torchrun $DISTRIBUTED_ARGS finetune.py \
--do_eval \
--tune_vision true \
--tune_llm true \
--model_max_length 2048 \
--model_max_length $MODEL_MAX_Length \
--max_slice_nums 9 \
--max_steps 10000 \
--eval_steps 1000 \