wehub-resource-sync
9 lines
250 B
Bash
9 lines
250 B
Bash
CUDA_VISIBLE_DEVICES=0,1 \
|
|
swift infer \
|
|
--model OpenGVLab/InternVL2_5-1B \
|
|
--infer_backend lmdeploy \
|
|
--val_dataset AI-ModelScope/captcha-images#1000 \
|
|
--lmdeploy_tp 2 \
|
|
--lmdeploy_vision_batch_size 8 \
|
|
--max_new_tokens 2048
|