internvl3_5-4b-iconclass-calib

LoRA adapter for OpenGVLab/InternVL3_5-4B-HF fine-tuned with ms-swift on davanstrien/iconclass-vlm-brillfull (2000 training rows, config sft).

epochs	1.0
learning rate	0.0001
LoRA rank / alpha	8 / 32
effective batch	16
max length / pixels	2048 / 1003520
vision tower	frozen
final train loss	0.8623245239257813
eval loss	1.1366941928863525
training time	21.5 min

Use

swift infer --model OpenGVLab/InternVL3_5-4B-HF --adapters davanstrien/internvl3_5-4b-iconclass-calib --use_hf true --stream true

Reproduction

Produced on Hugging Face Jobs (gpu) with the swift-vlm-sft.py recipe from uv-scripts. Run it yourself:

hf jobs uv run --flavor a10g-large --timeout 1h --secrets HF_TOKEN https://huggingface.co/datasets/uv-scripts/finetune/raw/main/swift-vlm-sft.py davanstrien/iconclass-vlm-brillfull davanstrien/internvl3_5-4b-iconclass-calib --model OpenGVLab/InternVL3_5-4B-HF --config sft --max-samples 2000 --epochs 1 --batch-size 4 --grad-accum 4 --work-dir /data/iconclass-internvl-calib