AmirMohseni/curvebench-qwen3-vl-8b-only-tree

Add arxiv field to link model to paper page

Add Colab example notebook link

Increase max_tokens to 2048 in usage examples

Display training plots stacked for better readability

Fix model class for VLM (Qwen3VLForConditionalGeneration / AutoModelForImageTextToText)

Add reward plots, dataset load example, remove broken W&B badge

Add eval reward plot

Add training reward plot

Fix LoRA rank to r=4, remove VRAM note, clean up usage sections

Add CurveBench paper, datasets, LoRA usage guide

Model save

Training in progress, step 250

Training in progress, step 240

Training in progress, step 220

Training in progress, step 200

Training in progress, step 180

Training in progress, step 160

Training in progress, step 140

Training in progress, step 120

Training in progress, step 100

Training in progress, step 80

Training in progress, step 60

Training in progress, step 40

Training in progress, step 20

initial commit