YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Think / no-think final model weights
12 final models covering UCF, SAT, Puzzle (Qwen2.5-VL-3B and Qwen3-VL-4B), SAT VisionOPD, and Math40k.
Each tar archive contains one final Hugging Face model directory with weights, configuration, and tokenizer/processor assets. Archive names preserve experiment and thinking-mode identifiers. Training optimizer states and intermediate checkpoints are excluded.
On macOS, verify downloads using shasum -a 256 -c SHA256SUMS, then extract with tar -xf ARCHIVE.tar.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support