Pradheep1647
/

kibitzer-clean-joint-scratch

state-space-model

Model card Files Files and versions

Pradheep1647/kibitzer-clean-joint-scratch

Kibitzer clean-rebuild checkpoint.

Training objective: policy_value
Checkpoint: joint_scratch_final.pt
Architecture: 32.1M-parameter position-only transformer/SSM hybrid
Policy output: 4,672 AlphaZero-style move logits
Value output: side-to-move scalar in [-1, 1]
Source: https://github.com/Mantissagithub/kibitzer/tree/clean-rebuild

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support