longertime
/

jmaxcool-ycle1

Model card Files Files and versions

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

BKN1890 DPO Cycle 1

On-policy DPO fine-tune from bkn1890-vision-only-v3 base (100 pairs, eval 653f269f).

Train: LoRA DPO, lr=5e-5, 2 epochs, 100 king-vs-challenger pairs
Fingerprint vs base v3: 0.942 (61 language shared_expert tensors changed)
Re-validated and re-uploaded: 2026-09-04

Downloads last month: 15

Safetensors

Model size

36B params

Tensor type

BF16

·

Inference Providers NEW

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