tya-math-ling-nb-t06

Same weights as rita-cohere/tya-math-ling-hyp-5327, with the notebook decode stack:

  • USER_INSTRUCTIONS (+ COT + /think) — no FORMAT-IS-CRITICAL / per-task hints
  • think 2048 @ T=0.6; force-close <|END_THINKING|><|START_RESPONSE|>
  • answer 512 @ T=0.6
  • marker-gated FINAL ANSWERS: parser

script.py reads /tmp/data/test.csv → writes submission.csv (id,pred).

Downloads last month
40
Safetensors
Model size
3B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for rita-cohere/tya-math-ling-nb-t06

Finetuned
(2)
this model