Steer, Don't Solve: Training Small Critic Models for Large Code Agents
Paper • 2606.21811 • Published • 1
Steer, Don't Solve: critic models and data
Critic models and SFT corpora from 'Steer, Don't Solve: Training Small Critic Models for Large Code Agents' (arXiv 2606.21811).
-
-
code-critic-model/Qwen3-8B-Critic-SFT
Text Generation • 1B • Updated • 341 -
code-critic-model/Qwen3-4B-Critic-SFT
Text Generation • 4B • Updated • 403 -
code-critic-model/Qwen3-4B-Critic-SFT-DPO
Text Generation • 4B • Updated • 373 -
code-critic-model/Qwen3-8B-Critic-SFT-CWM-only
Text Generation • 1B • Updated • 345 -
code-critic-model/Qwen3-8B-Critic-SFT-Qwen-only
Text Generation • 8B • Updated • 334 -
code-critic-model/Qwen3-4B-Critic-SFT-Qwen-only
Text Generation • 4B • Updated • 432 -
code-critic-model/Qwen3-8B-Critic-SFT-Detailed-Prompt
Text Generation • 1B • Updated • 340 -
code-critic-model/critic-sft-cwm-qwen
Viewer • Updated • 6.45k • 50 -
code-critic-model/critic-sft-cwm-only
Viewer • Updated • 4.53k • 44 -
code-critic-model/critic-sft-qwen-only
Viewer • Updated • 1.92k • 44 -
code-critic-model/critic-sft-cwm-only-detailed-prompt
Viewer • Updated • 3.14k • 166