sql-agent-q35-2b-r1012
Final HF snapshot (hf/step_29/) of a short multi-turn SQL-agent GRPO run of
Qwen/Qwen3.5-2B@15852e8c16360a2fea060d615a32b45270f8a8fc on Spider databases with Miles v0.1.0 (SGLang rollouts,
thinking enabled, one execute_sql tool, up to 8 turns, 30 rollout steps, cost-aware reward). Same layout as the base
model; loadable by SGLang and vLLM. Part of an AgentRL-Bench task environment.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support