AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

BIRD-RL: Scaling Agentic Reinforcement Learning over Stateful Data-Centric Environments

Ge Qu, Jinyang Li, Xiaohan Xu, Xiaolong Li, Bowen Qin, Nan Huo, Ziwei Tang, Chenhao Ma, Reynold Cheng

NeurIPS 2026 · Reinforcement learning › Visual and representation RL · approach: Probabilistic and autonomous RL · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes BIRD-RL, an agentic reinforcement learning approach for language agents operating over live, stateful database environments, where success requires reasoning about evolving database states and repairing state-mutating operations.

Problem

Language agents must operate over live, stateful database environments, where success depends on reasoning about evolving database state, using execution feedback, and repairing state-mutating operations rather than only generating initial SQL.

Approach

Scales agentic reinforcement learning over stateful data-centric environments; the abstract is truncated, so the specific mechanism is not stated.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Stage 4 stateful interactive agentic RL after Stages 1-3 (ablation)BIRD-Talon-7B · BIRD-CRITIC-SQLiteSuccess Rate (%): 40.60 -> 44.40Improved
add Stage 3 adaptive reasoning RL on top of Stages 1-2 (ablation)BIRD-Talon-7B · BIRD-CRITIC-SQLiteSuccess Rate (%): 39.20 -> 40.60Improved
add Stage 2 tool-use imitation on top of Stage 1 (ablation)BIRD-Talon-7B · BIRD-CRITIC-SQLiteSuccess Rate (%): 35.20 -> 39.20Improved
use purely execution-based reward instead of adding partial-correctness proxy reward (ablation)BIRD-Talon-7B Stage 4 · BIRD-CRITIC-SQLiteSuccess Rate (%): 42.80 -> 44.40Improved
use purely execution-based reward instead of LLM-as-a-judge proxy reward (ablation)BIRD-Talon-7B Stage 4 · BIRD-CRITIC-SQLiteSuccess Rate (%): 41.40 -> 44.40Improved
use purely execution-based reward instead of heuristic rubric proxy reward (ablation)BIRD-Talon-7B Stage 4 · BIRD-CRITIC-SQLiteSuccess Rate (%): 42.00 -> 44.40Improved
replace per-rollout containers with trajectory-scoped reusable database poolBIRD-RL infrastructure · SQLite rollout, B=16, R=16Online database I/O: 2.94 TiB -> 236 GiB (12.8x)Improved
apply BIRD-RL training on top of SQL-Act scaffold and supervised BIRD-FixerQwen2.5-Coder-7B · BIRD-CRITIC-SQLiteSuccess Rate (%): 33.60 -> 44.40Improved
use SQL-Act multi-turn scaffold alone with untrained base modelQwen2.5-Coder-7B · BIRD-CRITIC-SQLiteSuccess Rate (%): 26.20 -> 29.80Improved
use multi-turn agentic inference with proprietary models instead of single-turnGeneral-purpose LLMs · BIRD-CRITIC-SQLiteSuccess Rate (%): Claude-Opus-4.6 46.20 -> 48.20; Qwen3-Coder-480B 41.60 -> 42.60Mixed
use debug-heavy 3:1 debugging-to-generation data mix rather than balanced or generation-heavy (ablation)BIRD-Zeno-7B · BIRD-CRITIC-SQLite and BIRD-MiniSuccess Rate / Execution AccuracyMixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy