TIER: Trajectory-Invariant Execution Rewards for Multi-Step Tool Composition
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911690885431296 |
|---|---|
| author | Kulkarni, Anay Lu, ChiaEn Mekala, Dheeraj Srinivasa, Jayanth Liu, Gaowen Shang, Jingbo |
| author_facet | Kulkarni, Anay Lu, ChiaEn Mekala, Dheeraj Srinivasa, Jayanth Liu, Gaowen Shang, Jingbo |
| contents | Tool use enables large language models to solve complex tasks through sequences of API calls, yet existing reinforcement learning approaches fail to scale to multi-step composition settings. Outcome-based rewards provide only sparse feedback, while trajectory-supervised rewards depend on annotated reference solutions, penalizing valid alternatives and limiting scalability. We propose TIER: Trajectory-Invariant Execution Rewards, a reward framework that derives supervision directly from function schemas and runtime execution, rather than from reference trajectories. The reward decomposes into format validity, schema adherence, execution success, and answer correctness, providing dense, interpretable sequence-level feedback derived from fine-grained verification of individual steps of tool use. This design allows any valid execution path to receive credit, naturally supporting multiple solution strategies and adapting to evolving tool interfaces. On DepthBench, a compositional benchmark stratified by depth (1 to 6 steps), TIER achieves >90% accuracy across steps, where trajectory-supervised rewards collapse beyond step-4. We further demonstrate consistent gains on benchmarks like BFCL v3 and NestFUL. Ablation studies confirm that all reward components are necessary, highlighting the importance of multi-level supervision for compositional reasoning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_16790 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TIER: Trajectory-Invariant Execution Rewards for Multi-Step Tool Composition Kulkarni, Anay Lu, ChiaEn Mekala, Dheeraj Srinivasa, Jayanth Liu, Gaowen Shang, Jingbo Machine Learning Artificial Intelligence Computation and Language Tool use enables large language models to solve complex tasks through sequences of API calls, yet existing reinforcement learning approaches fail to scale to multi-step composition settings. Outcome-based rewards provide only sparse feedback, while trajectory-supervised rewards depend on annotated reference solutions, penalizing valid alternatives and limiting scalability. We propose TIER: Trajectory-Invariant Execution Rewards, a reward framework that derives supervision directly from function schemas and runtime execution, rather than from reference trajectories. The reward decomposes into format validity, schema adherence, execution success, and answer correctness, providing dense, interpretable sequence-level feedback derived from fine-grained verification of individual steps of tool use. This design allows any valid execution path to receive credit, naturally supporting multiple solution strategies and adapting to evolving tool interfaces. On DepthBench, a compositional benchmark stratified by depth (1 to 6 steps), TIER achieves >90% accuracy across steps, where trajectory-supervised rewards collapse beyond step-4. We further demonstrate consistent gains on benchmarks like BFCL v3 and NestFUL. Ablation studies confirm that all reward components are necessary, highlighting the importance of multi-level supervision for compositional reasoning. |
| title | TIER: Trajectory-Invariant Execution Rewards for Multi-Step Tool Composition |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2605.16790 |