# Feedback That Backfires: Why Small Language Model Agents Repeat the Call They Just Watched Fail arXiv: https://arxiv.org/abs/2608.23651 · HTML: https://mlsanity.com/paper/2608.23651/ ## Key facts - Cited by 4 papers in the map (data through 14 Oct 2026). - Placed in Tool learning and function calling, one of 16 directions in LLM agents and tool use. - Builds on 41 of its 48 references that are in the map. - First posted on arXiv 24 Aug 2026. ## Placed in - LLM agents and tool use / [Tool learning and function calling](https://mlsanity.com/agents/tool-learning-and-function-calling/) ## Builds on (most cited first, 30 of 41) - [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024) - [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025) - [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022) - [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024) - [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023) - [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021) - [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023) - [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023) - [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023) - [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019) - [WebArena: A Realistic Web Environment for Building Autonomous Agents](https://mlsanity.com/paper/2307.13854/) (2023) - [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024) - [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024) - [Gorilla: Large Language Model Connected with Massive APIs](https://mlsanity.com/paper/2305.15334/) (2023) - [AgentBench: Evaluating LLMs as Agents](https://mlsanity.com/paper/2308.03688/) (2023) - [GAIA: a benchmark for General AI Assistants](https://mlsanity.com/paper/2311.12983/) (2023) - [Large Language Models Cannot Self-Correct Reasoning Yet](https://mlsanity.com/paper/2310.01798/) (2023) - [Large Language Monkeys: Scaling Inference Compute with Repeated Sampling](https://mlsanity.com/paper/2407.21787/) (2024) - [Teaching Large Language Models to Self-Debug](https://mlsanity.com/paper/2304.05128/) (2023) - [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://mlsanity.com/paper/1909.05858/) (2019) - [Neural Text Generation with Unlikelihood Training](https://mlsanity.com/paper/1908.04319/) (2019) - [AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents](https://mlsanity.com/paper/2407.18901/) (2024) - [SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model](https://mlsanity.com/paper/2502.02737/) (2025) - [MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback](https://mlsanity.com/paper/2309.10691/) (2023) - [Small Language Models are the Future of Agentic AI](https://mlsanity.com/paper/2506.02153/) (2025) - [Efficient Guided Generation for Large Language Models](https://mlsanity.com/paper/2307.09702/) (2023) - [Can Large Language Models Really Improve by Self-critiquing Their Own Plans?](https://mlsanity.com/paper/2310.08118/) (2023) - [Learning to Break the Loop: Analyzing and Mitigating Repetitions for Neural Text Generation](https://mlsanity.com/paper/2206.02369/) (2022) - [Copy Suppression: Comprehensively Understanding an Attention Head](https://mlsanity.com/paper/2310.04625/) (2023) - [The Limits of Inference Scaling Through Resampling](https://mlsanity.com/paper/2411.17501/) (2024) ## Built on by (most cited first, 4 of 4) - [What Stops a Small Language Model From Driving a Database Agent](https://mlsanity.com/paper/2609.21341/) (2026) - [Jailbreak Context Lingers: Divergent Safety Routing and Its Cross-Task Predictability in Tool Agents](https://mlsanity.com/paper/2609.34686/) (2026) - [When History Fails to Become Experience: Action Calibration in Language Agents](https://mlsanity.com/paper/2610.02769/) (2026) - [Visible but Not Avoided: State-Dependent Re-Proposal of Rejected Candidates in Catalog-Based LLM Optimization Loops](https://mlsanity.com/paper/2610.15738/) (2026) ## Abstract (arXiv) Agent harnesses record a failed tool call and its error message in the transcript and ask the model to continue, on the assumption that the error is corrective information. We measure whether it is. Defining the corrective gain of a failure record as the change in log-probability of re-emitting the action that just failed, we find the gain is negative for every instruction-tuned model we tested (6 checkpoints, 135M-1.7B, 4 families) in two environments: simulated tool calling and MBPP program repair. Normalised by action length the effect is about -1.03 nats per action token, a factor of 2.8 in the odds of each token, and holds on 90%-100% of individual items, not only on average. Over a fixed candidate set the probability of repeating the failed call rises from 0.06 to 0.54, and greedy decoding reproduces it token for token on 19% of items after the failure versus 0% before. Counterfactuals pairing the same call with a failure message, a success message, or a neutral acknowledgement separate two effects: the failed call's surface form accounts for 83% of the damage, while the semantic contribution of marking it failed is small and inconsistent in sign across environments. The problem is in the harness, not the model's grasp of error messages, and that predicts which remedies work. Replacing the verbatim call with a runtime-generated description of the failure removes 76% of the inversion at no token cost, and making previously-failed strings unreachable at the decoder acts on the same term. Two plausible remedies do not: an explicit "do not repeat" instruction leaves the measured quantity where it was, and deleting the failed attempt to retry from a clean context, the standard prescription for context contamination, is the worst harness we measured for repetition, because it restores the context that produced the failure. The study runs end to end on a CPU; all artefacts are released.