Artificial Intelligence · arXiv:2610.07979 v1 · submitted 6 Oct 2026 · 50 references
Learning from Revision Consequences: Hindsight Meta-Experience Distillation for Self-Improving Agents
The paper says
“As agents continuously improve by generating and revising Skills, the process that discovers and refines those Skills becomes a learnable object in its own right.”
more of the abstract
Task-Skills directly act on task execution, whereas Meta-Skills govern how agents discover and improve future Skills; their value therefore emerges through the subsequent search processes they induce. Existing approaches improve Meta-Skills from observed raw Skill-search trajectories and branch outcomes. However, branch performance entangles the effects of the initial discovery state and the Meta-Skill revision that generated the search process, making it difficult to characterize what a particular revision actually changed, and pushing updates toward revisions that benefit from favorable states rather than those that improve the process. We introduce HMED (Hindsight Meta-Experience Distillation), a mechanism for constructing Meta-Experience for self-improving agents. HMED revisits the completed event from which a revision originates and re-executes the incumbent and revised Meta-Skills from the same restored discovery state, so that the changes associated with the revision can be observed under a shared condition. Each comparison is distilled into a Meta-Experience, a structured record that can be reused by future updates, so that even revisions that are not ultimately retained still contribute a learning signal. Across three interactive agent benchmarks and both open-source and closed-source models, HMED consistently improves Skill discovery performance over strong baselines, shifting Meta-Skill learning beyond branch outcomes toward the consequences of changing the improvement process.
from the abstract
Pith says
Machine review · deepseek-v4-flash · 7 Oct 2026 · what was read was not recorded · not checked by a person
It can be wrong. Challenge this review or reply to it.
Holds up, with conditions. Genuinely new revision-level learning signal for Meta-Skill evolution, but Table 2's ablation cells exactly match baselines and the feedback-loop claim is unsupported until that is explained. This line is wrong
3 major objections · 3 minor
How often objections from reviews like this one are right: not yet measured
Not checked: figures, code, what the 50 references say.
The paper says, at 4.3
The review did not quote this passage.
Pith says: Objection 1 of 3 · major
The ablation rows in Table 2 exactly match three baseline rows in Table 1 to two decimals: Comparison-only (49.25) equals SkillEvolver, Outcome-only (51.49) equals MetaSkill-Evolve, and No hindsight grounding (47.76) equals Static Skill. Section 4.3 states that all ablation variants retain matched comparison, so these exact matches are not explained by the experimental design: a variant that runs matched continuations …
more of objection 1
… and an adoption rule should not coincide numerically with a baseline that does not use matched comparison, let alone with Static Skill, which has no evolving mechanism. If these cells were inherited from Table 1 or are transcription artifacts, the 4.86-point gap used to argue that consequence feedback contributes beyond selection is not a measured effect, and the central claim that 'its benefit comes not only from comparison-based selection, but also from the learning signal' (b171) is unsupported. The authors must explain the provenance of these cells, rerun the ablations with independently logged seeds, and report per-run values and variances.
The paper says, at 4.1
The review did not quote this passage.
Pith says: Objection 2 of 3 · major
All headline comparisons rest on means over three independent runs with no error bars, confidence intervals, or significance tests. Margins are small in places, notably the 0.55-point reuse margin on DeepSeek-v4-Flash in Table 4 and the 1.16-point margin on Qwen3-32B, so the claim of 'consistently improves' requires at least per-seed values or a paired test across runs.
more of objection 2
In addition, BiomniBench-DA success is scored by DeepSeek-v4-Flash, the same model family used as one of the evaluated agent backbones, and no inter-annotator or judge-agreement check is reported; since the raw judge score is itself the dependent variable, the reader cannot tell how much of the 43.19 versus 36.80 gap is attributable to judge noise.
The paper says, at 3.3
The review did not quote this passage.
Pith says: Objection 3 of 3 · major
The matched-comparison mechanism is only as valid as its state restoration and its sampling. Section 3.3 defines x_src = Restore(e_t) without saying what the restored state contains, how it is serialized, or whether the underlying LLM context is recovered exactly; Section 3.4 then computes Delta(x) from single continuations S(x,m) and S(x,m') with no repeated sampling.
more of objection 3
The appendix's own trace shows deltas as small as 0.04 (Tables 5 and 7), which is comparable to plausible run-to-run LLM variance. If restoration loses context or if a single continuation per arm is noisy, both the adoption rule mt+1 and the distilled Meta-Experience records inherit that noise, and the measured 'revision consequence' is partly noise. Please specify the restoration procedure, report whether the same restored state reproduces the same continuation when re-run, and add repeated sampling or a variance estimate for Delta at least for the ALFWorld ablation.
3 minor objections
The paper says, at 4.4
Pith says: minor objection 1 of 3
The agreement analysis would be easier to interpret if the total number of evaluated revisions and the number of ties were reported for each benchmark; the current table reports only counts for Agree and Disagree together with a percentage, and it is unclear how many revisions had zero or indeterminate Delta.
The paper says, at Appendix A
Pith says: minor objection 2 of 3
The case-study prose is appropriately hedged, but Tables 6–9 would benefit from a column stating whether each comparison was used in the adoption decision or only in the feedback record; as printed, the reader must reconstruct this from the text.
show 1 more
The paper says, at Front matter
Pith says: minor objection 3 of 3
Please fix typographical issues: 'reults' in the Table 4 caption and 'Innovation Institue' in the author affiliation block.
What would settle it
Request the per-seed run records for the three Table 2 ablation rows and verify that each row was produced by a distinct pipeline invocation with its own logs and seeds, rather than by reusing a Table 1 baseline result. If the rows are independent, recompute their means from the raw per-task outcomes and check whether any row can legitimately match a Table 1 cell to two decimals; then re-derive the Section 4.3 conclusion from the corrected values.
from the skeptic's reading
How the review went
4 readings by deepseek-v4-flash · 7 Oct 2026
- Reader holds up, with conditions
- Skeptic partly agrees
- Editor wrote the verdict above
- Referee major revision: 3 major objections, 3 minor
Not run on this paper: the simulated authors' defence.
Pith did not check this review's objections against the paper.
The review in full
The review in brief, machine-written
Short version: the mechanism is real, and the feedback-loop claim is currently overclaimed. HMED's matched re-execution of incumbent and revised Meta-Skills from a restored discovery state is a genuinely different learning signal from the branch-level outcomes used in SkillEvolver and MetaSkill-Evolve, and the paper is appropriately careful that the case-study trace is not causal (§A). The main results table shows HMED with the highest mean in all six model–benchmark configurations, and the reuse protocol (frozen Meta-Skill, reset Task-Skill library) is the right kind of transfer check.
The stress-test concern holds up, and it's load-bearing. In Table 2, three ablation cells are numerically identical to Table 1 baselines to two decimals: Comparison-only 49.25 = SkillEvolver, Outcome-only 51.49 = MetaSkill-Evolve, No hindsight grounding 47.76 = Static Skill. Section 4.3 says all variants retain matched comparison, so these exact matches are unexplained. Independent three-seed runs landing on the same two-decimal values as unrelated baselines would be a striking coincidence. If those cells were inherited or transcribed, the 4.86-point gap used to separate selection from learning is not a measured effect, and the only internal evidence for the feedback loop disappears (the appendix trace explicitly does not establish causation). The provenance of those cells has to be resolved before that claim is accepted.
Secondary soft spots: no error bars or significance tests anywhere; margins as small as 0.55 points in the reuse table; BiomniBench-DA uses judge-based scoring with no agreement check; no code, prompts, or logs released. Also, the matched comparison uses a single continuation per state, and appendix gains as small as 0.04 suggest LLM run-to-run variance could be comparable to the deltas; Restore(et) is underspecified. These are real but secondary.
This paper is for the self-improving-agent and Meta-Skill evolution crowd. The mechanism is a legitimate contribution and deserves a serious referee. My recommendation: review it, but require an explanation of Table 2, per-run numbers, and at least basic variance reporting before the feedback claim is accepted.
The strongest objection
The central claim has two components: HMED outperforms branch-level Meta-Skill learning, and part of the gain comes from returned Meta-Experience rather than from matched comparison used only for selection. The second component rests on the Section 4.3 ablation in Table 2. As printed, three Table 2 rows are numerically identical to unrelated Table 1 baselines at two decimals: Comparison-only (49.25) equals SkillEvolver; Outcome-only (51.49) equals MetaSkill-Evolve; and No hindsight grounding (47.76) equals Static Skill. Section 4.3 states that all ablation variants retain matched comparison, so these exact matches are unexplained. Independent three-seed runs matching to two decimals would be a striking coincidence, and the paper offers no explanation. If these cells were inherited from the baselines or are transcription artifacts, the 4.86-point gap used to argue for 'selection versus learning' is not a measured effect. The Appendix trace is explicitly non-causal (b265), so Table 2 is the only internal evidence for the feedback component. The provenance of those cells must be resolved before the main claim can be accepted.
The reader's conditional verdict is appropriate, and the overall assessment that the mechanism is well motivated and the ablations target the right question is fair. My stress test, however, lands on a more specific internal-consistency problem than the reader's weakest assumption about restoration fidelity and LLM noise. Section 4.3 states that all ablation variants retain matched comparison, yet Table 2's Comparison-only value (49.25) equals Table 1's SkillEvolver cell, Outcome-only (51.49) equals MetaSkill-Evolve, and No hindsight grounding (47.76) equals Static Skill. Exact equality at two decimals across independent three-seed runs is effectively impossible without shared data, and the paper gives no explanation. Because the 4.86-point difference between full HMED and Comparison-only is the only quantitative support for the 'feedback, not just selection' component of the central claim, these coincidences must be resolved before the claim is accepted. The Appendix trace is explicitly non-causal, so it cannot fill the gap. This is not an accusation of misconduct; it is a request for the provenance of those cells. If the cells are intentional baseline-equivalent checks, the interpretation changes but must be stated; if they are errors, the corrected table should be reported. A secondary concern remains: Delta in Section 3.4 is computed from single continuations, so run-to-run variance is unknown; the requested per-seed records would also address this by showing spread. With that evidence, the verdict can remain CONDITIONAL; without it, the main claim's empirical support is not established.
The referee's report
If the empirical claims held up, the paper would make a useful conceptual contribution: it identifies a genuine confound in branch-level Skill-evolution signals and proposes a concrete, falsifiable remedy in the form of matched re-executions from restored discovery states. The distinction between selection and learning from revision consequences is well motivated, and the paper is careful to state the limits of its own case study: the appendix explicitly says the trace shows that records reach the updater but does not establish that they caused the subsequent gain (b254, b265). The main weakness is that the central empirical evidence for the feedback component rests on Table 2, whose numbers appear to be identical to unrelated baselines, and the quantitative support elsewhere is reported as three-run means without error bars or significance tests. The method itself is coherent, and the paper ships detailed trace-level examples, but the evidence as printed is not yet sufficient to support the abstract's claim that HMED 'consistently improves' over strong baselines.
Pith's reading of the paper (a machine's summary, not the authors')
When an agent improves itself by revising the policy that governs its own Skill discovery (the paper's "Meta-Skill"), what should count as the unit of experience? This matters because experience is what later revisions learn from, and the paper argues that the usual unit is the wrong one: the outcome of a whole search branch mixes the starting discovery state with the revision, so a good branch is not necessarily a good improver. The paper's answer is Hindsight Meta-Experience Distillation (HMED), which revisits the completed event that motivated a revision, restores that event's discovery state, re-runs the incumbent and revised Meta-Skills from the same starting point, and distills each paired comparison into a structured Meta-Experience record that future updates read, so even a rejected revision leaves a reusable signal. On ALFWorld, ScienceWorld, and BiomniBench-DA, with Qwen3-32B and DeepSeek-v4-Flash, HMED reports the highest mean downstream Skill discovery performance in all six configurations, with ablation evidence that part of the gain comes from the returned feedback rather than from comparison-based selection alone. The authors also report that a frozen HMED Meta-Skill still guides discovery at new origins, and they note that their case-study trace shows the record reaches the updater without by itself proving it caused the later gain.
What carries the argument. The load-bearing mechanism is the matched continuation: HMED re-runs the incumbent and revised Meta-Skills from the same restored discovery state so that only the Meta-Skill varies, computing $\Delta(x) = g(S(x, m'_t)) - g(S(x, m_t))$ for each state $x$ and aggregating over a small set $X_t$ of evaluation states. Two supporting objects carry the rest of the argument: the Meta-Experience record $E_t = (e_t, \Delta m_t, \Delta_\text{process}, \Delta_\text{discovery})$, which ties the motivating event to the Meta-Skill change, the process-level trajectory differences, and the capability-discovery differences; and the adoption rule $m_{t+1} = m'_t$ iff $\Delta_t > 0$ and validity checks pass. This design makes one comparison produce two outputs — a selection signal and a reusable record — so the comparison does double duty and rejected directions still leave traces that shape later proposals.
Core claim. The paper's central claim is that a good branch is not necessarily a good improver. Because a branch outcome is $y = f(x, m)$ — a function of the discovery state $x$ and the Meta-Skill $m$ — observing many branches cannot separate the two influences, so updates drift toward revisions that happened to start from favorable states. HMED changes the experience unit from the branch to the revision: each revision is grounded in the completed discovery event $e_t$ that motivated it, its discovery state is restored as $x^\text{src}_t = \text{Restore}(e_t)$ (plus one coverage state), and the incumbent and revised Meta-Skills are re-run from that shared state as $\tau^-(x) = S(x, m_t)$ and $\tau^+(x) = S(x, m'_t)$. The revision's consequence is $\Delta_t = \frac{1}{|X_t|} \sum_{x \in X_t} \Delta(x)$, where $\Delta(x) = g(S(x, m'_t)) - g(S(x, m_t))$ and $g$ measures best-so-far utility improvement; the revision is adopted only if $\Delta_t > 0$ and validity checks pass. Every comparison, adopted or rejected, is distilled into a Meta-Experience record $E_t = (e_t, \Delta m_t, \Delta_\text{process}, \Delta_\text{discovery})$ that future updates read. The paper reports three consequences of this shift: branch outcomes and matched $\Delta$ disagree in sign in 45.5% of comparable cases on ALFWorld and 75% on ScienceWorld; removing the feedback channel costs 4.86 points on ALFWorld; and a frozen HMED Meta-Skill still guides discovery at new origins after the Task-Skill library is reset.
The premise it rests on. The whole comparison rests on the premise that a completed event can be faithfully turned back into its discovery state — the paper defines $x^\text{src}_t = \text{Restore}(e_t)$ without specifying what the restored state contains or how it is serialized — and that a single paired run per state is informative enough that the measured $\Delta_t$, often as small as 0.04, reflects the revision's consequence rather than run-to-run LLM variance.
If the paper is right (Pith's inferences, not the authors')
Skill-evolution frameworks can be rebuilt around revision-level records: the HMED pipeline — ground the revision in a completed event, restore its state, re-run both arms, distill, return — is a template that does not require replacing the underlying discovery procedure.
Rejected revisions stop being wasted computation: a failed direction is distilled into a Meta-Experience record that later updates read, so the improvement-policy space keeps accumulating experience even when nothing is adopted.
Branch-level signals risk rewarding state-luck: the reported sign disagreement between branch outcomes and matched $\Delta$ (up to 75% on ScienceWorld) implies that branch-based Meta-Skill updates can select revisions that merely started from favorable states.
A learned Meta-Skill is claimed to be a reusable improvement policy rather than a fit to its own trajectories: with the Meta-Skill frozen and the Task-Skill library reset, HMED still guides discovery at new origins, with margins of 1.16 and 0.55 points over MetaSkill-Evolve.
Returning feedback to later updates is part of the mechanism, not decoration: the ALFWorld ablation attributes 4.86 points to consequence feedback over comparison-only selection, so using matched comparisons purely for adoption leaves a learnable signal unused.
My reading: the matched comparison turns each revision into a paired trial, which invites variance-reduction upgrades the paper does not pursue — repeated sampling of both arms per state, more coverage states, and a significance threshold on $\Delta_t$ before adoption.
A testable extension the paper leaves implicit: Meta-Experience records are structured (event, change, process delta, discovery delta), so they can accumulate into a training corpus for the updater itself, letting better revision proposals be learned directly from past revisions and their measured consequences.
Because the appendix deltas are small (for example, 0.04 in the traced comparison), within-arm run-to-run variance is the key unknown; if single paired runs are too noisy, averaging repeated continuations would be the minimal fix before the records can be trusted as learning signals.
Check this review
Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.
How often reviews like this one are right
Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.
How often objections from reviews like this one are right: not yet measured
Reviews by people
This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.
The paper
Abstract
As agents continuously improve by generating and revising Skills, the process that discovers and refines those Skills becomes a learnable object in its own right. Task-Skills directly act on task execution, whereas Meta-Skills govern how agents discover and improve future Skills; their value therefore emerges through the subsequent search processes they induce. Existing approaches improve Meta-Skills from observed raw Skill-search trajectories and branch outcomes. However, branch performance entangles the effects of the initial discovery state and the Meta-Skill revision that generated the search process, making it difficult to characterize what a particular revision actually changed, and pushing updates toward revisions that benefit from favorable states rather than those that improve the process. We introduce HMED (Hindsight Meta-Experience Distillation), a mechanism for constructing Meta-Experience for self-improving agents. HMED revisits the completed event from which a revision originates and re-executes the incumbent and revised Meta-Skills from the same restored discovery state, so that the changes associated with the revision can be observed under a shared condition. Each comparison is distilled into a Meta-Experience, a structured record that can be reused by future updates, so that even revisions that are not ultimately retained still contribute a learning signal. Across three interactive agent benchmarks and both open-source and closed-source models, HMED consistently improves Skill discovery performance over strong baselines, shifting Meta-Skill learning beyond branch outcomes toward the consequences of changing the improvement process.
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.
abstract compiled · 2026-10-07
The authors' main claim, as the classifier picked it out 1.00
“Across three interactive agent benchmarks and both open-source and closed-source models, HMED consistently improves Skill discovery performance over strong baselines, shifting Meta-Skill learning beyond branch outcomes toward the consequences of changing the improvement process.”
How the main claim is stated: asserted 1.00
What the abstract itself states, as probabilities: claims first 0.08 · states limitation 0.09 · human participants 0.04 · controlled comparison 0.88 · randomized assignment 0.02 · states quantitative result 0.06 · compares with prior results 0.96 · mentions code or data availability 0.04
The abstract, sentence by sentence
This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.
| Sentence | Role | The authors' own work | States the main conclusion | Strength |
|---|---|---|---|---|
| As agents continuously improve by generating and revising Skills, the process that discovers and refines those Skills becomes a learnable object in its own right. | background 0.99 | 0.10 | 0.06 | asserted 1.00 |
| Task-Skills directly act on task execution, whereas Meta-Skills govern how agents discover and improve future Skills; their value therefore emerges through the subsequent search processes they induce. | background 1.00 | 0.09 | 0.04 | asserted 1.00 |
| Existing approaches improve Meta-Skills from observed raw Skill-search trajectories and branch outcomes. | background 1.00 | 0.03 | 0.02 | asserted 1.00 |
| However, branch performance entangles the effects of the initial discovery state and the Meta-Skill revision that generated the search process, making it difficult to characterize what a particular revision actually changed, and pushing updates toward revisions that benefit from favorable states rather than those that improve the process. | gap 0.91 | 0.11 | 0.04 | asserted 0.97 |
| We introduce HMED (Hindsight Meta-Experience Distillation), a mechanism for constructing Meta-Experience for self-improving agents. | contribution 0.98 | 0.89 | 0.07 | asserted 1.00 |
| HMED revisits the completed event from which a revision originates and re-executes the incumbent and revised Meta-Skills from the same restored discovery state, so that the changes associated with the revision can be observed under a shared condition. | method 0.99 | 0.86 | 0.05 | asserted 1.00 |
| Each comparison is distilled into a Meta-Experience, a structured record that can be reused by future updates, so that even revisions that are not ultimately retained still contribute a learning signal. | method 0.94 | 0.83 | 0.08 | asserted 0.99 |
| Across three interactive agent benchmarks and both open-source and closed-source models, HMED consistently improves Skill discovery performance over strong baselines, shifting Meta-Skill learning beyond branch outcomes toward the consequences of changing the improvement process. | result 0.99 | 0.95 | 0.90 | asserted 1.00 |
What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.95
What was examined, and what was not
Sources read
- arXiv listing abstract v1
Not examined
- full text
- figures and tables
- references and citing sentences
- relations to other papers
Levels run
- P abstract
Not run
- S sections
- T body sentences
- C citations
- R cross-paper relations
How the labels were made
Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-10-07: 1 request, 6,428 input tokens.
Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON
Figures (2)
References (50)
- [1] MedExpMem: Adapting Experience Memory for Differential Diagnosis , author=. 2026 , doi= on Pith
- [2] Wang, Zefeng and Yan, Minxi and Bi, Jinhe and Yan, Sikuan and Tresp, Volker and Ma, Yunpu , year =. 2607.05297 , archivePrefix = on PithPith's reviewarXiv
- [3] Wu, Shujin and Qian, Cheng and Chen, Xiusi and Ji, Heng , year =. Teaching. 2607.21971 , archivePrefix = on PithPith's reviewarXiv
- [4] What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents , author =. 2026 , eprint = on Pith
- [5] Yeo, Woongyeong and Choi, Yumin and Ki, Taekyung and Hwang, Sung Ju , year =. 2605.17873 , archivePrefix = on PithPith's reviewarXiv
- [6] Shi, Qiming and Dou, Yibo and Zhu, Jiawen and Tao, Yulong and Jin, Linbo and Kang, Zhaolu and Zhou, Yunfan and Weng, Di , year =. 2607.28048 , archivePrefix = on PithPith's reviewarXiv
- [7] Lin, Hehai and Yang, Qi and Qin, Chengwei , year =. 2606.18837 , archivePrefix = on PithPith's reviewarXiv
- [8] A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence , author =. 2025 , eprint = on Pith
- [9] Fang, Jinyuan and Peng, Yanwen and Zhang, Xi and Wang, Yingxu and Yi, Xinhao and Zhang, Guibin and Xu, Yi and Wu, Bin and Liu, Siwei and Li, Zihao and Ren, Zhaochun and Aletras, Nikos and Wang, Xi and Zhou, Han and Meng, Zaiqiao , year =. A Comprehensive Survey of Self-Evolvin... on PithPith's reviewarXiv
- [10] Alzubi, Salaheddin and Provenzano, Noah and Bingham, Jaydon and Chen, Weiyuan and Vu, Tu , year =. 2603.02766 , archivePrefix = on PithPith's reviewarXiv
- [11] Zhang, Ziao and Shi, Kou and Huang, Shiting and Nie, Avery and Zeng, Yu and Zhao, Yiming and Fang, Zhen and Su, Qishen and Qiu, Haibo and Yang, Wei and Ren, Qingnan and Zou, Shun and Huang, Wenxuan and Chen, Lin and Chen, Zehui and Zhao, Feng , year =. 2604.17308 , archivePrefix = on PithPith's reviewarXiv
- [12] Zhou, Xiaolin and Liu, Jinbo and Li, Li and Rossi, Ryan A. and Hu, Xiyang , year =. Counterfactual Trace Auditing of. 2605.11946 , archivePrefix = on PithPith's reviewarXiv
- [13] Wu, Rong and Wang, Xiaoman and Mei, Jianbiao and Cai, Pinlong and Fu, Daocheng and Yang, Cheng and Wen, Licheng and Yang, Xuemeng and Shen, Yufan and Wang, Yuxin and Shi, Botian , year =. 2510.16079 , archivePrefix = on PithPith's reviewarXiv
- [14] Reflexion: Language Agents with Verbal Reinforcement Learning , author =. Advances in Neural Information Processing Systems , year = on Pith
- [15] Zhao, Andrew and Huang, Daniel and Xu, Quentin and Lin, Matthieu and Liu, Yong-Jin and Huang, Gao , year =. 2308.10144 , archivePrefix = on PithPith's reviewarXiv
- [16] Voyager: An Open-Ended Embodied Agent with Large Language Models , author =. 2023 , eprint = on Pith
- [17] Promptbreeder: Self-Referential Self-Improvement via Prompt Evolution , author =. 2023 , eprint = on Pith
- [18] Agrawal, Lakshya A. and Tan, Shangyin and Soylu, Dilara and Ziems, Noah and Khare, Rishi and Opsahl-Ong, Krista and Singhvi, Arnav and Shandilya, Herumb and Ryan, Michael J. and Jiang, Meng and Potts, Christopher and Sen, Koushik and Dimakis, Alexandros G. and Stoica, Ion and ... on PithPith's reviewarXiv
- [19] Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models , author =. 2025 , eprint = on Pith
- [20] ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory , author =. 2025 , eprint = on Pith
show 30 more
- [21] Automated Design of Agentic Systems , author =. International Conference on Learning Representations , year = on Pith
- [22] Zhang, Jenny and Hu, Shengran and Lu, Cong and Lange, Robert and Clune, Jeff , year =. Darwin G. 2505.22954 , archivePrefix = on PithPith's reviewarXiv
- [23] Feng, Yuru and Chen, Yaoqi and Zhao, Beidi and Zhang, Qianxi and Wang, Xinjiang and Lu, Jianan and Wang, Zhirui and Xu, Shusen and Li, Zengzhong and Chen, Qi , year =. 2608.05628 , archivePrefix = on PithPith's reviewarXiv
- [24] SkillCommit: Evolving Agent Skills through Behaviorally Validated Scope Expansion , author =. 2026 , eprint = on Pith
- [25] Tian, Yu and Chen, Jiawei and Zheng, Lifan and Tao, Mingxiang and Zeng, Xinyi and Yin, Zhaoxia and Su, Hang and Sun, Xian , year =. Skills-Coach: A Self-Evolving Skill Optimizer via Training-Free. 2604.27488 , archivePrefix = on PithPith's reviewarXiv
- [26] Guan, Tianyi and Wang, Yiding and Yang, Haotong and Cao, Siyuan and Liu, Shirui and Hu, Yi and Li, Jiaqi and Zhang, Muhan , year =. ContinualSkillBench: Can. 2608.03874 , archivePrefix = on PithPith's reviewarXiv
- [27] Taming the Monster: A Fast and Simple Algorithm for Contextual Bandits , author =. Proceedings of the 31st International Conference on Machine Learning , pages =. 2014 , volume = on Pith
- [28] Instrument-Armed Bandits , author =. Proceedings of Algorithmic Learning Theory , pages =. 2018 , volume = on Pith
- [29] High Confidence Policy Improvement , author =. Proceedings of the 32nd International Conference on Machine Learning , pages =. 2015 , volume = on Pith
- [30] Pham, Thinh and Nguyen, Nguyen and Zunjare, Pratibha and Chen, Weiyuan and Tseng, Yu-Min and Vu, Tu , year =. 2506.01062 , archivePrefix = on PithPith's reviewarXiv
- [31] Opsahl-Ong, Krista and Singhvi, Arnav and Collins, Jasmine and Zhou, Ivan and Wang, Cindy and Baheti, Ashutosh and Oertell, Owen and Portes, Jacob and Havens, Sam and Elsen, Erich and Bendersky, Michael and Zaharia, Matei and Chen, Xing , year =. 2603.08655 , archivePrefix = on PitharXiv
- [32] Shridhar, Mohit and Yuan, Xingdi and C. International Conference on Learning Representations , year = on Pith
- [33] Learning to Learn by Gradient Descent by Gradient Descent , author =. Advances in Neural Information Processing Systems , year =. 1606.04474 , archivePrefix = on PithPith's reviewarXiv
- [34] Hindsight Experience Replay , author =. Advances in Neural Information Processing Systems , year =. 1707.01495 , archivePrefix = on PithPith's reviewarXiv
- [35] Monte Carlo Theory, Methods and Examples , author =. 2013 , publisher = on Pith
- [36] Optimization as a Model for Few-Shot Learning , author =. International Conference on Learning Representations , year = on Pith
- [37] Discovering Reinforcement Learning Algorithms , author =. Advances in Neural Information Processing Systems , year =. 2007.08794 , archivePrefix = on PithPith's reviewarXiv
- [38] Evolving Reinforcement Learning Algorithms , author =. International Conference on Learning Representations , year =. 2301.09846 , archivePrefix = on PithPith's reviewarXiv
- [39] Dream to Control: Learning Behaviors by Latent Imagination , author =. International Conference on Learning Representations , year =. 1912.01603 , archivePrefix = on PithPith's reviewarXiv
- [40] Deep Reinforcement Learning at the Edge of the Statistical Precipice , author =. Advances in Neural Information Processing Systems , year =. 2108.13264 , archivePrefix = on PithPith's reviewarXiv
- [41] Connecting Large Language Models with Evolutionary Algorithms Yields Powerful Prompt Optimizers , author =. International Conference on Learning Representations , year =. 2310.06201 , archivePrefix = on PithPith's reviewarXiv
- [42] Feng, Qianhan and Huang, Zhongzhen and Zhu, Yakun and Gu, Yannian and Chu, Chiu Wing Winnie and Zhang, Xiaofan and Dou, Qi , year =. 2605.22872 , archivePrefix = on PithPith's reviewarXiv
- [43] Kim, Zae Myung and Lee, Young-Jun and Jwa, Seungyeon and Kang, Dongyeop , year =. 2608.24735 , archivePrefix = on PithPith's reviewarXiv
- [44] Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses , author =. 2026 , eprint = on Pith
- [45] Li, Yang and Yavuz, Semih and Joty, Shafiq , year =. 2609.05295 , archivePrefix = on PithPith's reviewarXiv
- [46] Hyperagents , author =. 2026 , eprint = on Pith
- [47] Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks , author =. International Conference on Machine Learning , year =. 1703.03400 , archivePrefix = on PithPith's reviewarXiv
- [48] Zhang, Genrui and Zhu, Erle and Zhou, Jinfeng and Jia, Caiyan and Wang, Hongning , year =. 2605.10500 , archivePrefix = on PithPith's reviewarXiv
- [49] Wang, Ruoyao and Jansen, Peter and C. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , year =. 2203.07540 , archivePrefix = on PithPith's reviewarXiv
- [50] Qu, Yuanhao and Lu, Yingzhou and Tu, Xinming and Zhang, Serena and She, Tianwei and Shaw, Alexander Glenn and Shih, Jou-Ho and Zhao, Bingqing and Shen, Minjie and Yang, Haochen and Yan, Jielin and Zhang, Rongchuan and Wu, Xinze and Li, Tingting and Zhou, Bin and Wang, Ning and... on Pith
Challenges and discussion
Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.
Come back by email
A short list of the week's newly reviewed papers. Nothing is sent until you confirm from your inbox, and every email has a link that stops it. No account.


No comments yet.
Sign in to comment