Pith. sign in

REVIEW 3 major objections 5 minor 3 cited by

Tiny Data Injection Fixes Most LLM Grammar Failures

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · glm-5.2

2026-07-05 14:57 UTC pith:MEMTMOH2

load-bearing objection Clean data-intervention experiment on GPT-2/BLiMP; main concern is whether improvements reflect genuine rule acquisition or surface-pattern matching to the benchmark format. the 3 major comments →

arxiv 2604.17931 v4 pith:MEMTMOH2 submitted 2026-04-20 cs.AI

LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent

Wanli Li , Bince Qu , Bo Pan , Jianyu Zhang , Zheng Liu , Pan Zhang , Wei Chen , Bo Zhang This is my paper
classification cs.AI
keywords trainingagenticsearchdeepreal-worldresearchscalableagent
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper asks why language models master some grammatical rules near-perfectly while failing badly on others, even after training on trillions of tokens. The authors test whether these failures come from inherent architectural limitations or simply from insufficient exposure to specific grammatical constructions in the training data. They pre-train a GPT-2 Small model (124M parameters) on 100M tokens sampled from the FineWeb corpus and then inject as little as 1% targeted synthetic data covering nine of the worst-performing BLiMP paradigms — a benchmark that tests whether a model prefers grammatical sentences over minimally different ungrammatical ones. The intervention substantially improves performance on 8 of the 9 targeted paradigms, with one case (only_npi_scope) jumping from 20.9% to 69.4% accuracy. In one striking result, just 0.01% synthetic data (roughly 10K tokens) lifted principle_A_reconstruction from 37.2% to 72.3%. The 124M-parameter model with this tiny intervention even outperformed Llama-3 70B (trained on 15+ trillion tokens) on specific paradigms. One paradigm, principle_A_c_command, resisted improvement entirely. The authors conclude that the heterogeneity in formal linguistic competence is largely a data composition problem, not an architecture problem — standard transformer architectures have the latent capacity to acquire formal syntax if the training data contains sufficient distributional evidence for the relevant constructions.

Core claim

The central discovery is that injecting as little as 1% targeted synthetic data into pre-training can rescue a small language model's performance on grammatical paradigms where it previously performed below chance, while preserving overall benchmark performance. This holds across 8 of 9 tested BLiMP paradigms, and in one case a mere 0.01% injection was sufficient for a large jump. The one persistent failure (principle_A_c_command) suggests that some constructions may require either much higher data density or genuinely different inductive biases. The finding that a 124M model with targeted data can beat a 70B model on specific paradigms reframes the bottleneck from model scale to data compos

What carries the argument

The central mechanism is targeted synthetic data injection: the authors generate synthetic text documents designed to contain specific grammatical constructions (drawn from nine BLiMP paradigms) across diverse genres and subgenres, then mix a small fraction (0.01%–1%) of this synthetic data into a 100M-token pre-training corpus. The BLiMP benchmark itself is the evaluation instrument — it tests formal linguistic competence by presenting minimal pairs (a grammatical sentence and an ungrammatical variant differing in one feature) and checking whether the model assigns higher probability to the grammatical version.

Load-bearing premise

The paper assumes that improved BLiMP accuracy after synthetic data injection reflects genuine acquisition of the underlying grammatical rule, rather than the model learning to recognize the surface patterns of the synthetic data or the minimal-pair format used by the benchmark itself.

What would settle it

If a model trained on the synthetic data passes BLiMP minimal pairs but fails on genuinely novel sentences testing the same grammatical rule in different surface forms, the improvement would be memorization rather than competence acquisition.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Data composition, not raw scale, may be the primary lever for closing specific competence gaps in language models — which would redirect effort from scaling laws toward curated data design.
  • If the result generalizes beyond GPT-2 Small, billion-parameter models could potentially fix their own grammatical blind spots with very small, targeted data additions rather than retraining from scratch.
  • The persistent failure of principle_A_c_command suggests a boundary case where either the construction is genuinely hard for transformers to learn from surface statistics, or the synthetic data generation method does not adequately capture the relevant distributional signal.
  • The finding that targeting one paradigm sometimes improves or degrades other non-targeted paradigms (Appendix D) points to transfer effects — both positive and negative — that could inform a more principled approach to curriculum design in pre-training.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This paper investigates whether the heterogeneous performance of language models on BLiMP paradigms stems from data scarcity rather than architectural limitations. The authors pre-train GPT-2 Small (124M) on 100M tokens from FineWeb and inject small fractions (0.01%–1%) of synthetic data targeting the 9 worst-performing BLiMP paradigms. They find that 8 of 9 paradigms improve substantially (e.g., only_npi_scope: 20.9%→69.4%), while principle_A_c_command remains below chance. The paper also reports cross-paradigm transfer effects and an ablation over intervention magnitudes. Code and data are open-sourced.

Significance. The question of whether formal linguistic competence failures are data-driven or architecture-driven is important for understanding language model capabilities and for data-centric AI research. The experimental design—controlled pre-training with targeted synthetic data injection—is a clean intervention. The authors deserve credit for open-sourcing code and data, for including an ablation over intervention magnitudes (Table 3), and for reporting the full 67-paradigm breakdown (Table 5) including cross-paradigm transfer effects. The finding that ~30 synthetic examples can boost principle_A_reconstruction from 37.2% to 72.3% is striking and warrants careful scrutiny.

major comments (3)
  1. §4–5, Table 3: The central claim—that improvements reflect acquisition of linguistic rules rather than surface-pattern recognition—depends on the synthetic training data not sharing the structural format of BLiMP test items. BLiMP evaluates via minimal pairs (grammatical vs. minimally ungrammatical sentences). If the synthetic data for, e.g., only_npi_scope contains sentences with the same NPI-scope structures and vocabulary distribution as the BLiMP test items, the model may be learning to recognize the test format rather than the underlying grammatical rule. The principle_A_reconstruction result (37.2%→72.3% with ~30 examples at 0.01% injection) is especially consistent with this concern: such a dramatic jump from so few examples is more typical of cue-latching than robust rule acquisition. The paper does not include any evaluation on held-out probes with different surface forms, nor a
  2. §5.3, Table 3: Several paradigms show non-monotonic responses to intervention magnitude (only_npi_scope: 20.9→41.9→31.0→69.4; existential_there_quantifiers_2: 23.4→13.2→25.2→51.8). The paper notes these anomalies but does not analyze them. Non-monotonicity is important because it is more consistent with unstable, cue-driven learning than with systematic generalization. If the model were acquiring a robust grammatical rule, one would generally expect monotonic improvement with more evidence. The authors should discuss what these non-monotonic patterns imply for their central claim, or at minimum acknowledge them as a limitation on the strength of the 'data scarcity' conclusion.
  3. §6: The comparison between GPT-2 Small (124M, 100M tokens + intervention) and Llama-3 70B (15T tokens, no intervention) on specific paradigms is presented as evidence that 'data composition may matter more than data scale.' This comparison is not controlled: Llama-3 was not trained with the same intervention, so the comparison conflates data composition with model size, training data, and training procedure. The claim would be strengthened by applying the same intervention to a larger model, or by framing the comparison more carefully as an existence proof rather than evidence that data composition dominates scale.
minor comments (5)
  1. The arXiv abstract and title ('LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent') do not match the paper content ('Heterogeneity in Formal Linguistic Competence of Language Models'). This appears to be a submission error but should be corrected.
  2. Appendix C describes genre/subgenre taxonomy for synthetic data diversity but does not specify how the linguistic target structures are embedded in the synthetic text. A brief example of a synthetic document for one paradigm would help readers assess the contamination risk.
  3. Figures 1–2 (Appendix B) are difficult to read; the axis labels and legend text are very small. Consider increasing font sizes or splitting into more figures.
  4. Table 5 is large and dense; consider highlighting the targeted paradigms or separating targeted vs. non-targeted results for clarity.
  5. The paper does not report variance or confidence intervals for the BLiMP accuracy numbers in Tables 3 and 5. Since BLiMP paradigms contain 1000 items each, binomial confidence intervals would be informative, especially for near-chance scores.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful and constructive report. All three major comments are well-taken. Comment 1 (surface-form overlap between synthetic data and BLiMP test items) identifies a genuine gap in our evaluation: we did not include held-out probes with different surface forms, and we will add such an evaluation. Comment 2 (non-monotonic responses to intervention magnitude) is correct that we noted but did not analyze these patterns; we will add a substantive discussion acknowledging them as a limitation on the strength of the data-scarcity conclusion. Comment 3 (uncontrolled GPT-2 vs. Llama-3 comparison) is also correct; we will reframe the comparison as an existence proof rather than evidence that data composition dominates scale. All three comments lead to revisions.

read point-by-point responses

  1. Referee: §4–5, Table 3: The central claim depends on synthetic training data not sharing the structural format of BLiMP test items. If synthetic data for e.g. only_npi_scope contains sentences with the same NPI-scope structures and vocabulary distribution as BLiMP test items, the model may be learning to recognize the test format rather than the underlying grammatical rule. The principle_A_reconstruction result (37.2%→72.3% with ~30 examples at 0.01% injection) is especially consistent with cue-latching. The paper does not include evaluation on held-out probes with different surface forms.

    Authors: The referee raises a legitimate and important concern. We acknowledge that our current evaluation does not rule out the possibility that some portion of the improvement reflects surface-form matching between the synthetic training data and the BLiMP minimal-pair test format, rather than robust acquisition of the underlying grammatical rule. The principle_A_reconstruction result—37.2% to 72.3% from approximately 30 examples—is indeed the case most susceptible to this alternative explanation, and we agree it warrants careful scrutiny. To address this, we will add a new evaluation using held-out probes with different surface forms and vocabulary distributions from the BLiMP test items. Specifically, we will generate novel minimal pairs for the targeted paradigms using different lexical items and sentence frames than those in BLiMP, and evaluate whether the improvements transfer. We will also add an explicit discussion of the cue-latching concern, particularly for the principle_A_reconstruction paradigm, and acknowledge that without such held-out probes, our claim that improvements reflect rule acquisition rather than surface-pattern recognition is not fully established. We note that the cross-paradigm transfer effects reported in Table 5 (where targeting one paradigm improves related but non-targeted paradigms) provide partial evidence against pure cue-latching, since the transfer paradigms use different surface forms than the targeted synthetic data—but we agree this is not a substitute for a dedicated held-out evaluation. revision: yes

  2. Referee: §5.3, Table 3: Several paradigms show non-monotonic responses to intervention magnitude (only_npi_scope: 20.9→41.9→31.0→69.4; existential_there_quantifiers_2: 23.4→13.2→25.2→51.8). The paper notes these anomalies but does not analyze them. Non-monotonicity is more consistent with unstable, cue-driven learning than with systematic generalization. The authors should discuss what these non-monotonic patterns imply for their central claim, or at minimum acknowledge them as a limitation.

    Authors: We agree that the non-monotonic response patterns deserve more than the brief mention they currently receive. The referee is correct that non-monotonicity is more consistent with unstable, cue-driven learning than with systematic generalization, and that this has implications for the strength of our data-scarcity conclusion. We will revise Section 5.3 to include a substantive discussion of these patterns. Specifically, we will acknowledge that the non-monotonic responses in only_npi_scope and existential_there_quantifiers_2 complicate the interpretation that improvements reflect robust rule acquisition, and that they are at least partially consistent with the cue-latching concern raised in the referee's first comment. We will also note that the paradigms showing monotonic improvement (e.g., principle_A_reconstruction, coordinate_structure_constraint_complex_left_branch, left_branch_island_echo_question) provide stronger evidence for the data-scarcity hypothesis than those showing non-monotonic patterns. We will add an explicit statement in the Limitations section that the non-monotonic responses constrain the strength of the conclusion that data scarcity is the sole bottleneck for all paradigms, and that the interaction between targeted synthetic data and existing corpus statistics may produce paradigm-specific dynamics that our framework does not fully explain. revision: yes

  3. Referee: §6: The comparison between GPT-2 Small (124M, 100M tokens + intervention) and Llama-3 70B (15T tokens, no intervention) is not controlled: Llama-3 was not trained with the same intervention, so the comparison conflates data composition with model size, training data, and training procedure. The claim would be strengthened by applying the same intervention to a larger model, or by framing the comparison more carefully as an existence proof rather than evidence that data composition dominates scale.

    Authors: The referee is correct that the comparison between GPT-2 Small with intervention and Llama-3 70B without intervention is not controlled, and that presenting it as evidence that 'data composition may matter more than data scale' conflates multiple variables. We cannot run the same intervention on Llama-3 70B, as we do not have access to its pre-training pipeline, so this concern cannot be fully resolved within the scope of the current work. However, we will revise Section 6 to reframe the comparison explicitly as an existence proof—demonstrating that a small model with targeted data intervention can achieve competence on specific paradigms that a much larger model without such intervention does not—rather than as controlled evidence that data composition dominates scale. We will also add a clear statement that this comparison conflates model size, training data, training procedure, and data composition, and that disentangling these factors would require applying the same intervention strategy to larger models, which we identify as a direction for future work. This is also already partially acknowledged in our Limitations section, but we will strengthen the language there to match the more careful framing in Section 6. revision: yes

Circularity Check

0 steps flagged

No significant circularity; the paper is an empirical intervention study with external benchmarks and standard architecture.

full rationale

The paper's derivation chain is straightforward and self-contained: (1) pre-train GPT-2 Small on FineWeb, (2) inject synthetic data targeting specific linguistic phenomena, (3) evaluate on BLiMP. The central claim — that data scarcity rather than architecture is the bottleneck — is supported by the empirical observation that targeted data injection improves BLiMP scores. There is no fitted parameter renamed as a prediction, no self-definitional loop, no load-bearing self-citation chain, and no uniqueness theorem invoked. The synthetic data is the experimental intervention itself, not a circular dependency: the paper is explicitly testing whether adding distributional evidence for a phenomenon improves performance on that phenomenon, which is a standard causal design. The skeptic's concern about train-test format contamination (synthetic data sharing surface structure with BLiMP test items) is a validity risk about whether BLiMP improvements reflect genuine competence versus surface-pattern recognition — that is a correctness concern, not a circularity concern. The paper relies on external benchmarks (BLiMP, FineWeb), standard architectures (GPT-2), and external prior work for framing. Score of 2 reflects the minor observation that the synthetic data generation methodology is entirely author-designed and could share structural properties with the evaluation set, but this does not constitute circularity in the derivation chain.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The paper introduces no new entities, particles, or forces. It uses standard architectures (GPT-2), standard benchmarks (BLiMP), and standard corpora (FineWeb). The free parameters are experimental design choices (intervention magnitude, token budget, paradigm selection) rather than fitted model constants.

free parameters (3)
  • Intervention magnitude = 1%
    The fraction of synthetic data injected (0.01%, 0.1%, 1%) is chosen by the authors. 1% is the main reported intervention.
  • Training token budget = 100M
    The 100M-token budget is chosen to approximate human-scale input, following the BabyLM challenge framing.
  • Targeted paradigms = 9
    The 9 worst-performing BLiMP paradigms are selected for intervention, which is a post-hoc selection based on baseline performance.
axioms (3)
  • domain assumption BLiMP accurately measures formal linguistic competence
    The entire evaluation framework relies on BLiMP being a valid measure of grammatical knowledge. This is a standard assumption in the field but is an axiom the paper depends on (Section 3).
  • domain assumption GPT-2 Small architecture is representative enough to draw conclusions about architectural limitations
    The paper acknowledges this limitation but the central claim about 'data scarcity rather than inherent architectural limitations' rests on this assumption (Section 6, Limitations).
  • domain assumption Synthetic data targeting a paradigm provides genuine distributional evidence for the underlying grammatical rule
    The paper assumes that the synthetic data injection teaches the grammatical rule rather than just the surface pattern of the BLiMP minimal pairs (Section 4).

pith-pipeline@v1.1.0-glm · 11835 in / 2541 out tokens · 373858 ms · 2026-07-05T14:57:05.616603+00:00 · methodology

0 comments
read the original abstract

Reinforcement Learning (RL) has emerged as a powerful training paradigm for LLM-based agents. However, scaling agentic RL for deep research remains constrained by two coupled challenges: hand-crafted synthetic data fails to elicit genuine real-world search capabilities, and real-world search dependency during RL training introduces instability and prohibitive cost, which limits the scalability of Agentic RL. LiteResearcher is a training framework that makes Agentic RL scalable: by constructing a lite virtual world that mirrors real-world search dynamics, we enable a continuously improving training recipe that empowers a tiny search agent to outperform large-scale open-source and commercial models (e.g., Tongyi DeepResearch and Claude-4.5 Sonnet). Specifically, on common benchmarks such as GAIA and Xbench, our LiteResearcher-4B achieves open-source state-of-the-art results of 71.3% and 78.0% respectively, demonstrating that scalable RL training is a key enabler for Deep Research Agents.

Figures

Figures reproduced from arXiv: 2604.17931 by Bince Qu, Bo Pan, Bo Zhang, Jianyu Zhang, Pan Zhang, Wanli Li, Wei Chen, Zheng Liu.

Figure 1
Figure 1. Figure 1: Performance of LiteResearcher. Left: Accuracy comparison on the Xbench DeepSearch benchmark across models of various scales. Right: Average rollout latency and cost per turn. *Equal contribution. Work done during internship at Simplex AI. †Corresponding authors. 1 arXiv:2604.17931v2 [cs.AI] 22 Apr 2026 [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: System architecture overview. (a) Corpus Extension and QA Synthesis: An iterative data engine which also enriches local webpage corpus, powering stable, local tools for zero-cost agent RL training. (b) Reinforcement Curriculum Learning: Synthetic tasks are leveled by complexity to guide the agent through progressive training stages. This reinforcement learning loop utilizes local tool interactions, scaling… view at source ↗
Figure 3
Figure 3. Figure 3: On-Policy vs. Off-Policy training reward. On-policy training is more stable and continues to improve throughout training. algorithm, where each rollout batch is split into multiple mini-batches (e.g., 256 samples into 4 mini-batches) and used for several successive updates. As shown in [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Stage 1 vs. Stage 2. GAIA accuracy (EMA smoothed) during RL training. The two-stage curriculum overcomes the Stage 1 plateau. 9 [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Corpus domain category distribution. The enriched corpus spans 18 domain categories covering 1M+ unique domains, with Academic, Regional, and Encyclopedia sources forming the largest segments. This broad coverage ensures that the local search environment reflects diverse real-world web structure. B SFT Details B.1 Data Composition The SFT dataset consists of 68,231 high-quality search trajectories from thr… view at source ↗
Figure 6
Figure 6. Figure 6: shows the distribution of the final 68K trajectories after processing: the mean token length is 12.4K with a long tail extending to ∼45K, and the mean number of interaction turns is 8.7, concentrated around 5–8 turns. The long tail motivates our choice of 64K max sequence length to cover 100% of samples. 0 10k 20k 30k 40k Token Length (per sample) 0 1000 2000 3000 4000 5000 6000 Number of Samples N = 68,23… view at source ↗
Figure 7
Figure 7. Figure 7: RL suppresses repetitive actions inherited from SFT. (a) Mean reward increases from ∼0.42 to ∼0.70, confirming improved task accuracy. (b–d) Mean response length (∼18K→12K tokens), interaction turns (∼30→24), and length clip ratio (∼0.28→0.02) all decrease, reflecting elimination of redundant action loops. No explicit length or repetition penalty is used. C.5 Training Dynamics We track several metrics acro… view at source ↗
Figure 8
Figure 8. Figure 8: Training dynamics during RL. (a) GAIA validation accuracy. (b) Policy entropy (Stage 1: temp = 0.7; Stage 2: temp = 1.0). (c) Average tool calls per sample. (d) Average trajectory total tokens. Dashed vertical lines mark the Stage 1→2 transition at step 220. D Infrastructure Details [PITH_FULL_IMAGE:figures/full_fig_p023_8.png] view at source ↗

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. SearchEyes: Towards Frontier Multimodal Deep Search Intelligence via Search World Simulation

    cs.AI 2026-07 unverdicted novelty 6.0

    SearchEyes unifies multimodal search-agent training via Perception-Knowledge Chains on Wikidata5M and Hop-Anchored Policy Optimization, claiming a 6.2-point average gain over the strongest open-source baseline on six ...

  2. AgenticRL: Self-Refining Agentic Reinforcement Learning for Vision-Conditioned UAV Navigation

    cs.RO 2026-06 unverdicted novelty 6.0

    AgenticRL deploys a multimodal GPT agent in a closed-loop process to autonomously design and refine reward functions for PPO-trained vision-conditioned UAV navigation policies, reporting 71% policy improvement and 91%...

  3. MetaResearcher: Scaling Deep Research via Self-Reflective Reinforcement Learning in Adversarial Virtual Environments

    cs.AI 2026-06 unverdicted novelty 5.0

    MetaResearcher is a proposed multi-component framework for scaling deep research agent training via adversarial virtual worlds, discovery tasks, meta-rewards, and multi-agent collaboration.

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages · cited by 3 Pith papers · 27 internal anchors

  1. [1]

    Introducing claude sonnet 4.5

    Anthropic. Introducing claude sonnet 4.5. https://www.anthropic.com/news/claude-sonnet-4-5, 2025

    work page 2025

  2. [2]

    Agentcpm-explore: Realizing long-horizon deep exploration for edge-scale agents

    Haotian Chen, Xin Cong, Shengda Fan, Yuyang Fu, Ziqin Gong, Yaxi Lu, Yishan Li, Boye Niu, Chengjun Pan, Zijun Song, et al. Agentcpm-explore: Realizing long-horizon deep exploration for edge-scale agents. arXiv preprint arXiv:2602.06485, 2026

    work page arXiv 2026

  3. [3]

    M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation

    Jianlv Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. Bge m3-embedding: Multi-lingual, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. arXiv preprint arXiv:2402.03216, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  4. [4]

    xbench: Tracking Agents Productivity Scaling with Profession-Aligned Real-World Evaluations

    Kaiyuan Chen, Yixin Ren, Yang Liu, Xiaobo Hu, Haotong Tian, Tianbao Xie, Fangfu Liu, Haoye Zhang, Hongzhang Liu, Yuan Gong, et al. xbench: Tracking agents productivity scaling with profession-aligned real-world evaluations. arXiv preprint arXiv:2506.13651, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  5. [5]

    Process Reinforcement through Implicit Rewards

    Ganqu Cui, Lifan Yuan, Zefan Wang, Hanbin Wang, Yuchen Zhang, Jiacheng Chen, Wendi Li, Bingxiang He, Yuchen Fan, Tianyu Yu, et al. Process reinforcement through implicit rewards. arXiv preprint arXiv:2502.01456, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  6. [6]

    MegaScience: Pushing the Frontiers of Post-Training Datasets for Science Reasoning

    Run-Ze Fan, Zengzhi Wang, and Pengfei Liu. Megascience: Pushing the frontiers of post-training datasets for science reasoning. arXiv preprint arXiv:2507.16812, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  7. [7]

    arXiv preprint arXiv:2508.07976

    Jiaxuan Gao, Wei Fu, Minyang Xie, Shusheng Xu, Chuyi He, Zhiyu Mei, Banghua Zhu, and Yi Wu. Beyond ten turns: Unlocking long-horizon agentic search with large-scale asynchronous rl. arXiv preprint arXiv:2508.07976, 2025

    work page arXiv 2025

  8. [8]

    DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Peiyi Wang, Qihao Zhu, Runxin Xu, Ruoyu Zhang, Shirong Ma, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  9. [9]

    Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model

    Jingcheng Hu, Yinmin Zhang, Qi Han, Daxin Jiang, Xiangyu Zhang, and Heung-Yeung Shum. Open-reasoner-zero: An open source approach to scaling up reinforcement learning on the base model. arXiv preprint arXiv:2503.24290, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  10. [10]

    R2E-Gym: Procedural Environments and Hybrid Verifiers for Scaling Open-Weights SWE Agents

    Naman Jain, Jaskirat Singh, Manish Shetty, Liang Zheng, Koushik Sen, and Ion Stoica. R2e-gym: Procedural environments and hybrid verifiers for scaling open-weights swe agents. arXiv preprint arXiv:2504.07164, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  11. [11]

    Diskann: Fast accurate billion-point nearest neighbor search on a single node

    Suhas Jayaram Subramanya, Fnu Devvrit, Harsha Vardhan Simhadri, Ravishankar Krishnawamy, and Rohan Kadekodi. Diskann: Fast accurate billion-point nearest neighbor search on a single node. Advances in neural information processing Systems, 32, 2019

    work page 2019

  12. [12]

    Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning

    Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. Search-r1: Training llms to reason and leverage search engines with reinforcement learning. arXiv preprint arXiv:2503.09516, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  13. [13]

    Fact, fetch, and reason: A unified evaluation of retrieval-augmented generation

    Satyapriya Krishna, Kalpesh Krishna, Anhad Mohananey, Steven Schwarcz, Adam Stambler, Shyam Upadhyay, and Manaal Faruqui. Fact, fetch, and reason: A unified evaluation of retrieval-augmented generation. In NAACL, 2025

    work page 2025

  14. [14]

    Websailor-v2: Bridging the chasm to proprietary agents via synthetic data and scalable reinforcement learning.arXiv preprint arXiv:2509.13305, 2025a

    Kuan Li, Zhongwang Zhang, Huifeng Yin, Rui Ye, Yida Zhao, Liwen Zhang, Litu Ou, Dingchu Zhang, Xixi Wu, Jialong Wu, et al. Websailor-v2: Bridging the chasm to proprietary agents via synthetic data and scalable reinforcement learning. arXiv preprint arXiv:2509.13305, 2025 a

    work page arXiv 2025

  15. [15]

    Chain-of-Agents: End-to-End Agent Foundation Models via Multi-Agent Distillation and Agentic RL

    Weizhen Li, Jianbo Lin, Zhuosong Jiang, Jingyi Cao, Xinpeng Liu, Jiayu Zhang, Zhenqiang Huang, Qianben Chen, Weichen Sun, Qiexiang Wang, et al. Chain-of-agents: End-to-end agent foundation models via multi-agent distillation and agentic rl. arXiv preprint arXiv:2508.13167, 2025 b

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  16. [16]

    Search-o1: Agentic search-enhanced large reasoning models

    Xiaoxi Li, Guanting Dong, Jiajie Jin, Yuyao Zhang, Yujia Zhou, Yutao Zhu, Peitian Zhang, and Zhicheng Dou. Search-o1: Agentic search-enhanced large reasoning models. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pp.\ 5420--5438, 2025 c

    work page 2025

  17. [17]

    WebThinker: Empowering Large Reasoning Models with Deep Research Capability

    Xiaoxi Li, Jiajie Jin, Guanting Dong, Hongjin Qian, Yongkang Wu, Ji-Rong Wen, Yutao Zhu, and Zhicheng Dou. Webthinker: Empowering large reasoning models with deep research capability. arXiv preprint arXiv:2504.21776, 2025 d

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  18. [18]

    DeepSeek-V3 Technical Report

    Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, et al. Deepseek-v3 technical report. arXiv preprint arXiv:2412.19437, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  19. [19]

    Proceedings of the 44th Interna- tional ACM SIGIR Conference on Research and De- velopment in Information Retrieval

    Junteng Liu, Yunji Li, Chi Zhang, Jingyang Li, Aili Chen, Ke Ji, Weiyu Cheng, Zijia Wu, Chengyu Du, Qidi Xu, et al. Webexplorer: Explore and evolve for training long-horizon web agents. arXiv preprint arXiv:2509.06501, 2025 a

    work page arXiv 2025

  20. [20]

    ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models

    Mingjie Liu, Shizhe Diao, Ximing Lu, Jian Hu, Xin Dong, Yejin Choi, Jan Kautz, and Yi Dong. Prorl: Prolonged reinforcement learning expands reasoning boundaries in large language models. arXiv preprint arXiv:2505.24864, 2025 b

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  21. [21]

    Gaia: a benchmark for general ai assistants

    Gr \'e goire Mialon, Cl \'e mentine Fourrier, Thomas Wolf, Yann LeCun, and Thomas Scialom. Gaia: a benchmark for general ai assistants. In The Twelfth International Conference on Learning Representations, 2023

    work page 2023

  22. [22]

    Minimax m2 & agent: Ingenious in simplicity

    MiniMax AI . Minimax m2 & agent: Ingenious in simplicity. https://www.minimax.io/news/minimax-m2, 2025

    work page 2025

  23. [23]

    Kimi-researcher: End-to-end rl training for emerging agentic capabilities

    Moonshot AI . Kimi-researcher: End-to-end rl training for emerging agentic capabilities. https://moonshotai.github.io/Kimi-Researcher/, 2025

    work page 2025

  24. [24]

    SFR-DeepResearch: Towards Effective Reinforcement Learning for Autonomously Reasoning Single Agents

    Xuan-Phi Nguyen, Shrey Pandit, Revanth Gangi Reddy, Austin Xu, Silvio Savarese, Caiming Xiong, and Shafiq Joty. Sfr-deepresearch: Towards effective reinforcement learning for autonomously reasoning single agents. arXiv preprint arXiv:2509.06283, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  25. [25]

    Introducing gpt-5

    OpenAI . Introducing gpt-5. https://openai.com/index/introducing-gpt-5/, 2025

    work page 2025

  26. [26]

    Sealqa: Raising the bar for reasoning in search-augmented language models

    Thinh Pham, Nguyen Nguyen, Pratibha Zunjare, Weiyuan Chen, Yu-Min Tseng, and Tu Vu. Sealqa: Raising the bar for reasoning in search-augmented language models. In ICLR, 2026

    work page 2026

  27. [27]

    Humanity's Last Exam

    Long Phan, Alice Gatti, Ziwen Han, Nathaniel Li, et al. Humanity's last exam. arXiv preprint arXiv:2501.14249, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  28. [28]

    DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, et al. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  29. [29]

    TaskCraft: Automated Generation of Agentic Tasks

    Dingfeng Shi, Jingyi Cao, Qianben Chen, Weichen Sun, Weizhen Li, Hongxuan Lu, Fangchen Dong, Tianrui Qin, King Zhu, Minghao Yang, Jian Yang, Ge Zhang, Jiaheng Liu, Changwang Zhang, Jun Wang, Yuchen Eleanor Jiang, and Wangchunshu Zhou. Taskcraft: Automated generation of agentic tasks, 2025. URL https://arxiv.org/abs/2506.10055

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  30. [30]

    ZeroSearch: Incentivize the Search Capability of LLMs without Searching

    Hao Sun, Zile Qiao, Jiayan Guo, Xuanbo Fan, Yingyan Hou, Yong Jiang, Pengjun Xie, Yan Zhang, Fei Huang, and Jingren Zhou. Zerosearch: Incentivize the search capability of llms without searching. arXiv preprint arXiv:2505.04588, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  31. [31]

    Beyond turn limits: Training deep search agents with dynamic context window

    Qiaoyu Tang, Hao Xiang, Le Yu, Bowen Yu, Yaojie Lu, Xianpei Han, Le Sun, WenJuan Zhang, Pengbo Wang, Shixuan Liu, et al. Beyond turn limits: Training deep search agents with dynamic context window. arXiv preprint arXiv:2510.08276, 2025

    work page arXiv 2025

  32. [32]

    Kimi Team, Angang Du, Bofei Gao, Bowei Xing, Changjiu Jiang, Cheng Chen, Cheng Li, Chenjun Xiao, Chenzhuang Du, Chonghua Liao, et al. Kimi k1. 5: Scaling reinforcement learning with llms. arXiv preprint arXiv:2501.12599, 2025 a

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  33. [33]

    MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling

    MiroMind Team, Song Bai, Lidong Bing, Carson Chen, Guanzheng Chen, Yuntao Chen, Zhe Chen, Ziyi Chen, Jifeng Dai, Xuan Dong, et al. Mirothinker: Pushing the performance boundaries of open-source research agents via model, context, and interactive scaling. arXiv preprint arXiv:2511.11793, 2025 b

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  34. [34]

    Mirorl: An mcp-first reinforcement learning framework for deep research agent

    MiroMind Foundation Model Team and MiroMind AI Infra Team. Mirorl: An mcp-first reinforcement learning framework for deep research agent. https://github.com/MiroMindAI/MiroRL, 2025

    work page 2025

  35. [35]

    Tongyi DeepResearch Technical Report

    Tongyi DeepResearch Team, Baixuan Li, Bo Zhang, Dingchu Zhang, Fei Huang, Guangyu Li, Guoxin Chen, Huifeng Yin, Jialong Wu, Jingren Zhou, et al. Tongyi deepresearch technical report. arXiv preprint arXiv:2510.24701, 2025 c

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  36. [36]

    Milvus: A purpose-built vector data management system

    Jianguo Wang, Xiaomeng Yi, Rentong Guo, Hai Jin, Peng Xu, Shengjun Li, Xiangyu Wang, Xiangzhou Guo, Chengming Li, Xiaohai Xu, et al. Milvus: A purpose-built vector data management system. In Proceedings of the 2021 international conference on management of data, pp.\ 2614--2627, 2021

    work page 2021

  37. [37]

    Chain-of-thought prompting elicits reasoning in large language models

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35: 0 24824--24837, 2022

    work page 2022

  38. [38]

    BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents

    Jason Wei, Zhiqing Sun, Spencer Papay, Scott McKinney, Jeffrey Han, Isa Fulford, Hyung Won Chung, et al. Browsecomp: A simple yet challenging benchmark for browsing agents. arXiv preprint arXiv:2504.12516, 2025 a

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  39. [39]

    SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution

    Yuxiang Wei, Olivier Duchenne, Jade Copet, Quentin Carbonneaux, Lingming Zhang, Daniel Fried, Gabriel Synnaeve, Rishabh Singh, and Sida I Wang. Swe-rl: Advancing llm reasoning via reinforcement learning on open software evolution. arXiv preprint arXiv:2502.18449, 2025 b

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  40. [40]

    WebDancer: Towards Autonomous Information Seeking Agency

    Jialong Wu, Baixuan Li, Runnan Fang, Wenbiao Yin, Liwen Zhang, Zhengwei Tao, Dingchu Zhang, Zekun Xi, Gang Fu, Yong Jiang, et al. Webdancer: Towards autonomous information seeking agency. arXiv preprint arXiv:2505.22648, 2025 a

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  41. [41]

    WebWalker: Benchmarking LLMs in Web Traversal

    Jialong Wu, Wenbiao Yin, Yong Jiang, et al. Webwalker: Benchmarking llms in web traversal. arXiv preprint arXiv:2501.07572, 2025 b

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  42. [42]

    Qwen3 Technical Report

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. Qwen3 technical report. arXiv preprint arXiv:2505.09388, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  43. [43]

    React: Synergizing reasoning and acting in language models

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. React: Synergizing reasoning and acting in language models. In The eleventh international conference on learning representations, 2022

    work page 2022

  44. [44]

    Tree of thoughts: Deliberate problem solving with large language models

    Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Tom Griffiths, Yuan Cao, and Karthik Narasimhan. Tree of thoughts: Deliberate problem solving with large language models. Advances in neural information processing systems, 36: 0 11809--11822, 2023

    work page 2023

  45. [45]

    DAPO: An Open-Source LLM Reinforcement Learning System at Scale

    Qiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan, Xiaochen Zuo, Yu Yue, Weinan Dai, Tiantian Fan, Gaohong Liu, Lingjun Liu, et al. Dapo: An open-source llm reinforcement learning system at scale. arXiv preprint arXiv:2503.14476, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  46. [46]

    GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

    Aohan Zeng, Xin Lv, Qinkai Zheng, Zhenyu Hou, Bin Chen, Chengxing Xie, Cunxiang Wang, Da Yin, Hao Zeng, Jiajie Zhang, et al. Glm-4.5: Agentic, reasoning, and coding (arc) foundation models. arXiv preprint arXiv:2508.06471, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  47. [47]

    Deepresearcher: Scaling deep research via reinforcement learning in real-world environments

    Yuxiang Zheng, Dayuan Fu, Xiangkun Hu, Xiaojie Cai, Lyumanshan Ye, Pengrui Lu, and Pengfei Liu. Deepresearcher: Scaling deep research via reinforcement learning in real-world environments. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pp.\ 414--431, 2025

    work page 2025

  48. [48]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

    work page