Skip to content

archived edition · 2026-09-18

PaperraftFind the AI research
worth testing.

Understand what a paper changes, what it costs, and whether it fits your next build. Explore research on faster LLM inference, agent memory, and evaluation, with the original sources in reach.

Explore the researchHow papers are assessed

Start with a brief. Follow the evidence. Choose a test.

edition
2026-09-18
briefs
42
research areas
17
independent implementations
36

19 of 42 papers in the index (45%) have no independent implementations.

Research index

Search by topic, browse a research area, or sort by date. Read a brief, open the original paper, and inspect the available evidence before deciding what to test.

showing30 of 42
sort by
arXiv 2608.26480
First page of Zero-Shot Self-Orchestration with Ledger-Based Control for Improved LLM Coding Performance
multi-agent orchestrationtest

Zero-Shot Self-Orchestration with Ledger-Based Control for Improved LLM Coding Performance

The method replaces single-pass LLM code generation with a manager-worker scaffold in which short worker calls share state through a filesystem ledger, requiring no training or per-benchmark tuning. It roughly triples the token bill per problem, though it can buy accuracy more cheaply than upgrading to a larger model and is self-hostable on a 24 GB GPU only via small open-weight workers such as Qwen-27B-class models. It can fail because gains are strongly model-conditional: some models show null or negative deltas (e.g., Qwen3.6-35B, -1 to -9 points with reasoning off), and large reasoning-enabled models gain little, so the extra token cost may buy nothing on a given workload.

observed signal
2impl.
1stars
0citations
5.2xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2608.21690
First page of Context as an Environment: Programmatic Context Management for Long-Horizon Agents
memory and contexttest

Context as an Environment: Programmatic Context Management for Long-Horizon Agents

Scroll replaces prompt-side history compression and fixed memory representations with an append-only event log plus a persistent sandboxed Python kernel, where the model writes code to search and materialize state and only printed projections enter the working context. It costs engineering effort to build and sandbox the kernel and eviction index, adds code-execution latency per turn, and its reported results depend on a strong proprietary backbone. It can fail if the model writes faulty state-management code, if eviction landmarks mislead retrieval of evicted regions, or if sandboxing is insufficient against unsafe exec output.

observed signal
5impl.
42stars
0citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19383
First page of FCx: An algorithm for finding Feasible Counterfactual Explanations
othertest

FCx: An algorithm for finding Feasible Counterfactual Explanations

FCx replaces standard counterfactual explanation generators (e.g., Wachter-style optimization or VAE-based methods) that can output infeasible changes, by adding hard user-specified constraints and soft causal constraints inferred from the data, with realism enforced via LOF density and a multi-factor loss. The cost is training and tuning a modified VAE per dataset with a multi-term objective, plus the need for domain knowledge or reliable causal inference to define feasibility, which fits a single 24 GB GPU but adds pipeline complexity. It can fail if the inferred soft constraints encode incorrect causal assumptions, if LOF density poorly characterizes the data manifold on sparse or high-dimensional tabular data, or if feasibility constraints leave no valid counterfactual in dense regions.

observed signal
1impl.
4stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20050
First page of The Missing Complement: State-Conditioned Minimal Sufficient Evidence for Coding Agents
retrieval and RAGtest

The Missing Complement: State-Conditioned Minimal Sufficient Evidence for Coding Agents

It replaces per-passage similarity ranking and reranking in coding-agent retrieval with a set-construction policy: three semantic calls propose a jointly sufficient evidence set, search for what the current agent state still lacks, and return 4-8 intact source units within a 6,144-token budget. The cost is three sequential LLM-class semantic calls per retrieval step plus agent-state capture infrastructure, which adds latency and API spend relative to a single embedding query, and the operating configuration must be fixed on calibration data. It can fail when the captured state does not reflect what the decision actually requires, when repositories fall outside the 45-repository calibration distribution, and because the reported gains come from a single 500-state benchmark whose sufficiency annotations may not transfer to other codebases or executors.

observed signal
2impl.
14stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19657
First page of PrefixBench-H100: Characterizing Prefix Reuse and Time-to-First-Token in H100 LLM Serving
KV cachetest

PrefixBench-H100: Characterizing Prefix Reuse and Time-to-First-Token in H100 LLM Serving

The paper replaces ad-hoc assumptions about KV-cache prefix reuse with a reproducible benchmark measuring time-to-first-token, throughput, and cache-hit behavior on a single H100 across vLLM and TensorRT-LLM. Adoption costs only benchmarking time, since prefix caching is already available in vLLM, but acting on the results requires matching the benchmark's workload dimensions (shared-prefix length, concurrency, cache size) to one's own traffic. The findings can fail to transfer because they were measured on an H100 at serving scales beyond a single 24 GB GPU, where memory pressure and cache-granularity regimes differ, and residual benefits depend on the runtime's scheduling layer.

observed signal
1impl.
4stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19754
First page of AutoData: Agentic Search for Pre-training Data Selection
efficient trainingwatch

AutoData: Agentic Search for Pre-training Data Selection

AutoData replaces hand-designed pre-training data curation heuristics (mixture weights, quality filters) with an LLM agent that iteratively searches a program space of scoring, stratification, and stochastic selection rules, refined against validation feedback from a small proxy model. The cost is a search loop requiring repeated proxy-model pre-training runs (an 'overnight search' in the paper), plus agent orchestration and evaluation infrastructure, which exceeds a single-GPU constrained budget. The method can fail through overfitting the selection recipe to the proxy model or proxy benchmark, non-transfer to the reader's data distribution, and selection rules that exploit quirks of per-document features (lexical statistics, perplexity) rather than true data quality.

observed signal
2impl.
131stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20715
First page of Don't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL
coding agentstest

Don't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL

ActObs replaces the standard SFT convention of computing loss only on agent action tokens by also supervising the environment observation tokens already present in each trajectory, before running GRPO. It adds no data, parameters, sequence tokens, or forward passes; the cost is a modified loss mask in SFT and the measured pass@1 reliability regression at 8B. It can fail when the deployment budget allows only a single sample (pass@1), since the benefit concentrates in pass@k and exploration diversity, and results are demonstrated only on Qwen3 4B/8B with GRPO on terminal and code-editing tasks.

observed signal
1impl.
10stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20804
First page of An Empirical Study of Harness Design for Coding Agents
coding agentsadopt

An Empirical Study of Harness Design for Coding Agents

This replaces treating coding harnesses as monolithic systems with component-level selection: rule-based elision staged before LLM summarization, optional planning, and a bash-only or predefined-tool action space chosen per model. The cost is modest implementation effort for context management plus summarization tokens, and recoverable-elision machinery that models rarely use and that yields no accuracy gain. Failures include context-overflow crashes under tight budgets if management is omitted, degraded performance when bash-only interfaces are given to models with weak bash proficiency, and planning overhead that does not help weaker models on cost or stronger models on accuracy.

observed signal
1impl.
10stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19969
First page of DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression
KV cachewatch

DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression

The method replaces a conventional full per-layer KV cache with cross-layer KV reuse in CSA2 attention combined with FP4 KV quantization, plus a replay strategy that shrinks what must persist on SSD or host memory. The cost is deep architectural coupling: the compression is baked into a custom Causal Encoder-Decoder MoE design with asymmetric prefill/decode activation, so it requires training and serving support for this exact architecture rather than being a drop-in change. It can fail downstream if FP4 KV quantization or cross-layer sharing degrades quality on long-context retrieval or agentic tasks outside the reported benchmarks, and none of the claims transfer to third-party API models the reader actually uses.

observed signal
2impl.
124stars
—citations
4xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20519
First page of SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
coding agentstest

SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness

SoL-Pi replaces the default agent harness (action execution, context compaction, observation handling, and delegated reading in tools such as Codex or Claude Code) with four mechanisms selected by recursive auto-research rollouts to cut token traffic at equal task performance. The cost is integration effort at the harness layer and dependence on mechanisms discovered through automated selection, plus added complexity in how observations and context are compacted for the model. Failure modes include compaction or delegated reading dropping information needed on task distributions unlike EdgeBench, mechanisms that may not transfer to other model providers or harnesses, and selection bias toward the 51-task evaluation used to validate them.

observed signal
2impl.
130stars
—citations
1.81xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19705
First page of SoK: Trading Agents or Market Crashers? Dissecting Robustness and Security Failures in Academic Financial LLM Trading Schemes
safety and guardrailswatch

SoK: Trading Agents or Market Crashers? Dissecting Robustness and Security Failures in Academic Financial LLM Trading Schemes

FARSIGHT replaces ad-hoc or absent red-teaming of financial LLM trading agents with a scheme-level evaluation harness covering robustness under market turbulence and three attack classes: compromised information sources, direct attacks on agents, and agent-as-attacker behaviors. The cost is integration effort: adapting one's agent to the harness, running turbulence and adversarial scenarios, and remediating findings, with no claim that it reduces inference cost or latency. It can fail by producing false assurance, since passing its scenarios does not guarantee safety against novel adversaries or live-market conditions outside the tested distributions, and the paper itself is an evaluation study rather than a deployable defense.

observed signal
1impl.
14stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20152
First page of MTVA-Bench: Evaluating the Language Model Inside Cascaded Voice Agents
agent evaluationtest

MTVA-Bench: Evaluating the Language Model Inside Cascaded Voice Agents

MTVA-Bench replaces end-to-end voice pipeline scoring and generic LLM benchmarks with an isolated evaluation of the language model under realistic call conditions (ASR errors, split utterances, script and language constraints), using an LLM caller, a mock backend, deterministic tool-call checks, and two citation-requiring LLM judges across 490 scenarios in 7 languages. Adopting it costs the engineering effort of integrating an agent harness and mock backends plus recurring LLM judge inference, with no model retraining required. It can fail through judge unreliability or citation gaming, limited scenario coverage relative to a real production domain, and overfitting to the benchmark's simulated caller rather than actual human callers.

observed signal
2impl.
131stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19721
First page of LearnActCoder: Role-Aware Error Memory for Adaptive Clinical Coding Agents
memory and contexttest

LearnActCoder: Role-Aware Error Memory for Adaptive Clinical Coding Agents

Replaces one-shot or raw-reflection prompting with a structured Mistake Knowledge Database built from a small labeled batch, routing false-negative lessons to a Coder and false-positive lessons to a Judge, all at inference time with no weight updates. Costs a labeled error-learning batch, an extra Judge pass per case (more API calls or local inference), and curation/maintenance of the memory store. Can fail through precision-recall shifts rather than net F1 gains (as on MIMIC-IV), through weak transfer of lessons across datasets or code systems, and because the evidence base is a small retrospective evaluation with low absolute CPT performance.

observed signal
2impl.
134stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19759
First page of Rethinking Multi-Agent Collaboration: When More Is Less
multi-agent orchestrationtest

Rethinking Multi-Agent Collaboration: When More Is Less

SAIGE replaces fixed multi-agent pipelines and monolithic single-agent harnesses with a dynamically evolving collaboration graph in which agents are spawned on demand and edges are semantic dependencies established via content-based retrieval, reducing context overhead. The cost is added orchestration complexity, retrieval infrastructure for dependency edges, and the engineering burden of incremental graph management. It can fail on tightly coupled sequential workflows, where the paper itself finds single-agent harnesses superior, and scaling the agent pool or recursion depth does not reliably improve outcomes.

observed signal
2impl.
134stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20089
First page of UnifiedPlayers: Enhance Tool-Integrated Reasoning in Agentic Reinforcement Learning
tool usewatch

UnifiedPlayers: Enhance Tool-Integrated Reasoning in Agentic Reinforcement Learning

UnifiedPlayers replaces static verifiers and self-consistency reward signals in self-evolving tool-integrated reasoning with three jointly trained players (task planner, trajectory executor, executable-verifier constructor) coordinated under GRPO. It costs a full reinforcement learning pipeline training three interacting model roles on model backbones, which exceeds a single 24 GB GPU budget and adds substantial engineering complexity around reward design and co-adaptation stability. It can fail through coordination instability, where each player's updates shift the data and feedback distribution of the others, and through verifier reward hacking if the evaluation player is exploited by the executor.

observed signal
2impl.
134stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19868
First page of Zarya: A Hybrid Autoregressive--Masked Diffusion Language Model with Flexible Training and Dual-Mode Inference
speculative decodingtest

Zarya: A Hybrid Autoregressive--Masked Diffusion Language Model with Flexible Training and Dual-Mode Inference

Zarya replaces a purely autoregressive decoding stack with a single model trained jointly on AR and masked-diffusion objectives, offering either parallel MDM sampling or slotted speculative decoding that interleaves diffusion-based slot selection with AR infilling while reusing the KV cache. The cost is a custom training curriculum (variable-size slots with increasing granularity) and a non-standard inference interface; pure MDM mode still forfeits KV-cache reuse and incurs its characteristic compute overhead, and no off-the-shelf serving stack supports either decoding mode. Failures after adoption include incoherent generation in diffusion mode, immature tooling and reproducibility risk for the speculative mode, and unverified speedups, since the abstract reports benchmark participation but no quantified latency or quality advantage over a standard AR baseline at the released 0.6B–4B scales.

observed signal
1impl.
120stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20761
First page of Agile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Control
othernot applicable

Agile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Control

Agile-WAM replaces large pretrained generative backbones in tactile world action models with a compact shared-latent architecture that directly flow-matches vision-tactile observations to action chunks and future latents, using multi-horizon supervision tuned to each modality's timescale. The cost is a full perception-to-control training pipeline requiring synchronized visual and tactile robot data, plus contact-rich hardware (tactile sensors, manipulators) that is not part of a typical 24 GB GPU or API-based stack. Failure modes include sim-to-real transfer gaps on the nine simulated tasks, tactile sensor drift or noise degrading the fine-grained contact prediction, and brittleness outside the demonstrated task distribution.

observed signal
1impl.
107stars
—citations
1.294xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20441
First page of Cross-Architecture Foundation-Model Distillation for Edge Flood Segmentation
distillationadopt

Cross-Architecture Foundation-Model Distillation for Edge Flood Segmentation

The method replaces direct supervised training of a small model on a fixed labeled set, and replaces edge deployment of a large foundation model, with teacher-generated pseudo-labels on unlabeled imagery used to train a compact student. The cost is a full teacher fine-tuning and inference pass over unlabeled data plus quantization-aware training and TensorRT export, with a residual quality gap on the primary benchmark (0.787 vs 0.822 water IoU). Failure modes include the student remaining below the teacher on the harder WorldFloods-v2 distribution, pseudo-label noise propagating at larger pool scales, and the possibility that a fixed MNDWI threshold matches the learned models on clean external data, limiting the practical advantage of the model.

observed signal
1impl.
120stars
—citations
428.6xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19880
First page of D-Quant: Driftable Entropy Coding for KV Cache Quantization
KV cachetest

D-Quant: Driftable Entropy Coding for KV Cache Quantization

D-Quant replaces fixed-width KV cache quantization with entropy-coded representations whose variable-length outputs are converted, via a drift mechanism, into fixed-size bitstreams so attention kernels retain regular memory access and parallel dequantization. It costs implementation complexity: rotation and normalization of KV values, custom dequantization kernels handling the drifted fixed-size streams, and potential dequantization latency if kernel integration is not optimized. It can fail if the near-normal distributional assumptions do not hold across models or layers, if the drift overhead erodes the bit savings over simpler 4-bit or 8-bit baselines, or if no maintained kernel exists for the reader's inference stack.

observed signal
1impl.
124stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20186
First page of To Copy or Not to Copy: Controlling Speculative Decoding via Intrinsic Model Signals
speculative decodingtest

To Copy or Not to Copy: Controlling Speculative Decoding via Intrinsic Model Signals

The method replaces fixed speculative-decoding drafting strategies (pure neural drafting such as EAGLE3, or heuristic n-gram copy triggering) with a runtime switch driven by lightweight probes trained on the target model's hidden states to detect copy intent. It costs the existing EAGLE-style drafter infrastructure plus training and maintaining per-model probes, and adds probe-inference overhead at each switching decision; memory cost is small but engineering complexity increases. It can fail if probe accuracy degrades on domains or models different from the training distribution, and the 15% gain depends on workloads containing genuine copy-intensive spans, offering little benefit on free-form generation.

observed signal
1impl.
124stars
—citations
1.15xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20474
First page of How Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents
planning and decompositionadopt

How Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents

Replaces free-form or unguided agent prompting and implicit self-judgment of completion with prewritten task-specific plans and a read-only terminal verifier that checks completion before acceptance. Costs are low: plan authoring per task type and under one cent per episode for verification, with no additional model training or infrastructure. The verifier produces false rejections (17% of correct episodes withheld), plan quality degrades on tasks outside the authored templates, and the measured gains come from Retail and Airline tool-use benchmarks and may not transfer to other domains.

observed signal
1impl.
130stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20625
First page of Chronicle: Cut-Point Replay for Regression Testing of LLM Agents
agent evaluationtest

Chronicle: Cut-Point Replay for Regression Testing of LLM Agents

The method replaces ad-hoc manual reproduction of agent failures and full live re-runs with recorded envelopes at model and tool boundaries, replayed selectively so that only the code under test executes live, turning each incident into a deterministic CI regression test. The cost is instrumentation of all non-deterministic boundaries, storage of immutable run records, and 23 microseconds of recording overhead per crossing, plus the engineering effort to maintain replay-compatible boundaries as the agent evolves. It can fail when trajectories drift so that the recorded suffix no longer matches the new code's live prefix, when tools or models change their interfaces, and because the evidence base is only six incidents with simulated model boundaries, leaving real-model fidelity unproven.

observed signal
1impl.
130stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.20807
First page of Score Centering Stabilizes Off-policy Reinforcement Learning
efficient trainingtest

Score Centering Stabilizes Off-policy Reinforcement Learning

Score centering replaces (or complements) importance-sampling corrections for training-inference mismatch in RL post-training of LLMs by adding an additive drift-canceling term to the score rather than reweighting samples. It costs little compute or memory since the correction is additive, but adds implementation complexity in the RL training loop and requires a reliable estimate of the persistent bias between engines. It can fail if the mismatch is not primarily drift (e.g., high-variance noise rather than persistent bias), if the drift estimate itself is noisy or unstable, or in setups where training and inference engines can simply be kept consistent.

observed signal
1impl.
130stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19243
First page of Randomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices
othertest

Randomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices

The methods replace the full SVD used in normalized spectral co-clustering of word-document matrices with a random-projection randomized SVD or a partial SVD combined with element-wise sampling. The cost is approximation error in the spectral embedding, which can degrade cluster quality, and the added implementation complexity of maintaining two algorithm variants. The sampling-based variant can fail on already sparse text matrices, where it provides limited benefit, so the choice of approximation depends on the sparsity structure of the data.

observed signal
0impl.
4stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19670
First page of CoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting
othertest

CoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting

CoRe replaces pointwise MSE-style supervision in direct multivariate time-series forecasting with two output-space constraints: a frequency-domain coherence loss aligning predicted and target spectra, and a low-rank relational graph loss matching pairwise differences in a PCA subspace. It adds no trainable parameters and only changes the loss, so compute and memory overhead are modest, but it introduces extra hyperparameters and per-batch spectral and PCA computations that must be implemented and tuned. It can fail when target spectra are noisy or nonstationary, when cross-variable relationships are weak or unstable so the PCA subspace misrepresents them, or on datasets where pointwise error is already well correlated with the deployment metric.

observed signal
0impl.
4stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19683
First page of MiX: Micro-Inverted-Scaling for End-to-End Low-Bit Vision-Language Model Acceleration
quantizationwatch

MiX: Micro-Inverted-Scaling for End-to-End Low-Bit Vision-Language Model Acceleration

MiX replaces shared-exponent microscaling block formats (NVFP4/MX-style) with a format grouping private per-element exponents under one shared mantissa, so multipliers in the datapath become shifters, and pairs it with a dual-format inference framework for VLM outlier topologies. The cost is that the claimed speed, energy, and area gains exist only on a custom accelerator that does not exist as a product; on the reader's GPU there is no hardware support for this format, so only the quantization accuracy result is indirectly relevant. It can fail in practice because accuracy parity was shown on selected VLMs and benchmarks, the dual-format path adds format-selection complexity, and simulated accelerator results often do not survive contact with real silicon or vendor toolchains.

observed signal
0impl.
4stars
—citations
4.5xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19702
First page of Understanding and Exploiting Diagonal Attention Sparsity in Autoregressive Image Generation
attention and kernelstest

Understanding and Exploiting Diagonal Attention Sparsity in Autoregressive Image Generation

Replaces dense attention during decoding in autoregressive image generation by skipping KV entries along the diagonal attention direction within a recent window, exploiting spatial locality of visual tokens. Costs custom kernel implementation and integration effort (built on FlexGen, FlashAttention-2, and custom kernels) plus up to roughly 2% output quality degradation. Can fail if the reader's model or workload does not exhibit the measured diagonal sparsity pattern, if quality-sensitive applications cannot tolerate the degradation, or if the engineering overhead of maintaining custom kernels outweighs the throughput gain at small request volumes.

observed signal
0impl.
4stars
—citations
3.1xclaimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19743
First page of Syndrome Decoding for Silent Data Corruption in Quantized Integer GPU Arithmetic
failure recoverywatch

Syndrome Decoding for Silent Data Corruption in Quantized Integer GPU Arithmetic

SProbe replaces binary checksum-based ABFT (which detects but cannot localize or correct errors, and is blind to axis-cancelling error patterns) with a trailing verification kernel that pinpoints the column and exact magnitude of errors via Reed-Solomon syndrome decoding, then repairs or recomputes the GEMM. The cost is 11-49% of vendor GEMM time depending on size, roughly 30% end-to-end inference throughput, plus integration of a nontrivial verification and decoding pipeline. It can fail to be worth adopting at all: it addresses transient hardware faults in INT32 tensor-core accumulators, which are rare events whose relevance depends on deployment criticality, and the method was validated on an H100 that the reader may not possess, with correctness gains that matter mainly for safety-critical quantized inference rather than typical LLM serving.

observed signal
0impl.
4stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19830
First page of Dual-Axis Policy Optimization for LLM Agents: Bayesian Feedback Attribution and Trajectory Mass Normalization
efficient trainingtest

Dual-Axis Policy Optimization for LLM Agents: Bayesian Feedback Attribution and Trajectory Mass Normalization

It replaces the credit-assignment and batch-aggregation components of standard GRPO/GiGPO with Bayesian feedback attribution within trajectories and equal-mass normalization across trajectories, leaving the underlying RL loop and base model unchanged. The cost is additional inference-time posterior computation per step, rollout collection for RL, and extra hyperparameters, all within a training workflow that still demands substantial sampling. Failure modes include mis-specified feedback posteriors in environments with sparse or noisy rewards, instability from equalizing trajectory mass when task difficulty is highly heterogeneous, and the possibility that gains measured on small agentic benchmarks do not transfer to production task distributions.

observed signal
0impl.
10stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗
arXiv 2609.19906
First page of Learning and Transferring Closed-Loop Robot Software
coding agentstest

Learning and Transferring Closed-Loop Robot Software

Replaces manual design and tuning of closed-loop robot policy code (observation processing, state management, situation-dependent branching) with coding-agent generation that is iteratively refined by simulation feedback, archived, and reused as reference implementations when acquiring policies for new tasks. Costs consist of repeated coding-agent API calls and simulation rollouts during optimization and archive construction, plus the need for task demonstrations and a relevant simulator; deployment itself is inexpensive because the selected policy is frozen and executes without further model calls. Transfer is inconsistent (initial unoptimized references were better on two of nine target tasks), all evidence is simulation-only so the sim-to-real gap is unquantified, and the 57% mean success rate is below the reliability most production settings require.

observed signal
0impl.
0stars
—citations
—claimed gain
abstract onlyReview decisionDownload skill ↓Original paper ↗

Research signals

Three views of the same evidence: adoption against attention, publication cadence by research area, and reported gains when coverage is sufficient.

color = research area

01 · attention × adoption

The research frontier

Look to the upper left for independent implementation before mass attention.

compare by
Each point represents one paper. Higher means more independent implementations; farther right means more GitHub stars.Context as an Environment: Programmatic Context Management for Long-Horizon Agents: 42 GitHub stars; 5 independent implementationsZero-Shot Self-Orchestration with Ledger-Based Control for Improved LLM Coding Performance: 1 GitHub star; 2 independent implementationsRandomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices: 4 GitHub stars; 0 independent implementationsFCx: An algorithm for finding Feasible Counterfactual Explanations: 4 GitHub stars; 1 independent implementationPrefixBench-H100: Characterizing Prefix Reuse and Time-to-First-Token in H100 LLM Serving: 4 GitHub stars; 1 independent implementationCoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting: 4 GitHub stars; 0 independent implementationsMiX: Micro-Inverted-Scaling for End-to-End Low-Bit Vision-Language Model Acceleration: 4 GitHub stars; 0 independent implementationsUnderstanding and Exploiting Diagonal Attention Sparsity in Autoregressive Image Generation: 4 GitHub stars; 0 independent implementationsSoK: Trading Agents or Market Crashers? Dissecting Robustness and Security Failures in Academic Financial LLM Trading Schemes: 14 GitHub stars; 1 independent implementationLearnActCoder: Role-Aware Error Memory for Adaptive Clinical Coding Agents: 134 GitHub stars; 2 independent implementationsSyndrome Decoding for Silent Data Corruption in Quantized Integer GPU Arithmetic: 4 GitHub stars; 0 independent implementationsAutoData: Agentic Search for Pre-training Data Selection: 131 GitHub stars; 2 independent implementationsRethinking Multi-Agent Collaboration: When More Is Less: 134 GitHub stars; 2 independent implementationsDual-Axis Policy Optimization for LLM Agents: Bayesian Feedback Attribution and Trajectory Mass Normalization: 10 GitHub stars; 0 independent implementationsZarya: A Hybrid Autoregressive--Masked Diffusion Language Model with Flexible Training and Dual-Mode Inference: 120 GitHub stars; 1 independent implementationD-Quant: Driftable Entropy Coding for KV Cache Quantization: 124 GitHub stars; 1 independent implementationLearning and Transferring Closed-Loop Robot Software: 0 GitHub stars; 0 independent implementationsDeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression: 124 GitHub stars; 2 independent implementationsThe Missing Complement: State-Conditioned Minimal Sufficient Evidence for Coding Agents: 14 GitHub stars; 2 independent implementationsMarginal utility, matrix factorization, and the Key-Value (KV) cache: a unified information-economic framework for sovereign geo-mining inference: 120 GitHub stars; 0 independent implementationsMATCH: Model-Aware Tool Learning with Curriculum Scheduling and Hierarchically Gated Rewards: 4 GitHub stars; 0 independent implementationsUnifiedPlayers: Enhance Tool-Integrated Reasoning in Agentic Reinforcement Learning: 134 GitHub stars; 2 independent implementationsLocal Sparsity Enables Unsupervised LLM Safety Detection: 4 GitHub stars; 0 independent implementationsMTVA-Bench: Evaluating the Language Model Inside Cascaded Voice Agents: 131 GitHub stars; 2 independent implementationsQUALS: Corpus Equilibrium for Universal Forecasting via Pattern Quantization and Learnability Synchronization: 0 GitHub stars; 0 independent implementationsTo Copy or Not to Copy: Controlling Speculative Decoding via Intrinsic Model Signals: 124 GitHub stars; 1 independent implementationCross-Architecture Foundation-Model Distillation for Edge Flood Segmentation: 120 GitHub stars; 1 independent implementationSpotlights: Discovering Improvement Opportunities in Software Repositories: 0 GitHub stars; 0 independent implementationsHow Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents: 130 GitHub stars; 1 independent implementationScaling Fourier-Based Sparse Matrix Analysis on GPUs: 0 GitHub stars; 0 independent implementationsTruncated automatic sparse differentiation for machine learning interatomic potentials: 1 GitHub star; 0 independent implementationsSoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness: 130 GitHub stars; 2 independent implementationsLanguage-model groups overstate consensus when replaying human deliberation on a reasoning task: 0 GitHub stars; 0 independent implementationsInference-Engine Fingerprinting Attacks are Practical: Exploring Model-Driven Environmental Discovery, Exploitation, and Escape: 0 GitHub stars; 0 independent implementationsChronicle: Cut-Point Replay for Regression Testing of LLM Agents: 130 GitHub stars; 1 independent implementationDon't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL: 10 GitHub stars; 1 independent implementationOn-Demand Attention: Language Models Know When to Recall: 120 GitHub stars; 0 independent implementationsAgile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Control: 107 GitHub stars; 1 independent implementationAn Empirical Study of Harness Design for Coding Agents: 10 GitHub stars; 1 independent implementationScore Centering Stabilizes Off-policy Reinforcement Learning: 130 GitHub stars; 1 independent implementationQuantifying Overclaiming Propensity in Frontier LLM Agents: 10 GitHub stars; 0 independent implementationsCoding Agents with an Obstacle-Aware Harness for Safe Robot Manipulation: 10 GitHub stars; 0 independent implementations

Papers above 1000 stars or 200 citations are not scored. They have already broken through the attention threshold, while Paperraft is designed to identify earlier signals.

02 · cadence

Research areas over time

Compare monthly publication volume on a shared scale and calendar.

coding agentsmax 5Aug 26: 0 papersSep 26: 5 papersAug 26Sep 26agent evaluationmax 5Aug 26: 0 papersSep 26: 4 papersAug 26Sep 26KV cachemax 5Aug 26: 0 papersSep 26: 3 papersAug 26Sep 26speculative decodingmax 5Aug 26: 0 papersSep 26: 2 papersAug 26Sep 26distillationmax 5Aug 26: 0 papersSep 26: 1 paperAug 26Sep 26sparsity and pruningmax 5Aug 26: 0 papersSep 26: 2 papersAug 26Sep 26attention and kernelsmax 5Aug 26: 0 papersSep 26: 2 papersAug 26Sep 26memory and contextmax 5Aug 26: 1 paperSep 26: 1 paperAug 26Sep 26multi-agent orchestrationmax 5Aug 26: 1 paperSep 26: 1 paperAug 26Sep 26othermax 5Aug 26: 0 papersSep 26: 4 papersAug 26Sep 26planning and decompositionmax 5Aug 26: 0 papersSep 26: 1 paperAug 26Sep 26quantizationmax 5Aug 26: 0 papersSep 26: 1 paperAug 26Sep 26failure recoverymax 5Aug 26: 0 papersSep 26: 1 paperAug 26Sep 26retrieval and RAGmax 5Aug 26: 0 papersSep 26: 1 paperAug 26Sep 26safety and guardrailsmax 5Aug 26: 0 papersSep 26: 4 papersAug 26Sep 26efficient trainingmax 5Aug 26: 0 papersSep 26: 5 papersAug 26Sep 26tool usemax 5Aug 26: 0 papersSep 26: 2 papersAug 26Sep 26

Every panel uses the same calendar and vertical scale. An empty column means no papers were recorded.

Method under scrutiny

Inspect the highest-ranked paper, every repository behind its signal, and the rule used to classify each one.

Zero-Shot Self-Orchestration with Ledger-Based Control for Improved LLM Coding Performance

multi-agent orchestration · test · 2 of 2 implementations are independent · 5.2x gain in cost (reported by the authors; not independently verified)

The method replaces single-pass LLM code generation with a manager-worker scaffold in which short worker calls share state through a filesystem ledger, requiring no training or per-benchmark tuning. It roughly triples the token bill per problem, though it can buy accuracy more cheaply than upgrading to a larger model and is self-hostable on a 24 GB GPU only via small open-weight workers such as Qwen-27B-class models. It can fail because gains are strongly model-conditional: some models show null or negative deltas (e.g., Qwen3.6-35B, -1 to -9 points with reasoning off), and large reasoning-enabled models gain little, so the extra token cost may buy nothing on a given workload.

Exclusions and controls

Every exclusion is counted and published.

Exclusion counts were not recorded for this edition.

0 previously indexed papers were rechecked in this run.

Paperraft tracks one signal: how many independent implementations a paper attracts on GitHub after author-owned repositories are removed. This is a discovery signal to investigate, not a measure of implementation quality.

It does not claim that a method works. Paperraft runs no reproduction benchmarks. Reported gains remain author claims until independently tested. Papers above the attention threshold are excluded because this publication is designed to identify research before consensus forms.