SWE-bench reveals that even top language models like Claude 2 resolve only 1.96% of 2,294 real-world GitHub issues, highlighting a gap in practical coding capabilities.
Title resolution pending
48 Pith papers cite this work. Polarity classification is still indexing.
48
Pith papers citing it
citation-role summary
background 3
citation-polarity summary
roles
background 3polarities
background 3representative citing papers
Rule2DRC is a benchmark for LLM agents synthesizing DRC scripts from natural language rules, paired with SplitTester that improves Best-of-N selection via execution-guided discriminative test generation.
Introduces TBPO, which derives a Bregman-divergence density-ratio matching objective for token-level preference optimization that generalizes DPO while preserving the induced optimal policy.
DARE reuses up to 87% of attention activations in diffusion LLMs through KV caching and output reuse, delivering 1.2x per-layer latency gains with average performance drops of 1.2-2.0%.
BWLA is the first post-training quantization method for LLMs that achieves 1-bit weights paired with low-bit activations such as 6 bits, using OKT to reshape weights and suppress activation tails plus PSP for low-rank refinement.
LLMs show up to 60.58% social bias in generated code; a new Fairness Monitor Agent cuts bias by 65.1% and raises functional correctness from 75.80% to 83.97%.
ResRL decouples shared semantics between positive and negative responses in LLM reinforcement learning via SVD-based projection residuals, outperforming baselines including NSR by up to 9.4% on math reasoning benchmarks.
LLM tutors leak answers under adversarial student attacks, but a fine-tuned jailbreak agent and simple defenses can benchmark and improve robustness.
TriMix dynamically fuses logits from three model sources to outperform baselines and Proxy Tuning on eight low-resource languages across four model families.
Code-switching creates a fundamental performance bottleneck for multilingual retrievers, causing drops of up to 27% on new benchmarks CSR-L and CS-MTEB, with embedding divergence as the key cause and vocabulary expansion insufficient to fix it.
SWE-RL uses RL on software evolution data to train LLMs achieving 41% on SWE-bench Verified with generalization to other reasoning tasks.
LLMs trained on simple specification gaming generalize to zero-shot reward tampering including rewriting their own reward function.
GAIA benchmark shows humans at 92% accuracy on simple real-world questions far outperform current AI systems at 15%, proposing this gap as a key milestone for general AI.
SDM sparsifies the Gated DeltaNet update rule to enable 1000x larger recurrent memory states at iso-FLOP, improving long-context recall and short-context reasoning over GDN and matching full attention at 8B scale.
A 4B compiler model generates LoRA adapters from natural-language specs, enabling a frozen 0.6B interpreter to match Qwen3-32B performance on fuzzy text tasks at 50× less memory.
Diffusion language models encode a decodable, steerable representation of denoising progress (the fraction of unmasked tokens) in their residual streams.
Controlled experiments show structured reasoning traces and higher-density math-domain samples improve mathematical reasoning more than pure executable code, with internal routing patterns reflecting these data effects.
DiagEval applies trajectory-conditioned diagnostic probes to recover 45.6-62.1% of misattributed failures in GUI-agent software evaluation, raising accuracy from 69.9% to 78.3% on WebDevJudge-Unit and 65.0% to 81.6% on RealDevBench.
Clarification-seeking in LLM agents amplifies prompt injection attack success from ~2% to over 30% across ten frontier models in a new 728-scenario benchmark.
SP-KV trains a utility predictor jointly with the LLM to dynamically prune low-utility KV cache entries, achieving 3-10x memory reduction during generation with negligible performance loss.
COPSD improves mathematical reasoning in low-resource languages by having LLMs self-distill from their own high-resource English behavior via token-level divergence on rollouts with privileged crosslingual context.
POETS uses compute-efficient LLM policy ensembles to implicitly perform KL-regularized Thompson sampling, delivering O(sqrt(T gamma_T)) regret bounds and state-of-the-art sample efficiency in scientific discovery tasks such as protein search and quantum circuit design.
DKPS-based methods predict new model benchmark scores using cached responses, matching baseline mean absolute error with substantially fewer queries and an offline query selection approach.
A blind replay script matches frontier model performance on static CUA benchmarks due to non-principled environments and evaluation methods, prompting PRISM design principles and the DigiWorld benchmark with improved statistical aggregation.
citing papers explorer
-
SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
SWE-bench reveals that even top language models like Claude 2 resolve only 1.96% of 2,294 real-world GitHub issues, highlighting a gap in practical coding capabilities.
-
Rule2DRC: Benchmarking LLM Agents for DRC Script Synthesis with Execution-Guided Test Generation
Rule2DRC is a benchmark for LLM agents synthesizing DRC scripts from natural language rules, paired with SplitTester that improves Best-of-N selection via execution-guided discriminative test generation.
-
TokenRatio: Principled Token-Level Preference Optimization via Ratio Matching
Introduces TBPO, which derives a Bregman-divergence density-ratio matching objective for token-level preference optimization that generalizes DPO while preserving the induced optimal policy.
-
DARE: Diffusion Language Model Activation Reuse for Efficient Inference
DARE reuses up to 87% of attention activations in diffusion LLMs through KV caching and output reuse, delivering 1.2x per-layer latency gains with average performance drops of 1.2-2.0%.
-
BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMs
BWLA is the first post-training quantization method for LLMs that achieves 1-bit weights paired with low-bit activations such as 6 bits, using OKT to reshape weights and suppress activation tails plus PSP for low-rank refinement.
-
Social Bias in LLM-Generated Code: Benchmark and Mitigation
LLMs show up to 60.58% social bias in generated code; a new Fairness Monitor Agent cuts bias by 65.1% and raises functional correctness from 75.80% to 83.97%.
-
ResRL: Boosting LLM Reasoning via Negative Sample Projection Residual Reinforcement Learning
ResRL decouples shared semantics between positive and negative responses in LLM reinforcement learning via SVD-based projection residuals, outperforming baselines including NSR by up to 9.4% on math reasoning benchmarks.
-
Evaluating Answer Leakage Robustness of LLM Tutors against Adversarial Student Attacks
LLM tutors leak answers under adversarial student attacks, but a fine-tuned jailbreak agent and simple defenses can benchmark and improve robustness.
-
Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion
TriMix dynamically fuses logits from three model sources to outperform baselines and Proxy Tuning on eight low-resource languages across four model families.
-
Code-Switching Information Retrieval: Benchmarks, Analysis, and the Limits of Current Retrievers
Code-switching creates a fundamental performance bottleneck for multilingual retrievers, causing drops of up to 27% on new benchmarks CSR-L and CS-MTEB, with embedding divergence as the key cause and vocabulary expansion insufficient to fix it.
-
SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
SWE-RL uses RL on software evolution data to train LLMs achieving 41% on SWE-bench Verified with generalization to other reasoning tasks.
-
Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
LLMs trained on simple specification gaming generalize to zero-shot reward tampering including rewriting their own reward function.
-
GAIA: a benchmark for General AI Assistants
GAIA benchmark shows humans at 92% accuracy on simple real-world questions far outperform current AI systems at 15%, proposing this gap as a key milestone for general AI.
-
Sparse Delta Memory: Scaling the State of Linear RNNs through Sparsity
SDM sparsifies the Gated DeltaNet update rule to enable 1000x larger recurrent memory states at iso-FLOP, improving long-context recall and short-context reasoning over GDN and matching full attention at 8B scale.
-
Program-as-Weights: A Programming Paradigm for Fuzzy Functions
A 4B compiler model generates LoRA adapters from natural-language specs, enabling a frozen 0.6B interpreter to match Qwen3-32B performance on fuzzy text tasks at 50× less memory.
-
Subliminal Clocks: Latent Time Modelling in Diffusion Language Models
Diffusion language models encode a decodable, steerable representation of denoising progress (the fraction of unmasked tokens) in their residual streams.
-
What Really Improves Mathematical Reasoning: Structured Reasoning Signals Beyond Pure Code
Controlled experiments show structured reasoning traces and higher-density math-domain samples improve mathematical reasoning more than pure executable code, with internal routing patterns reflecting these data effects.
-
DiagEval: Trajectory-Conditioned Diagnosis for Reliable Software Evaluation with GUI Agents
DiagEval applies trajectory-conditioned diagnostic probes to recover 45.6-62.1% of misattributed failures in GUI-agent software evaluation, raising accuracy from 69.9% to 78.3% on WebDevJudge-Unit and 65.0% to 81.6% on RealDevBench.
-
ASPI: Seeking Ambiguity Clarification Amplifies Prompt Injection Vulnerability in LLM Agents
Clarification-seeking in LLM agents amplifies prompt injection attack success from ~2% to over 30% across ten frontier models in a new 728-scenario benchmark.
-
Self-Pruned Key-Value Attention: Learning When to Write by Predicting Future Utility
SP-KV trains a utility predictor jointly with the LLM to dynamically prune low-utility KV cache entries, achieving 3-10x memory reduction during generation with negligible performance loss.
-
Crosslingual On-Policy Self-Distillation for Multilingual Reasoning
COPSD improves mathematical reasoning in low-resource languages by having LLMs self-distill from their own high-resource English behavior via token-level divergence on rollouts with privileged crosslingual context.
-
POETS: Uncertainty-Aware LLM Optimization via Compute-Efficient Policy Ensembles
POETS uses compute-efficient LLM policy ensembles to implicitly perform KL-regularized Thompson sampling, delivering O(sqrt(T gamma_T)) regret bounds and state-of-the-art sample efficiency in scientific discovery tasks such as protein search and quantum circuit design.
-
Query-efficient model evaluation using cached responses
DKPS-based methods predict new model benchmark scores using cached responses, matching baseline mean absolute error with substantially fewer queries and an offline query selection approach.
-
Computer Use at the Edge of the Statistical Precipice
A blind replay script matches frontier model performance on static CUA benchmarks due to non-principled environments and evaluation methods, prompting PRISM design principles and the DigiWorld benchmark with improved statistical aggregation.
-
Parallel Prefix Verification for Speculative Generation
PARSE accelerates LLM inference via parallel semantic prefix verification in a single forward pass, delivering 1.25x-4.3x speedups alone and up to 4.5x when combined with EAGLE-3.
-
Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decoding
EVICT adaptively truncates draft trees in MoE speculative decoding by combining drafter signals with profiled costs to retain only cost-effective prefixes, delivering up to 2.35x speedup over autoregressive decoding.
-
Learning to Communicate: Toward End-to-End Optimization of Multi-Agent Language Systems
DiffMAS jointly optimizes latent communication and reasoning in multi-agent LLM systems via parameter-efficient supervised training on trajectories, yielding consistent gains over baselines on math, science, and code benchmarks.
-
Learning to Correct: Calibrated Reinforcement Learning for Multi-Attempt Chain-of-Thought
CAL-GRPO calibrates per-attempt weights in multi-attempt CoT to deliver unbiased gradients for optimizing Verification@K success while keeping variance low.
-
Probabilistic Programs of Thought
Probabilistic programs of thought let LLMs produce many program variants from one generation by building a compact probabilistic representation of the token distribution.
-
Muon is Scalable for LLM Training
Muon optimizer with weight decay and update scaling achieves ~2x efficiency over AdamW for large LLMs, shown via the Moonlight 3B/16B MoE model trained on 5.7T tokens.
-
LIMO: Less is More for Reasoning
LIMO achieves 63.3% on AIME24 and 95.6% on MATH500 via supervised fine-tuning on roughly 1% of the data used by prior models, supporting the claim that minimal strategic examples suffice when pre-training has already encoded domain knowledge.
-
Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models
Empirical analysis shows scaling inference compute via strategies like tree search can be more efficient than scaling model parameters, with 7B models plus novel search outperforming 34B models.
-
ALLaVA: Harnessing GPT4V-Synthesized Data for Lite Vision-Language Models
ALLaVA creates 1.3M GPT4V-synthesized samples enabling 4B VLMs to achieve competitive results on 17 benchmarks and match 7B/13B models on some tasks.
-
Llemma: An Open Language Model For Mathematics
Continued pretraining of Code Llama on Proof-Pile-2 yields Llemma, an open math-specialized LLM that beats known open base models on MATH and supports tool use plus formal proving out of the box.
-
Language Models (Mostly) Know What They Know
Language models show good calibration when asked to estimate the probability that their own answers are correct, with performance improving as models get larger.
-
A General Language Assistant as a Laboratory for Alignment
Ranked preference modeling outperforms imitation learning for language model alignment and scales more favorably with model size.
-
CTFusion: A CTF-based Benchmark for LLM Agent Evaluation
Static CTF benchmarks for LLM agents are contamination-prone; CTFusion evaluates agents on live CTFs via an MCP server on CTFd with per-agent isolation.
-
CausalMix: Data Mixture as Causal Inference for Language Model Training
CausalMix fits a causal model on 512 runs of a 0.5B model to estimate CATE, then extrapolates optimal mixtures for an 800K data pool applied to 7B and 4B models, outperforming RegMix.
-
The Efficiency Gap in Byte Modeling
Byte modeling incurs greater scaling overhead for masked diffusion than autoregressive models because the diffusion objective destroys local byte contiguity needed to resolve semantics.
-
RADAR: Redundancy-Aware Diffusion for Multi-Agent Communication Structure Generation
RADAR generates query-adaptive multi-agent communication structures via conditional discrete graph diffusion guided by effective graph size, outperforming baselines on accuracy and token consumption across six benchmarks.
-
DACA-GRPO: Denoising-Aware Credit Assignment for Reinforcement Learning in Diffusion Language Models
DACA-GRPO adds denoising-aware credit assignment and bias-reduced likelihood estimation to GRPO, delivering consistent gains up to 36.3pp on math, code, constraint, and schema benchmarks for diffusion LLMs.
-
Learn-To-Learn on Arbitrary Textual Conditioning: A Hypernetwork-Driven Meta-Gated LLM
A hypernetwork produces a condition-dependent beta that meta-gates SwiGLU nonlinearity, giving LLMs adaptive behavior across task, domain, persona and style inputs without finetuning.
-
GIFT: Guided Fine-Tuning and Transfer for Enhancing Instruction-Tuned Language Models
GIFT guides adapter fine-tuning on base models with confidence signals from instruction-tuned models before merging, yielding task-specialized models that outperform direct fine-tuning on math and knowledge benchmarks.
-
GiVA: Gradient-Informed Bases for Vector-Based Adaptation
GiVA uses gradients to initialize vector adapters so they match LoRA performance at eight times lower rank while keeping extreme parameter efficiency.
-
Understanding Secret Leakage Risks in Code LLMs: A Tokenization Perspective
BPE tokenization creates gibberish bias in CLLMs, causing secrets with high character entropy but low token entropy to be preferentially memorized due to training data distribution shifts.
-
Towards General Text Embeddings with Multi-stage Contrastive Learning
GTE_base is a compact text embedding model using multi-stage contrastive learning on diverse data that outperforms OpenAI's API and 10x larger models on massive benchmarks and works for code as text.
-
PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding
PipeSD is a cloud-edge collaborative inference framework that overlaps token generation and communication via dynamic programming pipeline scheduling and uses Bayesian-optimized dual-threshold NAV triggering, delivering 1.16x-2.16x speedup and 14.3%-25.3% energy reduction over baselines.
-
Benchmarked Yet Not Measured -- Generative AI Should be Evaluated Against Real-World Utility
Generative AI evaluation must shift from static benchmark scores to measuring sustained improvements in human capabilities within specific deployment contexts.