Sumi is an openly released 7B parameter uniform diffusion language model pretrained from scratch on 1.5T tokens that matches autoregressive models on several benchmarks.
super hub Mixed citations
gpt-oss-120b & gpt-oss-20b Model Card
Mixed citation behavior. Most common role is background (41%).
433
Pith papers citing it
Background
41% of classified citations
abstract
We present gpt-oss-120b and gpt-oss-20b, two open-weight reasoning models that push the frontier of accuracy and inference cost. The models use an efficient mixture-of-expert transformer architecture and are trained using large-scale distillation and reinforcement learning. We optimize the models to have strong agentic capabilities (deep research browsing, python tool use, and support for developer-provided functions), all while using a rendered chat format that enables clear instruction following and role delineation. Both models achieve strong results on benchmarks ranging from mathematics, coding, and safety. We release the model weights, inference implementations, tool environments, and tokenizers under an Apache 2.0 license to enable broad use and further research.
hub tools
citation-role summary
background 33
baseline 16
method 16
other 6
dataset 5
citation-polarity summary
claims ledger
- abstract We present gpt-oss-120b and gpt-oss-20b, two open-weight reasoning models that push the frontier of accuracy and inference cost. The models use an efficient mixture-of-expert transformer architecture and are trained using large-scale distillation and reinforcement learning. We optimize the models to have strong agentic capabilities (deep research browsing, python tool use, and support for developer-provided functions), all while using a rendered chat format that enables clear instruction following and role delineation. Both models achieve strong results on benchmarks ranging from mathematics,
authors
co-cited works
representative citing papers
TW-LegalBench evaluates 13 LLMs on over 30,000 Taiwanese legal tasks from exams and judgments, showing top models pass lawyer thresholds but struggle with exact statute citations.
UltraEP is the first exact-load real-time expert balancer for large-EP MoE training and serving on rack-scale nodes, reaching 94.3% of ideal throughput and 1.49x over no-balancing.
RobotValues is a benchmark of 10K value-conflict scenarios that reveals VLMs default to safety and accommodation while failing to follow instructions to prioritize other values 80% of the time.
Presents the first fully open pipeline for clinical LLMs by unifying eight public QA datasets with three clinician-vetted synthetic extensions and applying it to five base models to achieve benchmark gains while maintaining auditability.
MathAtlas is the first large-scale benchmark for autoformalizing graduate mathematics, where even strong models reach only 9.8% correctness on theorem statements and drop to 2.6% on the hardest dependency-deep subset.
LLMs lack temporal awareness of medical knowledge, showing gradual performance decline on up-to-date facts, much lower accuracy on historical knowledge (25-54% relative), and inconsistent year-to-year predictions.
Sieve dynamically schedules MoE experts across GPU and PIM hardware to handle bimodal token distributions, achieving 1.3x to 1.6x gains in throughput and interactivity over static prior PIM systems on three large models.
Soohak is a 439-problem mathematician-curated benchmark where frontier LLMs reach at most 30.4% on research math challenges and no model exceeds 50% on refusal for ill-posed problems.
MathConstraint generates scalable, automatically verifiable combinatorial problems where LLMs achieve 18.5-66.9% accuracy without tools but roughly double that with solver access.
IRIS-14B is the first LLM trained explicitly for GIMPLE-to-LLVM IR translation and outperforms much larger models by up to 44 percentage points on real-world C code.
RoundPipe achieves near-zero-bubble pipeline parallelism for LLM training on consumer GPUs by dynamically dispatching computation stages round-robin, yielding 1.48-2.16x speedups and enabling 235B model fine-tuning on 8x RTX 4090.
InfiniteScienceGym procedurally generates unbounded scientific repositories with exact ground-truth QA pairs to benchmark LLMs on data reasoning, abstention, and tool use without static datasets.
Large language models display the identifiable victim effect at roughly twice the human baseline, strongly amplified by instruction tuning and chain-of-thought prompting but inverted by reasoning-specialized models.
Tessera performs kernel-granularity disaggregation on heterogeneous GPUs, achieving up to 2.3x throughput and 1.6x cost efficiency gains for large model inference while generalizing beyond prior methods.
The SDE benchmark shows LLMs lag on scientific discovery tasks relative to general science tests, with diminishing scaling returns and shared weaknesses across models.
SpeechCombine produces instruction-following SLMs via speech pre-training followed by direct weight combination with the text LLM instruction delta, without any speech instruction tuning.
OpenSafeIntent benchmark shows models fail to calibrate safety across intent shifts in matched dual-use prompts, indicating current evaluations are insufficient.
A 0.6B LM with length-aware attention adjustments performs competitive in-context retrieval at million-token scale on MS MARCO, NQ, and LIMIT benchmarks.
LLM-generated research ideas cluster more around bridge-like opportunities and synthesis methods than the broader distribution seen in human papers.
ELDR reduces median TPOT by 5.9-13.9% in PD-disaggregated MoE serving via expert signatures from prefill, K-means partitioning, and locality-band routing with KV-co-indexed signature cache.
Introduces GenAI agent framework for auditing personalization algorithms via synthetic accounts with fixed personas, applied to X post-2024 election showing amplification of toxic and right-leaning content varying by ideology.
SABER-Math is an automated benchmark for mathematical IR that uses LLM summaries, topic similarities, and preference tournaments on 283K problems to create reranking tasks, showing embedding models outperform baselines but struggle in symbol-heavy areas and that MTEB does not predict math performanc
LLM agents often fail to abstain at the right time in uncertain multi-turn tasks, and the CONVOLVE context engineering method raises timely abstention rates on WebShop from 26.7 to 57.4 without parameter updates.
citing papers explorer
-
veScale-FSDP: Flexible and High-Performance FSDP at Scale
veScale-FSDP uses RaggedShard and structure-aware planning to support block-wise quantization and non-element-wise optimizers while delivering 5-66% higher throughput and 16-30% lower memory than prior FSDP systems at massive scale.
-
Disentangling Ambiguity from Instability in Large Language Models: A Clinical Text-to-SQL Case Study
CLUES decomposes semantic uncertainty into separate ambiguity and instability scores for clinical Text-to-SQL, with instability via Schur complement, outperforming Kernel Language Entropy on failure prediction while enabling diagnostic triage.
-
When LLMs get significantly worse: A statistical approach to detect model degradations
A McNemar-based statistical test detects real degradations in optimized LLMs with controlled false positives, even for accuracy changes as small as 0.3%.
-
Scalable Generation and Validation of Isomorphic Physics Problems with GenAI
GenAI framework generates isomorphic physics problem banks via prompt chaining and validates them with 17 language models that correlate with student performance (ρ up to 0.594), achieving homogeneous difficulty in 73% of deployed banks.
-
Temporal Leakage in Search-Engine Date-Filtered Web Retrieval: A Retrospective Forecasting Case Study
Date filters on major search engines frequently leak post-cutoff information, inflating Brier scores in retrospective forecasting from 0.24 to 0.10.
-
Compounding Disadvantage: Auditing Intersectional Bias in LLM-Generated Explanations Across Indian and American STEM Education
LLMs generate lower-quality STEM explanations for marginalized student profiles in Indian and American contexts, with intersectional compounding producing gaps of up to 2.55 grade levels.
-
MirrorBench: A Benchmark to Evaluate Conversational User-Proxy Agents for Human-Likeness
MirrorBench defines a reproducible benchmark combining lexical metrics (MATTR, Yule's K, HD-D) and LLM-judge metrics with calibration controls to measure human-likeness of user-proxy agents across four datasets.
-
ALL-FEM: Agentic Large Language models Fine-tuned for Finite Element Methods
ALL-FEM fine-tunes LLMs on a corpus of verified FEniCS scripts and uses multi-agent workflows to automate finite element code generation, achieving 71.79% success on 39 benchmarks across elasticity, flow, and coupled problems.
-
Understanding Structured Financial Data with LLMs: A Case Study on Fraud Detection
FinFRE-RAG combines importance-guided feature reduction with label-aware retrieval-augmented generation to boost LLM performance on tabular fraud detection across four public datasets while providing human-readable rationales.
-
Asynchronous Reasoning: Training-Free Interactive Thinking LLMs
Using properties of positional embeddings, reasoning LLMs can be made to think, listen, and generate outputs asynchronously without any additional training, cutting time to first token to under 5 seconds.
-
SQuARE: Structured Query & Adaptive Retrieval Engine For Tabular Formats
SQuARE is a hybrid retrieval system that uses a complexity score to route tabular queries between chunk-based and SQL-based paths, outperforming single-strategy baselines and GPT-4o on precision and accuracy for complex spreadsheets.
-
MermaidSeqBench: An Evaluation Benchmark for NL-to-Mermaid Sequence Diagram Generation
MermaidSeqBench is a new human-verified benchmark for evaluating LLMs on natural language to Mermaid sequence diagram generation, revealing significant capability gaps across models.
-
SnapAudit: Active Auditing of Differentially Private In-Context Learning via Snapshot-Based Simulation
SnapAudit decomposes DP-ICL into a deterministic snapshot stage and a stochastic noise stage, using bootstrap simulation to achieve 80-200x faster auditing and exposing privacy bound violations in existing Gaussian and embedding mechanisms.
-
MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling
MiroThinker shows that scaling agent-environment interactions via reinforcement learning lets a 72B open-source model reach up to 81.9% on GAIA and approach commercial performance on research benchmarks.
-
Graph-Based Alternatives to LLMs for Human Simulation
GEMS formulates close-ended human-behavior simulation as link prediction on a heterogeneous graph and matches or exceeds LLM performance with three orders of magnitude fewer parameters across three datasets and three evaluation settings.
-
Making Interpretable Discoveries from Unstructured Data: A High-Dimensional Multiple Hypothesis Testing Approach
A new framework combines AI-derived concept embeddings with high-dimensional selective inference to enable statistically principled, interpretable discovery from unstructured data in empirical economics.
-
Kimi Linear: An Expressive, Efficient Attention Architecture
Kimi Linear hybridizes linear attention with a new KDA module to beat full attention on tasks while slashing KV cache by 75% and speeding decoding up to 6x.
-
Don't Pass@k: A Bayesian Framework for Large Language Model Evaluation
A Dirichlet-prior Bayesian estimator for model success probability replaces Pass@k, delivering faster-converging and more stable rankings with credible intervals on math benchmarks.
-
Beyond Linear Probes: Dynamic Safety Monitoring for Language Models
TPCs allow term-by-term progressive polynomial evaluation on LLM activations for flexible safety monitoring that supports both stronger guardrails and low-cost adaptive cascades.
-
Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post Training
Post-training on reasoning tasks sparks the emergence of specialized attention heads that enable structured computation, with SFT adding stable heads while GRPO uses dynamic activation and pruning tied to reward signals, and controllable think models relying on compensatory heads instead of specific
-
Short window attention enables long-term memorization
Short sliding windows in hybrid attention-xLSTM models boost long-context performance by encouraging long-term memory use, and stochastic window sizing improves both short and long tasks.
-
ChatInject: Abusing Chat Templates for Prompt Injection in LLM Agents
ChatInject exploits LLM chat template structures to boost indirect prompt injection success rates on agents from ~5-15% to 32-52% across benchmarks, with multi-turn persuasion variants performing best.
-
Can Large Language Models Really Recognize Your Name?
LLMs exhibit 20-40% lower recall on ambiguous human names for PII detection, worsening under prompt injections, as shown via the new AmBench benchmark.
-
"Don't Say It!": Constraints, Compliance, and Communication when Language Models Play Taboo
LLMs exhibit different trade-offs between rule compliance and communicative success across prompting, generation constraints, and representation interventions, but remain substantially weaker than humans at guessing under lexical constraints.
-
NeuroCogMap Reveals Cognitive Organization of Large Language Models
NeuroCogMap maps LLM internal representations into stable functional parcels tied to cognitive functions, failure modes, and human cortical activity during language tasks.
-
Revising RVL-CDIP: Quantifying Errors and Test-Train Overlap
RVL-CDIP contains 12% label errors and 35% train-test duplicates; correcting labels improves OOD generalization while deduplication reduces in-distribution accuracy.
-
EntroRouter: Learning Efficient Model Routing via Entropy Regulation
EntroRouter applies entropy regulation in a single-round routing framework to decouple reasoning from routing, retaining 98.3% of top expert accuracy at 48.25% lower compute cost.
-
ToE: A Hierarchical and Explainable Claim Verification Framework with Dynamic Multi-source Evidence Retrieval and Aggregation
ToE is a hierarchical claim verification framework using RL-driven multi-source retrieval, evidence evaluation, and tree aggregation that reports 4-24 point gains over baselines especially on poisoned inputs.
-
Listening Like a Judge: A Music-Aware Framework for Automatic Singing Performance Evaluation
MusicJudge is a modality-guided framework that performs block-aligned multimodal analysis for singing quality assessment by coupling lyrics with pitch-rhythm fidelity via multi-signal matching and Modality-Guided LoRA fine-tuning.
-
Latent Confidence Alignment for LLM Self-Assessment
LCAE is introduced as a Rasch-model metric that aligns LLM self-reported confidence with latent error probability derived from ability and item difficulty, shown to improve calibration on a medical dataset across 20 models.
-
Qiskit Code Migration with LLMs
A taxonomy-guided RAG system with LLMs reduces hallucinations and improves migration suggestions for Qiskit code compared to unconstrained retrieval.
-
Conservation Laws for Modern Neural Architectures
Unified framework characterizes conservation laws for gradient flow in feedforward networks with GELU/SiLU/SwiGLU, multihead attention with positional encodings, and MoE models under various gating.
-
SpecAlign: Efficient Specification-Grounded Alignment of Large Language Models via Synthetic Data
SpecAlign synthesizes boundary-aware preference pairs directly from structured model specifications to train LLMs for improved rule compliance.
-
Know Your Limits : On the Faithfulness of LLMs as Solvers and Autoformalizers in Legal Reasoning
Formal structure boosts LLM accuracy on legal entailment but does not produce faithful reasoning, with scope laundering and other failures persisting across models on ContractNLI.
-
Reward Modeling for Multi-Agent Orchestration
OrchRM uses intermediate artifacts from multi-agent runs to create training pairs for a reward model that guides orchestrator training and test-time scaling, reporting up to 10x token efficiency and 8% accuracy gains across reasoning domains.
-
Constrained Semantic Decompression in LLMs through Persian Proverb-Conditioned Story Generation
Introduces PAND dataset for Persian proverbs and reports a persistent decompression gap in LLMs that explicit reasoning partially reduces.
-
Lung-R1: A Knowledge Graph-Guided LLM for Pulmonary Diagnostic Reasoning
Introduces the first structured pulmonary knowledge graph LungKG and uses it to train Lung-R1, which reaches SOTA on EMR-based pulmonary diagnosis tasks.
-
It Takes One to Bias Them All: Breaking Bad with One-Shot GRPO
One-shot GRPO on a single biased example induces generalizing stereotype bias in post-trained LLMs, with susceptibility varying by initial bias likelihood.
-
PBSD: Privileged Bayesian Self-Distillation for Long-Horizon Credit Assignment
PBSD derives autoregressive turn-level credit signals from outcome rewards via the posterior-to-prior ratio converted through Bayes' rule between student and privileged teacher models.
-
Overcoming the Regulatory Bottleneck via Agent-to-Agent Protocols: A Nuclear Case Study
RCP is a new agent-to-agent protocol that reduces nuclear regulatory review costs by 50-77% (to 21-44M USD) and timelines by 65% (to 15 months) versus an 89M USD, 42-month reconstructed baseline.
-
Better Literary Translation: A Multi-Aspect Data Generation and LLM Training Approach
Multi-aspect iterative refinement with specialized LLMs generates superior literary translation data, enabling SFT and GRPO to produce LitMT-8B and LitMT-14B models scoring 67.25 and 69.07 CEA100 on MetaphorTrans, competitive with Claude Sonnet 4.5.
-
Staying with the Uncertainty: Uncertainty-Scaffolding Strategies for Artificial Moral Advisors in LLM-to-LLM Simulated Conversations
LLM-simulated dialogues show uncertainty-scaffolding strategies sustain higher-quality engagement than controls without producing more stance revision.
-
Optimizing the Cost-Quality Tradeoff of Agentic Theorem Provers in Lean
An agentic theorem prover in Lean uses a control plane to route actions based on cost and success estimates, achieving 28.9% lower average cost than a fixed-step baseline on a PutnamBench subset while preserving performance.
-
E2LLM: Towards Efficient LLM Serving in Heterogeneous Edge/Fog Environments
E2LLM replicates LLMs across role-specialized device clusters in heterogeneous edge/fog environments, optimizes clusters with genetic algorithms and intra-cluster partitioning with dynamic programming, and reports over 50% reduction in average waiting time versus Splitwise under high demand.
-
SCOPE: Real-Time Natural Language Camera Agent at the Edge
SCOPE introduces an edge-deployable natural-language PTZ camera agent, a simulation benchmark, and evaluations showing that stronger small language models reduce hallucinations while perception remains the main bottleneck.
-
Transferable Self-Harm Surveillance from Emergency Department Triage Notes Using an Evidence-Augmented Machine Learning Approach
Three-stage evidence-augmented ML model detects self-harm in ED triage notes at AUPRC ~0.88, transfers across sites without retraining, and identifies primary method at 95% accuracy.
-
AgentCL: Toward Rigorous Evaluation of Continual Learning in Language Agents
AgentCL constructs controlled task streams with intentional reusability and introduces MemProbe to evaluate non-parametric memory designs for continual learning in language agents across coding, research, and reasoning tasks.
-
Qiskit QuantumKatas: Adapting Microsoft's Quantum Computing exercises for LLM evaluation
Adapts QuantumKatas to Qiskit yielding a 350-task benchmark across 26 categories and evaluates 16 LLMs in 39,200 runs, reporting performance gaps and prompting effects.
-
Inference Time Context Sparsity: Illusion or Opportunity?
Current LLMs remain robust to high levels of inference-time context sparsity across diverse tasks, enabling up to 10x acceleration via sparse kernels.
-
GradeLegal: Automated Grading for German Legal Cases
Reasoning-oriented LLMs reach up to 0.91 quadratic weighted kappa agreement with experts on public law cases when given sample solutions and grading rubrics, but only 0.60 on criminal law cases.