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.
hub Mixed citations
GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models
Mixed citation behavior. Most common role is background (69%).
91
Pith papers citing it
Background
69% of classified citations
abstract
We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes. Through multi-stage training on 23T tokens and comprehensive post-training with expert model iteration and reinforcement learning, GLM-4.5 achieves strong performance across agentic, reasoning, and coding (ARC) tasks, scoring 70.1% on TAU-Bench, 91.0% on AIME 24, and 64.2% on SWE-bench Verified. With much fewer parameters than several competitors, GLM-4.5 ranks 3rd overall among all evaluated models and 2nd on agentic benchmarks. We release both GLM-4.5 (355B parameters) and a compact version, GLM-4.5-Air (106B parameters), to advance research in reasoning and agentic AI systems. Code, models, and more information are available at https://github.com/zai-org/GLM-4.5.
hub tools
citation-role summary
background 19
baseline 3
method 3
other 1
citation-polarity summary
claims ledger
- abstract We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes. Through multi-stage training on 23T tokens and comprehensive post-training with expert model iteration and reinforcement learning, GLM-4.5 achieves strong performance across agentic, reasoning, and coding (ARC) tasks, scoring 70.1% on TAU-Bench, 91.0% on AIME 24, and 64.2% on SWE-bench Verified. With much fewer parameters than several competitors, GLM-4.5 ranks 3rd ove
co-cited works
representative citing papers
ReLibra uses pre-known token-to-expert routing from RL rollouts to perform inter-batch expert reordering and intra-batch replication, delivering up to 1.6x higher throughput than Megatron-LM and 1.2x over oracle-equipped EPLB while staying within 6-10% of an ideal balanced baseline.
WildTableBench is the first QA benchmark for naturally occurring table images, where 21 multimodal models were evaluated and only one exceeded 50% accuracy.
User-turn generation reveals that LLMs' interaction awareness is largely decoupled from task accuracy, remaining near zero in deterministic settings even as accuracy scales to 96.8% on GSM8K.
LatentOmni proposes a latent-space cross-modal reasoning framework that uses feature-level supervision and Omni-Sync Position Embedding to align and synchronize audio-visual latents, supported by a new 35K interleaved reasoning dataset and showing gains over text CoT baselines.
CopT reverses CoT by eliciting a draft answer first then using continuous-embedding contrastive verification and on-policy thinking to reflect and correct, yielding up to 23% higher accuracy and 57% fewer tokens without training.
PRISM benchmark of over 10k pairs shows LLMs have a 41% average drop from code execution success to spatial correctness in programmatic video generation.
PARAMΔ upcycles dense models to MoE for per-language experts and grafts post-training deltas to enable data-efficient language expansion while preserving original capabilities.
A genome-conditioned 4B LLM agent predicts microbial life boundaries and matches larger frontier models via token fusion, tool use, and a counterfactual gene-grounding reward.
StoryReward, trained on a new 100k story preference dataset, sets state-of-the-art performance on the introduced StoryRMB benchmark for aligning LLM stories with human preferences.
Hosted open-weight LLM APIs function as time-varying heterogeneous services rather than fixed model artifacts, with demand concentrated, supply-use mismatches, and task-specific routing yielding major cost and throughput gains.
AgentVisor cuts prompt injection success rate to 0.65% in LLM agents with only 1.45% utility loss via semantic privilege separation and one-shot self-correction.
Dr.Sai autonomously executed full physics analysis pipelines on real BESIII data to re-measure ten J/psi decay branching fractions, matching established benchmarks without any manual coding.
SportsTime benchmark and CoTR method improve multimodal AI's temporal compositional reasoning and evidence grounding in long-form sports videos.
FEPLB reduces token and GEMM stragglers in MoE training by 50-70% using nearly free Copy Engine communication on Hopper architecture.
Expert upcycling duplicates experts in an existing MoE checkpoint and continues pre-training to match fixed-size baseline performance with 32% less compute.
A single query-specific poisoned document, built by extracting and iteratively refining an adversarial chain-of-thought, can substantially degrade reasoning accuracy in retrieval-augmented LLM systems.
E2E-REME outperforms nine LLMs in accuracy and efficiency for end-to-end microservice remediation by using experience-simulation reinforcement fine-tuning on a new benchmark called MicroRemed.
ImplicitMemBench shows no LLM exceeds 66% on implicit memory tasks, with top models at 65%, far below humans and pointing to architectural limits beyond scaling.
Bridge-STG decouples spatio-temporal alignment via semantic bridging and query-guided localization modules to achieve state-of-the-art m_vIoU of 34.3 on VidSTG among MLLM methods.
Autopoiesis uses LLM-driven program synthesis to evolve serving policies online during deployment, delivering up to 53% and average 34% gains over prior LLM serving systems under runtime dynamics.
OmniCompliance-100K supplies 12,985 distinct rules and 106,009 associated real-world cases from 74 multi-domain regulations to benchmark LLM safety and compliance.
SPIRAL is a closed-loop think-act-reflect framework using PlanAgent, VideoGenerator, and CriticAgent plus GRPO self-evolution to improve long-horizon action-conditioned video generation, with new dataset and benchmark showing gains over open-loop baselines.
EvoESAP uses evolutionary search guided by a speculative-decoding-inspired ESAP metric to discover non-uniform layer-wise sparsity allocations for MoE expert pruning, improving generation accuracy up to 19.6% at 50% sparsity.
citing papers explorer
-
Sieve: Dynamic Expert-Aware PIM Acceleration for Evolving Mixture-of-Experts Models
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.
-
ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning
ReLibra uses pre-known token-to-expert routing from RL rollouts to perform inter-batch expert reordering and intra-batch replication, delivering up to 1.6x higher throughput than Megatron-LM and 1.2x over oracle-equipped EPLB while staying within 6-10% of an ideal balanced baseline.
-
WildTableBench: Benchmarking Multimodal Foundation Models on Table Understanding In the Wild
WildTableBench is the first QA benchmark for naturally occurring table images, where 21 multimodal models were evaluated and only one exceeded 50% accuracy.
-
Beyond the Assistant Turn: User Turn Generation as a Probe of Interaction Awareness in Language Models
User-turn generation reveals that LLMs' interaction awareness is largely decoupled from task accuracy, remaining near zero in deterministic settings even as accuracy scales to 96.8% on GSM8K.
-
LatentOmni: Rethinking Omni-Modal Understanding via Unified Audio-Visual Latent Reasoning
LatentOmni proposes a latent-space cross-modal reasoning framework that uses feature-level supervision and Omni-Sync Position Embedding to align and synchronize audio-visual latents, supported by a new 35K interleaved reasoning dataset and showing gains over text CoT baselines.
-
CopT: Contrastive On-Policy Thinking with Continuous Spaces for General and Agentic Reasoning
CopT reverses CoT by eliciting a draft answer first then using continuous-embedding contrastive verification and on-policy thinking to reflect and correct, yielding up to 23% higher accuracy and 57% fewer tokens without training.
-
PRISM: A Benchmark for Programmatic Spatial-Temporal Reasoning
PRISM benchmark of over 10k pairs shows LLMs have a 41% average drop from code execution success to spatial correctness in programmatic video generation.
-
A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$\Delta$ Integration into Upcycled MoE
PARAMΔ upcycles dense models to MoE for per-language experts and grafts post-training deltas to enable data-efficient language expansion while preserving original capabilities.
-
GGBound: A Genome-Grounded Agent for Microbial Life-Boundary Prediction
A genome-conditioned 4B LLM agent predicts microbial life boundaries and matches larger frontier models via token fusion, tool use, and a counterfactual gene-grounding reward.
-
StoryAlign: Evaluating and Training Reward Models for Story Generation
StoryReward, trained on a new 100k story preference dataset, sets state-of-the-art performance on the introduced StoryRMB benchmark for aligning LLM stories with human preferences.
-
When Is the Same Model Not the Same Service? A Measurement Study of Hosted Open-Weight LLM APIs
Hosted open-weight LLM APIs function as time-varying heterogeneous services rather than fixed model artifacts, with demand concentrated, supply-use mismatches, and task-specific routing yielding major cost and throughput gains.
-
AgentVisor: Defending LLM Agents Against Prompt Injection via Semantic Virtualization
AgentVisor cuts prompt injection success rate to 0.65% in LLM agents with only 1.45% utility loss via semantic privilege separation and one-shot self-correction.
-
Dr.Sai: An agentic AI for real-world physics analysis at BESIII
Dr.Sai autonomously executed full physics analysis pipelines on real BESIII data to re-measure ten J/psi decay branching fractions, matching established benchmarks without any manual coding.
-
Towards Temporal Compositional Reasoning in Long-Form Sports Videos
SportsTime benchmark and CoTR method improve multimodal AI's temporal compositional reasoning and evidence grounding in long-form sports videos.
-
FEPLB: Exploiting Copy Engines for Nearly Free MoE Load Balancing in Distributed Training
FEPLB reduces token and GEMM stragglers in MoE training by 50-70% using nearly free Copy Engine communication on Hopper architecture.
-
Expert Upcycling: Shifting the Compute-Efficient Frontier of Mixture-of-Experts
Expert upcycling duplicates experts in an existing MoE checkpoint and continues pre-training to match fixed-size baseline performance with 32% less compute.
-
AdversarialCoT: Single-Document Retrieval Poisoning for LLM Reasoning
A single query-specific poisoned document, built by extracting and iteratively refining an adversarial chain-of-thought, can substantially degrade reasoning accuracy in retrieval-augmented LLM systems.
-
E2E-REME: Towards End-to-End Microservices Auto-Remediation via Experience-Simulation Reinforcement Fine-Tuning
E2E-REME outperforms nine LLMs in accuracy and efficiency for end-to-end microservice remediation by using experience-simulation reinforcement fine-tuning on a new benchmark called MicroRemed.
-
ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models
ImplicitMemBench shows no LLM exceeds 66% on implicit memory tasks, with top models at 65%, far below humans and pointing to architectural limits beyond scaling.
-
Bridging Time and Space: Decoupled Spatio-Temporal Alignment for Video Grounding
Bridge-STG decouples spatio-temporal alignment via semantic bridging and query-guided localization modules to achieve state-of-the-art m_vIoU of 34.3 on VidSTG among MLLM methods.
-
Autopoiesis: A Self-Evolving System Paradigm for LLM Serving Under Runtime Dynamics
Autopoiesis uses LLM-driven program synthesis to evolve serving policies online during deployment, delivering up to 53% and average 34% gains over prior LLM serving systems under runtime dynamics.
-
OmniCompliance-100K: A Multi-Domain, Rule-Grounded, Real-World Safety Compliance Dataset
OmniCompliance-100K supplies 12,985 distinct rules and 106,009 associated real-world cases from 74 multi-domain regulations to benchmark LLM safety and compliance.
-
SPIRAL: Self-Evolving Action-Conditioned Video Generation via Reflective Planning Agents
SPIRAL is a closed-loop think-act-reflect framework using PlanAgent, VideoGenerator, and CriticAgent plus GRPO self-evolution to improve long-horizon action-conditioned video generation, with new dataset and benchmark showing gains over open-loop baselines.
-
EvoESAP: Non-Uniform Expert Pruning for Sparse MoE
EvoESAP uses evolutionary search guided by a speculative-decoding-inspired ESAP metric to discover non-uniform layer-wise sparsity allocations for MoE expert pruning, improving generation accuracy up to 19.6% at 50% sparsity.
-
SQuTR: A Robustness Benchmark for Spoken Query to Text Retrieval under Acoustic Noise
SQuTR aggregates 37k queries from six text retrieval datasets, synthesizes speech from 200 speakers, adds 17 noise categories at varying SNR, and shows that even large retrieval models degrade sharply under extreme acoustic noise.
-
A Benchmark for Evaluating Outcome-Driven Constraint Violations in Autonomous AI Agents
A new benchmark of 40 scenarios finds state-of-the-art LLMs exhibit outcome-driven constraint violations in 0-62.8% of cases under KPI pressure, with no consistent safety gains across model generations.
-
SWE-EVO: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarios
SWE-EVO shows GPT-5.4 with OpenHands reaching only 25% success on complex multi-file evolution tasks versus 72.8% on SWE-Bench Verified, and introduces Fix Rate as a partial-progress metric.
-
Dynamic Tool Dependency Retrieval for Lightweight Function Calling
DTDR dynamically retrieves relevant tools by modeling dependencies from demonstrations and conditioning on the evolving agent plan, improving function calling success rates by 23-104% over static retrievers across benchmarks.
-
MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning
MemSearcher trains LLMs to manage compact memory in multi-turn searches via multi-context GRPO for end-to-end RL, outperforming ReAct-style baselines with stable token counts.
-
SecureWebArena: A Holistic Security Evaluation Benchmark for LVLM-based Web Agents
SecureWebArena is a new benchmark suite for holistic security evaluation of LVLM-based web agents using diverse simulated environments, attack taxonomies, and multi-layered failure analysis across reasoning, behavior, and outcomes.
-
Talking Trees: Reasoning-Assisted Induction of Decision Trees for Tabular Data
Reasoning LLMs with minimal tools for tree construction and analysis induce decision trees that outperform CART, compete with ensembles on low-resource tabular data, and provide human-readable reasoning traces.
-
Maestro: Reinforcement Learning to Orchestrate Hierarchical Model-Skill Ensembles
Maestro uses outcome-based RL to train a lightweight policy that orchestrates ensembles of frozen expert models and skills, reporting 70.1% average accuracy across ten multimodal benchmarks and outperforming GPT-5 and Gemini-2.5-Pro while generalizing to unseen components.
-
Memory Grafting: Scaling Language Model Pre-training via Offline Conditional Memory
Memory Grafting improves language-model benchmarks by grafting offline hidden-state memory from a larger model into a recipient model using n-gram lookups and lightweight adapters, outperforming MoE and vanilla Engram baselines at 0.92B and 2.8B scales.
-
Post-Trained MoE Can Skip Half Experts via Self-Distillation
ZEDA injects zero-output experts and uses two-stage self-distillation to adapt post-trained MoE models into dynamic ones that skip over half the experts, yielding 1.2x inference speedup with small accuracy drops.
-
FinDocMRE: A Benchmark for Document-Level Financial Multimodal Reasoning Evaluation
FinDocMRE is a new multi-image document-level benchmark spanning 12 financial domains and 5 task types, showing that 11 tested LMMs all score below 65 overall with particular weaknesses in numerical estimation and cross-page grounding.
-
Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models
PUMA detects reasoning-level semantic redundancy to enable early exit in chains of thought, achieving 26.2% average token reduction across five LRMs and five benchmarks while preserving accuracy and CoT quality.
-
Artificial Intolerance: Stigmatizing Language in Clinical Documentation Skews Large Language Model Decision-Making
Frontier LLMs exhibit bias from stigmatizing language in clinical vignettes across four conditions, skewing decisions toward less aggressive management, with limited mitigation from Chain-of-Thought or self-debiasing prompts.
-
Code-Centric Detection of Vulnerability-Fixing Commits: A Unified Benchmark and Empirical Study
Code language models show no transferable security understanding from code diffs alone, rely on commit messages, miss over 93% of fixes at 0.5% false positive rate, and suffer large drops under group or temporal splits.
-
ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox
ComplexMCP benchmark shows top LLM agents achieve under 60% success on dynamic interdependent tool tasks versus 90% for humans, due to tool retrieval saturation, over-confidence, and strategic defeatism.
-
Edit-Based Refinement for Parallel Masked Diffusion Language Models
ME-DLM augments parallel masked diffusion models with edit-distance-supervised refinements to raise quality on coding and math benchmarks while using far fewer diffusion steps.
-
ORACLE: Anticipating Scams from Partial Trajectories in Streaming App Usage
ORACLE is a new agentic framework using adaptive context consolidation and teacher-student distillation to detect emerging scam patterns from incomplete, long-horizon app usage streams across 12 scam types.
-
Self-ReSET: Learning to Self-Recover from Unsafe Reasoning Trajectories
Self-ReSET is a reinforcement learning approach that lets large reasoning models learn to recover from their own unsafe reasoning trajectories, improving robustness to adversarial jailbreaks while preserving utility.
-
VeriContest: A Competitive-Programming Benchmark for Verifiable Code Generation
VeriContest supplies 946 problems with specs, code, proofs, and tests to benchmark verifiable code generation in Rust/Verus, showing models reach 92% on code but only 5% end-to-end on full verifiable synthesis.
-
WebTrap: Stealthy Mid-Task Hijacking of Browser Agents During Navigation
WebTrap uses multi-step instruction fusion and context-grounded generation to stealthily hijack browser agents mid-navigation while preserving original task success.
-
HTPO: Towards Exploration-Exploitation Balanced Policy Optimization via Hierarchical Token-level Objective Control
HTPO introduces hierarchical token-level objective control in RLVR to balance exploration and exploitation by grouping tokens according to difficulty, correctness, and entropy, yielding up to 8.6% gains on AIME benchmarks over DAPO.
-
Hard to Read, Easy to Jailbreak: How Visual Degradation Bypasses MLLM Safety Alignment
Degraded image resolution in MLLMs bypasses safety alignments via cognitive overload, raising jailbreak rates across perturbations.
-
Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less
Full finetuning with the pretraining optimizer reduces forgetting compared to other optimizers or LoRA while achieving comparable new-task performance.
-
Teaching Thinking Models to Reason with Tools: A Full-Pipeline Recipe for Tool-Integrated Reasoning
A training recipe for tool-integrated reasoning models achieves state-of-the-art open-source results on math benchmarks such as 96.7% and 99.2% on AIME 2025 at 4B and 30B scales by balancing tool-use trajectories and optimizing for pass@k during SFT before stable RLVR.
-
AlbumFill: Album-Guided Reasoning and Retrieval for Personalized Image Completion
AlbumFill retrieves identity-consistent references from personal albums via VLM-inferred semantic cues to support personalized image completion.
-
AMMA: A Multi-Chiplet Memory-Centric Architecture for Low-Latency 1M Context Attention Serving
AMMA is a memory-centric multi-chiplet architecture using HBM-PNM cubes, custom logic dies, hybrid parallelism, and reordered collectives that delivers 15.5X lower attention latency and 6.9X lower energy than NVIDIA H100 for 1M context serving.