全部收录论文
论文 · 第 38 页
共 7039 篇 · 按日期新到旧 · 每页 50 篇 · 第 38 / 141 页
查找论文请用站内搜索。
- Evidence Integration in Large Language Models
- Search2Skill: Skill Distillation Beyond Knowledge Boundaries Via Rubric-Based Reinforcement Learning
- PROGRESS: Coverage-guided RL to Train Search-augmented LLM Agent
- Cross-Domain Hybrid OPD for Generalizable Search Agents
- GRACE: Graph-Grounded Reflective Agent Copilot Engine for Expert-in-the-Loop Knowledge Expansion
- Advanced CUDA Graph Techniques in SGLang
- Fast Engine Recovery: Sub-Second Engine Restart for SGLang via Weight Cache Daemon
- HPC-Ops × SGLang: High-Performance Attention, Router GEMM, and MoE Kernels from Tencent Hunyuan
- Atlas: Optimizing Deployment of Compound AI Workflows on Heterogeneous Clusters
- Same Request, Different Answer: Quantization Amplifies Cache-Induced Divergence in LLM Serving
- Scale-QLoRA: Code-Invariant Adapter Merging for Native 4-bit Microscaling LLMs
- Unified Radix Cache: One Tree for Hybrid Model Prefix Caching
- CIERA: Cross-Iteration Exponent Reuse for Lossless Allgather in Sharded MoE Training
- Tuning Collective Patterns to Alleviate Congestion in Shared AI Clusters
- What Attention Recalls and Recurrence Controls in Hybrid Language Models
- LLM4AIGQ: LLM-based AI Guidance Query Generation Framework for Multi Interest Mining
- EPIC: Explicit Posterior Item Conditioning for Semantic ID Diffusion Recommendation
- SelfDR: Self-Distillation from Reasoning for LLM-Based Recommendation
- HypRQ-VAE: Hyperbolic Item Indexing for Long-Tail-Aware Generative Recommender Systems
- The Dice Roll Method: A Standardized Protocol for Repeated-Query Auditing of Large Language Model Brand Recommendations
- UniCon: A Unified Context-Centric Modeling Paradigm for CTR Prediction
- One Size Does Not Fit All: Adaptive Interest Representation Learning for Multi-interest Recommendation
- DISR: Distributional Item Representations for Sequential Recommendation
- Comparative Analysis and Unified Evaluation Framework of User Simulators for Realistic Behaviour Modelling
- DoPR: Reusable Compressed Document Prefixes for Efficient LLM Reranking
- From Topical Relevance to Answerability: Entailment Distillation for Conversational Retrieval
- CORE: Improving Compositional Reasoning in MLLM Embedding via Reranker Distillation
- Retrieval Grounding Latent Reasoning for Dense Retrieval
- Comparing Retrieval Methods for Academic Advisor Discovery: A Six-Method Study of 768 CS Faculty Profiles Across 9 US Universities
- CHSR-RRF: A curriculum-gated hybrid retrieval framework with reciprocal rank fusion and leakage-aware benchmarking for educational RAG
- Making Every Tool Call Count: Necessary Tool-Evidence Path Rewards for Agentic Vision-Language Models
- When Users Don't Ask: Benchmarking Context-Driven Memory Retrieval in Conversational Agents
- DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training
- Agentic Search. More accurate and efficient results from your AI systems.
- Building a Memory-Driven Agent with NVIDIA NemoClaw
- Fresh Memory, Stale Plans: Dependency-Scoped Validation for Distributed LLM-Agent Memory
- KC-Bench: A Dynamic Interactive Benchmark for Evaluating Knowledge Conflicts in LLM Agents
- Speculative Macro Commit for Faster Tool-Using Agents
- Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments
- Research acceleration: The view inside OpenAI
- Fast Embeddings on GPUs
- AInfer-PD: Communication-Safe In-Place Prefill-Decode Multiplexing for Distributed MoE Rollouts
- FlowTT: Exploiting Computation Flow Reuse in Irregular Tensor-Train Embedding
- MeanField Surrogate Modeling for Scalable Runtime Scheduling of Concurrent Heterogeneous AI Inference on Shared GPUs
- RT-HiSS: Ray Tracing Accelerated High Dimensional Vector Similarity Searches
- Latency-Aware Orchestration for Multi-Agent LLM Workflows on Heterogeneous GPUs
- Every Kernel Is a Join: Automatic Multi-GPU Parallelism for AI Computations in Einsummable
- mzCache: On-Device LLM Memory Management under Multitasking
- CREDIT: Cost-guided Reduction-reuse with Efficient DSMEM Inter-CTA Tiling
- Frontier Reasoning Reaches the Edge: How to Deploy and Optimize Models on NVIDIA Jetson