ADu2021/skillXiv - 第 13 页

goedel-prover-formal-theorem-proving

Train language models for formal theorem proving via expert iteration with verifier-guided self-correction and checkpoint merging.

原文语言：英语

golden-goose-task-synthesis

Synthesize unlimited verifiable training tasks from unverifiable text by converting reasoning passages into multiple-choice problems. Creates higher-quality training data for RLHF systems without requiring new human labels.

原文语言：英语

gorl-generative-online-rl

Separates generative policy optimization through latent encoder (standard RL algorithms) and conditional decoder (frozen then refined), using two-timescale alternating schedule to eliminate gradient instability from direct generative policy optimization.

原文语言：英语

governed-autonomy-drug-discovery

Balance LLM flexibility with domain rigor in scientific agents through dual-layer architecture. Enforce role-based access control and artifact-centric state management to prevent hallucinations, while preserving free-form reasoning for lower-risk tasks.

原文语言：英语

gradient-grouping-learning-rate-scaling

Improve adaptive learning rates by clustering gradient statistics within layers and applying cluster-specific scaling.

原文语言：英语

gradmem-context-memory-compression

Compress long context into compact memory tokens via iterative gradient descent. Learn to write information into prefix memory without storing full KV-caches, enabling efficient long-context reasoning and retrieval.

原文语言：英语

grao-unified-alignment

Unifies supervised fine-tuning and reinforcement learning through GRAO framework that combines multiple-output generation with group direct alignment loss for improved preference learning.

原文语言：英语

grape-group-position-encoding

Unify positional encoding methods via group action theory, encompassing RoPE and ALiBi as special cases. GRAPE enables exploration of cross-subspace feature coupling—ideal when you need principled positional encoding beyond standard implementations.

原文语言：英语

graph-optimization-test-time-compute

Optimize test-time computation through graph-based collaborative architecture where nodes represent models, edges represent information flow, and topology itself is optimizable via reinforcement learning to discover ideal model assignments and configurations.

原文语言：英语

green-vla-robotics

Train robot controllers via five-stage curriculum progressing from base vision-language models to embodiment-specific RL-refined policies. Unified action space enables cross-embodiment transfer with minimal performance loss.

原文语言：英语

group-rank-reranking-rl

Improve ranking quality via groupwise reranking with RL—process document groups jointly for within-group comparisons using GRPO with composite rewards (recall, ranking metrics, distribution alignment).

原文语言：英语

grove-moe-adjugate-experts

Enables efficient MoE architectures through heterogeneous expert sizing and dynamic activation mechanisms that adjust parameter count based on input complexity.

原文语言：英语

growing-transformers-modular-expansion

Build language models layer-by-layer on frozen embeddings, training new Transformer blocks sequentially while keeping lower layers frozen. Achieves 5% improvement over monolithic baselines on MMLU while fitting 740M trainable parameters per stage on single…

原文语言：英语

grpo-ma-multi-answer-cot-training

Stabilize and accelerate chain-of-thought RL training by sampling multiple answers per generated thought. GRPO-MA reduces gradient noise and improves convergence across math, code, vision, and manipulation tasks while cutting computational cost versus naive…

原文语言：英语

gtr-turbo-vlm

Eliminate expensive external teacher dependencies in VLM RL training via merged-checkpoint teachers. Uses TIES merging of historical RL checkpoints to create free, stable teacher models for step-level guidance—matching external teacher performance while…

原文语言：英语

guardians-of-hair-soft-boundaries

Recover fine details at soft boundaries (hair, fur) through depth refinement networks and view synthesis. Integrate plug-and-play with existing depth models via adaptive combination across monocular, stereo, and novel view tasks.

原文语言：英语

gui-360-desktop-agent-dataset

Enable training and evaluation of desktop computer-using agents through 1.2M action steps across diverse Windows applications, covering GUI grounding, screen parsing, and action prediction with hybrid GUI+API action space reflecting modern agent designs.

原文语言：英语

gui-actor-grounding

Enable GUI agents to ground actions without generating pixel coordinates by using attention-based patch-level alignment and a verifier for selecting optimal action regions from candidates.

原文语言：英语

gui-test-time-scaling

Improve GUI agent planning and action grounding through test-time scaling and reinforcement learning. Sample and evaluate multiple action candidates, then use RL to precisely target visual interface elements.

原文语言：英语

guidelines-to-prompt-large-language-models-for-cod

Implement techniques from Guidelines to Prompt Large Language Models for Code Generation: An Empirical Characterization. Large Language Models (LLMs) are nowadays extensively used for various types of software engineering tasks, primarily code generation

原文语言：英语

h-net-dynamic-chunking-hierarchical-sequence

Eliminate fixed tokenization by learning data-dependent segmentation jointly with the model through dynamic chunking, matching BPE-tokenized Transformers at equivalent compute while showing improved robustness and better downstream task performance without…

原文语言：英语

hardtests-code-verification

Generate comprehensive test cases for code problems that reliably detect wrong solutions through LLM-based edge case synthesis and test quality ranking.

原文语言：英语

hardware-agnostic-reranker-evaluation

Evaluate LLM-based document rerankers using hardware-agnostic FLOPs metrics instead of latency, enabling fair comparison of ranking quality per unit of computation across different models and deployment scenarios.

原文语言：英语

harmonyguard-safety-utility-agents

Multi-agent framework that balances safety compliance with task completion through adaptive policy extraction and dual-objective optimization. Achieves 38% improvement in policy compliance while maintaining 20% higher task completion.

原文语言：英语

hcapo-hindsight-credit-assignment

Compute step-level credit assignments via hindsight generative verification: condition the LLM on successful outcomes to compute importance ratios that filter credit by causal relevance. Synergizes macro stability with micro precision.

原文语言：英语

hermes-hierarchical-video-memory

Use KV cache as hierarchical memory for real-time video stream understanding with minimal GPU overhead, achieving 10x faster response times compared to standard methods. Use when processing continuous video streams where latency and memory efficiency are…

原文语言：英语

heteroscale-autoscaling

Scale disaggregated LLM inference (prefill-decode) via topology-aware scheduling and metric-driven policies, achieving 26.6% GPU utilization improvement and conserving hundreds of thousands GPU-hours daily

原文语言：英语

hierarchical-lvm-reasoning

Reason about long-horizon dynamics by disentangling structure and motion in video VAE latents. Learn continuous latent motion chains that preserve temporal coherence while predicting terminal keyframes, enabling efficient reasoning about multi-step scenarios.

原文语言：英语

high-entropy-minority-tokens-rl

Optimize only high-entropy tokens during RL training to achieve better reasoning performance with 80% fewer gradient updates.

原文语言：英语

higher-order-linear-attention-mechanism

Enable data-dependent higher-order interactions in attention using prefix-sufficient statistics that maintain linear time and constant state, replacing quadratic dot-product attention while preserving expressivity through compact matrix operations.

原文语言：英语

himap-travel

Solves long-horizon planning problems with global constraints by decoupling planning into strategic (resource allocation) and tactical (execution) levels. Prevents constraint drift through synchronized state tracking and cooperative bargaining.

原文语言：英语

hop-skip-overthink-diagnosis

Novel error categorization framework examining failures across hops (diversity), coverage, and overthinking. Combines human annotation with automated metrics to diagnose why reasoning models hallucinate on multi-step tasks.

原文语言：英语

hpsv3-human-preference-score-evaluation

A VLM-based preference scoring system trained on 1.17M annotated comparisons to evaluate text-to-image generation quality at scale. Uses uncertainty-aware ranking loss for fine-grained assessment across diverse images and supports iterative quality…

原文语言：英语

humanomni-multimodal-reasoning

Improve multimodal reasoning by requiring explicit context understanding before reasoning. Use specialized reward mechanisms and context-aware training to prevent information-skipping shortcuts.

原文语言：英语

hybrid-linear-attention

Design hybrid architectures mixing linear and full attention at optimal ratios. Superior standalone linear models don't necessarily excel in hybrids; recall improves significantly with more full attention layers at ratios below 3:1, enabling efficient…

原文语言：英语

hybrid-reinforcement-learning

Combine sparse verifier rewards with dense reward model scores using stratified normalization to overcome limitations of either approach alone.

原文语言：英语

hyperagents-self-improvement

Enable AI systems to recursively improve themselves by making the meta-level modification procedure itself editable, achieving open-ended capability growth.

原文语言：英语

hypergraph-memory-rag

Build hypergraph-structured memory systems for multi-step RAG that capture high-order relationships between facts, enabling stronger reasoning across long contexts. Use when combining multiple retrieved documents in complex reasoning chains that require…

原文语言：英语

i-grpo-self-feedback-reasoning

Improve LLM reasoning through iterative refinement where the model refines its best previous attempts. Two-stage training: exploratory draft generation, then conditioned refinement using GRPO. Dynamic conditioning signals evolve with policy, enabling…

原文语言：英语

ieap-image-editing

Enable robust image editing by decomposing free-form instructions into sequential atomic operations executed through a neural program interpreter.

原文语言：英语