ADu2021/skillXiv - 24ページ

ruscarl-rubric-scaffolded-rl

Guide LLM exploration through rubric-based scaffolding that gradually diminishes, enabling models to internalize reasoning patterns while maintaining exploration quality for robust RL training.

原文の言語: 英語

rwml-reinforcement-world-models

Train LLM agents to anticipate environment consequences by learning world models through reinforcement learning with embedding-space similarity rewards, avoiding task-specific labels while enabling robust environment adaptation.

原文の言語: 英語

safety-at-one-shot-lm-repair

Recover safety alignment in fine-tuned LLMs using only a single safety example, without sacrificing utility. Leverages low-rank structure of safety gradients to enable minimal-cost correction even when models have been compromised by extensive harmful…

原文の言語: 英語

saffron-safety-scaling

Implement Saffron's multifurcation reward model approach to achieve efficient inference-time safety scaling, improving robustness against prompt injection attacks while reducing computational overhead.

原文の言語: 英語

sage-long-video-agents

Enable agents to reason over long videos through multi-turn reasoning for complex questions and single-turn answering for simpler queries. Equip orchestrator VLM with multiple tools (web search, speech transcription, temporal grounding). Use synthetic data…

原文の言語: 英語

sage-retrieval-research-agents

Build effective retrieval systems for deep research agents by understanding query-retriever mismatch, preferring BM25 for keyword-oriented queries, and augmenting corpus documents with metadata and LLM-generated keywords to improve information discovery.

原文の言語: 英語

sage-rl-agent

Enable agents to continuously improve by accumulating reusable skills across sequential task chains. Train via GRPO across task sequences where skills persist and compound, provide dual rewards for both task completion and skill generation/reuse—improving…

原文の言語: 英語

sage-self-hinting-rl

Inject privileged hints during GRPO training to reshape rollout distributions when advantage collapses occur, increasing outcome diversity without changing task rewards. Hints removed at deployment; policy automatically learns when to use hints via online…

原文の言語: 英語

sail-rl-adaptive-reasoning

Teach multimodal models to determine when deep reasoning is necessary and how to reason effectively through dual-reward reinforcement learning, preventing both overthinking on simple tasks and underthinking on complex ones while reducing hallucinations.

原文の言語: 英語

salad-achieve-high-sparsity-attention-via-efficien

Implement techniques from SALAD: Achieve High-Sparsity Attention via Efficient Linear Attention Tuning for Video Diffusion Transformer. Diffusion Transformers have recently demonstrated remarkable performance in video generation

原文の言語: 英語

sample-more-think-less-gfpo

Group Filtered Policy Optimization (GFPO) reduces inference-time computation by sampling larger groups during training and filtering responses based on length and token efficiency to teach models efficient reasoning.

原文の言語: 英語

samtok-mask-tokenization

Tokenize any region mask into two special tokens, enabling pixel-wise tasks like segmentation and region understanding in multimodal language models without architectural changes. Use when adapting vision-language models to perform pixel-level understanding…

原文の言語: 英語

scale-selective-test-time

Decomposes math problems into sequential sub-problems, assesses difficulty, and allocates simple ones to fast System 1 reasoning while directing complex ones to deliberate System 2. Save 33-53% tokens while improving accuracy by up to 13.75 points on AIME.

原文の言語: 英語

scalecua-cross-platform-computer-use-agents

Build scalable open-source agents that autonomously operate graphical user interfaces across Windows, macOS, Linux, Android, iOS, and web platforms using large-scale cross-platform training data and vision-language models for GUI understanding, element…

原文の言語: 英語

scaler-reasoning-environment

SCALER: Synthetic Scalable Adaptive Learning Environment for Reasoning. From arXiv:2601.04809

原文の言語: 英語

scaling-behavior-cloning-games

Train video game-playing foundation models discovering that increasing training data and network depth enables learning more causal policies. Release 8300+ hours of gameplay data and open-source models for real-time consumer GPU inference.

原文の言語: 英語

scaling-latent-reasoning

Scales reasoning depth through internal iteration rather than explicit generation. Ouro models perform repeated computation in latent space with entropy-regularized objectives enabling learned depth allocation. Smaller 1.4B model matches 12B standard models…

原文の言語: 英語

scaling-laws-code

Establish language-specific scaling laws for multilingual code pre-training across 7 programming languages (0.2B-14B models, 1T tokens). Show interpreted languages scale better than compiled, synergy gains depend on syntax similarity, and parallel pairing…

原文の言語: 英語

scaling-laws-optimal-data-mixtures

Predict optimal data mixture proportions for multi-domain LLM training using scaling laws that require only 10-20 small experiments. Determine which domains should contribute how much data across model sizes (186M-7B), reducing computational waste in…

原文の言語: 英語

sched-fast-diffusion-decoding

Accelerate diffusion LLM decoding by halting when confidence stabilizes using progress-dependent thresholds. SchED achieves 3.8-4.0× speedup while retaining 99.8-100% quality—training-free and model-agnostic for any dLLM.

原文の言語: 英語

sciarena-evaluation-platform

Build community-driven evaluation platforms for scientific tasks using pairwise model comparisons and human voting. Assess foundation models on literature-grounded reasoning without automated metrics.

原文の言語: 英語

scientific-image-synthesis-benchmarking-methodolog

Implement techniques from Scientific Image Synthesis: Benchmarking, Methodologies, and Downstream Utility. While synthetic data has proven effective for improving scientific reasoning in the text domain, multimodal reasoning remains constrained by the…

原文の言語: 英語

scievalkit-an-open-source-evaluation-toolkit-for-s

Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.

原文の言語: 英語

sculptor-active-context-management

Framework enabling LLMs to actively manage internal working memory and context through tools for fragmentation, summarization, and semantic search. Mitigates proactive interference and improves reasoning at extended context lengths.

原文の言語: 英語

seacache-spectral-evolution-diffusion

Accelerate diffusion models through spectral-evolution-aware caching. Exploit insight that early timesteps establish low-frequency structure while later timesteps refine high-frequency details. Apply FFT-based frequency filtering to feature cache decisions:…

原文の言語: 英語

seagent-self-evolving-computer-use

Vision-based computer use agent that self-improves through experiential learning, curriculum generation, and reward-based RL on diverse software.

原文の言語: 英語

sealqa-reasoning-search

Evaluate search-augmented language models on fact-seeking questions with conflicting or unhelpful search results, revealing critical reasoning gaps in frontier models and testing robustness to noisy information.

原文の言語: 英語

search-r2-refinement-rl

Decompose search-integrated reasoning into actor (generates trajectories) and refiner (identifies and corrects flawed steps). Refiner performs surgical cut-and-regenerate correction preserving valid prefixes. Trained end-to-end with GRPO; adds minimal…

原文の言語: 英語

search-r3-reasoning

Unify LLM reasoning with embedding generation by training models to output embeddings as direct outputs of step-by-step reasoning. Trigger: improve performance on complex retrieval tasks requiring deep reasoning.

原文の言語: 英語

see-upo-sequence-level-rl

Train multi-turn AI agents with convergence guarantees using sequential backward-induction updates, eliminating the need for separate critic networks while maintaining theoretical optimality. Use for long-horizon agentic reasoning where monotonic improvement…

原文の言語: 英語

seed-diffusion-parallel-code-generation

Fast parallel code generation using discrete-state diffusion models with curriculum learning and trajectory optimization, achieving 2,146 tokens/second inference speed.

原文の言語: 英語

seed-prover-1-5

Enable LLM agents to incrementally construct formal proofs through strategic tool orchestration rather than one-shot generation. Combines natural language reasoning, sketch translation, and formal verification in an agentic workflow trained via PPO and Rubric…

原文の言語: 英語

seed-prover-automated-theorem-proving

Framework combining formal verification feedback with long chain-of-thought reasoning to enable deep and broad mathematical reasoning for automated theorem proving. Achieves 78.1% on formalized IMO problems through lemma-based refinement and test-time…

原文の言語: 英語

seedvr2-video-restoration

Achieves single-step video restoration at 1080p resolution with 4x speedup over multi-step diffusion approaches via adversarial training, adaptive window attention, and feature matching loss.

原文の言語: 英語

segment-policy-optimization

Segment-level credit assignment for RL in LLMs using Monte Carlo advantage estimation, enabling precise reward attribution without critic models for improved reasoning task performance.

原文の言語: 英語

self-eval-t2i

Train text-to-image models from scratch for any inference step count via self-evaluation mechanism. Model evaluates its own generated samples using current score estimates as dynamic self-teacher, enabling global distribution matching without external…

原文の言語: 英語

self-evolving-agents

Enable agents to learn continuously from execution experience through hierarchical memory and autonomous reflection. Trigger: improve agent performance on long-horizon tasks by accumulating and applying experience.

原文の言語: 英語

self-forcing-video

Address exposure bias in video generation by training on self-generated sequences, achieving 17 FPS real-time performance while matching quality of slower baselines.

原文の言語: 英語

self-rewarding-vlm

Enable VLMs to self-assess visual perception accuracy through decomposed two-stage reasoning: perception generation then validation, eliminating external supervision dependency

原文の言語: 英語

semantic-visual-reconstruction

Add explicit visual supervision to VLMs by training models to autoregressively reconstruct semantic image tokens, achieving 2-3% average gains and 10-point improvements on hallucination robustness.

原文の言語: 英語