ADu2021/skillXiv - Page 21

prefix-grouper-grpo

Implement Prefix Grouper to accelerate Group Relative Policy Optimization training by eliminating redundant prefix encoding, achieving up to 8x speedup for long-context scenarios.

presentation-video-generation

Transform lengthy documents into fully narrated presentation videos with synchronized audio-visual delivery. Automatically segments content, generates visuals, synthesizes speech, and composes final video.

pretrain-zero-active-pretraining

Bilevel min-max optimization where mask generator selects informative spans from pretraining data and mask predictor recovers them via chain-of-thought, enabling effective RL pretraining on noisy corpora without supervised fine-tuning.

pretraining-data-refinement

Improve pretraining efficiency by refining noisy data through expert-guided programs: learn to generate deletion operations that clean documents, achieving 2.6-7.2% performance gains with fewer training tokens.

pretraining-midtraining-rl-interplay

Understand when RL genuinely expands reasoning beyond pre-training through controlled experiments on synthetic tasks. Discover that RL works best at the edge of competence and process rewards reduce hacking—critical for designing effective reasoning model…

primo-r1-vision-critic

Transform video multimodal models into active process critics for robotic tasks. Use RL to incentivize explicit reasoning about progress toward goals and anchor reasoning temporally between initial and current states.

prism-benchmarking-phone-realization-in-speech

Phone recognition (PR) serves as the atomic interface for language-agnostic modeling for cross-lingual speech processing and phonetic analysis. Despite prolonged efforts in developing PR systems, current evaluations only measure surface-level transcription…

prism-diffusion-scaling

Scale inference efficiency for discrete diffusion language models through hierarchical trajectory search with adaptive pruning and self-verified feedback. Achieve 3-4× speedup versus best-of-N with equal quality.

prism-hypothesis

Unify semantic understanding and pixel-level detail in a single representation by decomposing features into frequency bands. Low frequencies encode semantics while high frequencies capture pixels—enabling one tokenizer for both understanding and generation…

prism-process-reward-guided-inference

Optimize multi-step reasoning by treating candidate solutions as particles in a process-reward energy landscape. Use PRM step-level scores to guide stochastic refinement and population resampling, achieving directional error correction without hallucination…

privileged-information-object-detection

Leverage training-time privileged information (depth, saliency maps) to improve student detector performance without inference overhead. Model-agnostic methodology applicable across detection architectures with no increase in inference complexity.

prl-process-reward-learning

Improves LLM reasoning by decomposing RL objectives into intermediate process rewards assigned to reasoning steps, improving both final accuracy and reasoning capacity without expensive Monte Carlo Tree Search.

proact-agentic-lookahead

Learn to ground LLM agent planning in real environment dynamics using Monte-Carlo Tree Search exploration combined with lightweight Monte-Carlo critics, reducing hallucination-driven planning failures in interactive tasks.

proactive-self-refinement

Enable models to refine outputs dynamically during generation based on internal signals, reducing token consumption by 41.6% while improving accuracy by 8.2%.

profuse-3d-semantic-understanding

Apply semantic understanding to 3D Gaussian Splatting scenes through dense correspondence-guided pre-registration without render-supervised fine-tuning. Achieve semantic understanding in ~5 minutes using cross-view clustering and direct language feature…

progressive-residual-warmup

Improves LLM convergence and downstream task performance by introducing time-dependent scaling to residual connections, enabling shallow layers to learn first before deeper layers activate. Apply during model pretraining to achieve 0.4-4.86 perplexity…

prompt-orchestration-markup

Use component-based markup with CSS-like styling to structure complex prompts, integrate diverse data types, and separate content from formatting for maintainable, version-control-friendly LLM applications.

promptquine-evolving-in-context-prompts

Automatically discovers optimal in-context learning prompts through evolutionary token pruning that removes redundant demonstrations to create effective 'gibberish' prompts. Matches state-of-the-art optimization with low-data regimes. Use for automated prompt…

prophet-diffusion-lm

Detect when diffusion language models converge on correct answers before completing refinement steps using confidence gap monitoring, achieving 3.4x decoding speedup

prorl-reasoning-expansion

Train LLMs to discover novel reasoning strategies beyond base model capabilities using prolonged RL with KL control, reference policy resets, and diverse task suites.

psa-pyramid-sparse-attention

Replaces binary keep/drop masks with multi-level pooled key-value representations, allowing queries to access larger receptive fields under same compute budget through hierarchical aggregation without discarding information.

puzzle-curriculum-grpo

Post-train vision-language models using automatically-verifiable puzzle environments (Jigsaw, Rotation, PatchFit) with graded rewards. Implement exploration-aware curriculum combining difficulty weighting with solution-space diversity metrics. Track…

pyramidal-wan-video-efficiency

Convert pretrained video diffusion models into pyramidal architectures via low-cost finetuning while preserving output quality. Explore step distillation for enhanced efficiency, enabling deployment of efficient inference without training from scratch.

pyvision-agentic-vision-dynamic-tools

Enable multimodal language models to autonomously generate and execute Python-based tools during visual reasoning, boosting performance on vision benchmarks by up to 31% through interactive problem-solving without relying on predefined tool sets.

q-tuning-joint-pruning-efficient-training

Dramatically reduce training data requirements (to 12.5% of original) while improving model performance using joint sample and token pruning guided by Error-Uncertainty plane diagnostics. Asymmetric pruning preserves calibration signals while removing…

qerl-quantization-rl

Combine NFVP4 quantization with LoRA to accelerate RL rollout phases while using quantization noise as implicit exploration bonus. Achieve 1.5x speedup and better strategy discovery through noise-enhanced policy entropy.

quantile-advantage-entropy-safe-reasoning

Stabilize LLM reasoning training by replacing mean-based advantage baselines with K-quantile baselines, preventing both entropy collapse and explosion while improving performance on mathematical benchmarks through response-level gating and asymmetric sample…

quantized-evolution-strategies

Fine-tune quantized LLMs directly in low-precision discrete parameter space using evolution strategies with accumulated error feedback. Overcome gradient stagnation in quantized models by accumulating fractional updates using Delta-Sigma modulation, achieving…

quco-rag

Replace unreliable model-internal confidence signals with objective corpus statistics to decide when RAG retrieval is necessary. Pre-evaluates entity rarity in training data and verifies entity co-occurrence at runtime, triggering retrieval only when…

query-bandits-hallucination

QueryBandits adaptively learns per-query rewriting strategies to reduce LLM hallucinations, achieving 87.5% improvement without model retraining.

qwen3-embedding

Leverages Qwen3 foundation models for text embedding and reranking via multi-stage training combining weakly-supervised pre-training on 150M synthetic pairs with supervised fine-tuning.

qwen3-vl-multimodal

State-of-the-art multimodal model advancing vision-language understanding and generation capabilities through improved visual encoders, dense token representations, and unified reasoning over images and text.

qwenlong-long-context-reasoning

Systematically post-train models for long-context reasoning through multi-hop data synthesis, stabilized RL with adaptive entropy control, and memory-augmented architecture supporting 4M+ token sequences. Achieves performance comparable to GPT-5 and…

r-horizon-long-horizon-reasoning

Construct multi-step reasoning benchmarks with interdependent problems to evaluate and improve long-horizon reasoning in large reasoning models. Enables evaluation of reasoning depth and breadth beyond single-step tasks.

r-wom-world-model

Ground LLM world models with retrieved current knowledge from tutorials and documentation. Reduce hallucination in environment prediction and improve long-horizon planning by 16-23% on web agent benchmarks.

r-zero-self-evolving-reasoning

Co-evolutionary framework where Challenger generates tasks and Solver solves them. Models evolve autonomously from scratch without human annotations. Improves math reasoning +6.49pts and general reasoning +7.54pts.

radial-attention-video

Accelerate video diffusion models using sparse radial attention that exploits energy decay patterns. Achieves 3.7× speedup on long videos while maintaining quality through O(n log n) complexity instead of O(n²).

ral-reinforced-attention-learning

Optimize where multimodal models attend by treating attention weights as a learnable policy, using policy gradients with advantage weighting to improve visual grounding and perception without changing model architecture.

ramp-adaptive-quantization

Learn optimal per-layer bit-width assignments for LLM quantization via RL, generalizing across models without retraining. Achieves superior compression under fixed bit budgets.

rank-one-safety-alignment

Apply rank-one weight modifications to amplify model safety via residual stream steering, requiring no fine-tuning and preserving utility on standard benchmarks