Menu
Skills dans ce dépôt
SkillsMP a collecté 1 228 skills depuis ADu2021/skillXiv. Ouvrez un skill pour examiner sa source et ses détails.
ADu2021/skillXivAffichage de 40 skills collectés sur 1 228.
Combines DeepSeek Sparse Attention (DSA) achieving O(Lk) complexity, scalable RL framework allocating 10%+ compute to post-training, and large-scale agentic task synthesis with 1,800+ environments. Matches proprietary models in reasoning;…
Langue du texte source : anglais
Enable efficient speculative decoding by training discrete diffusion language models for parallel draft generation. Use AR-style distillation and scribe refinement to train dLLMs. Eliminate left-to-right error accumulation through independent parallel…
Langue du texte source : anglais
Improve diffusion model alignment by assigning step-wise rewards during denoising instead of terminal rewards. Fixes sparse reward signal mismatch in multi-step generation processes through ODE-based reward estimation.
Langue du texte source : anglais
Extend monocular depth models to handle adverse conditions (weather, darkness, sensor noise) using only 540K training samples. Applies perturbation-based consistency learning and spatial constraints to maintain robust depth prediction across challenging…
Langue du texte source : anglais
Replace LayerNorm with Derf(x) = erf(αx + s) for improved generalization in transformers. Derf outperforms LayerNorm across vision, speech, and DNA modeling—ideal when normalization-free training provides benefits without architectural complexity.
Langue du texte source : anglais
Accelerate LLM inference 6x by using block diffusion for parallel token drafting with tight coupling to the target model's hidden representations, achieving higher speedups than existing speculative methods without quality loss.
Langue du texte source : anglais
Unlocks parallelism in recurrent memory transformers through diagonal batching of the layers-segments grid, achieving 3.3x speedup on 131K-token sequences without model retraining.
Langue du texte source : anglais
Train diffusion language models to generate optimized CUDA kernels using bi-phase reinforcement learning. First phase masks and regenerates core kernel logic with provided scaffolding to prevent PyTorch shortcuts. Second phase enables end-to-end generation.…
Langue du texte source : anglais
Recast chain-of-thought reasoning as iterative denoising using diffusion principles to overcome exposure bias in autoregressive reasoning. DiffCoT enables retrospective refinement of intermediate steps while maintaining temporal consistency through causal…
Langue du texte source : anglais
Improve state-space model performance on language modeling and retrieval by applying differential mechanisms to reduce noise in intermediate representations and enhance long-context reasoning.
Langue du texte source : anglais
Apply diffusion models as native generative agents for vision-centric reasoning tasks (sequential planning, constraint satisfaction, spatial configuration) instead of text-based LLM chains. Achieves 3x+ improvements over GPT-5 and Gemini-3 on visual…
Langue du texte source : anglais
Train masked diffusion models for code generation using coupled-GRPO to optimize non-autoregressive generation. Achieves 4.4% improvement on code benchmarks while reducing autoregressive bias.
Langue du texte source : anglais
Extract fine-grained visual features from pretrained text-to-image diffusion models to improve multimodal LLMs beyond CLIP, achieving up to 6% gains through strategic fusion of unconditional and conditional diffusion representations.
Langue du texte source : anglais
Convert pre-trained autoregressive vision-language models into diffusion VLMs without architectural modifications. Use block diffusion strategy enabling arbitrary-length generation and KV-cache reuse. Hybrid attention enforces bidirectional within blocks,…
Langue du texte source : anglais
Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
Langue du texte source : anglais
Apply reasoning-focused RL to vision foundation models using GRPO to develop deep visual understanding and abstract reasoning beyond visual recognition.
Langue du texte source : anglais
Improve diffusion model capacity efficiency by directly predicting clean data instead of noise—leverage the manifold assumption that natural data occupies low-dimensional space while noise spans full dimensionality.
Langue du texte source : anglais
Optimize diffusion models for preference alignment by learning directly from group-level preferences without stochastic policies, enabling efficient and stable training.
Langue du texte source : anglais
Enable effective RL for diffusion language models via DiPO (unbiased GRPO for dLLMs) and framework optimizations. FlexAttention accelerates blockwise training, LMDeploy optimizes inference, achieving training-inference consistency—improving dLLM math…
Langue du texte source : anglais
Enables diffusion LLMs to achieve 2.5× faster inference than autoregressive models through block-wise generation with parallel inter-block decoding.
Langue du texte source : anglais
Reformulate discrete diffusion using signal-to-noise ratio for hybrid masked-uniform noise scheduling. Derive compute-optimal scaling laws through careful hyperparameter tuning, showing uniform diffusion scales more favorably in token-constrained settings…
Langue du texte source : anglais
Understand fundamental limitations of applying diffusion to discrete text: position-agnostic corruption ignores linguistic structure, and token-wise training misses multi-token dependencies. Design text diffusion systems satisfying five essential properties:…
Langue du texte source : anglais
Evaluate and optimize tool-integrated reasoning in LLMs through empirical benchmarking, performance-cost metrics (PAC, AUC-PCC), and measurement frameworks for diverse reasoning tasks.
Langue du texte source : anglais
Convert inference-time feedback into persistent, retrievable guidelines stored as agent memory. Framework enables LLMs to improve performance over time by systematically accumulating and applying learned critiques. Augmented models rapidly match test-time…
Langue du texte source : anglais
Align latent distributions with arbitrary reference distributions via explicit matching constraints rather than fixed priors. DMVAE achieves gFID 3.2 on ImageNet with 64 epochs—when you need flexibility in latent representation design for image generation.
Langue du texte source : anglais
Synthesize diverse, verifiable training tasks by executing real tools first, then reverse-deriving tasks from execution traces. Ensure diversity across tools and reasoning patterns while maintaining grounding by construction.
Langue du texte source : anglais
Train models to decompose complex problems into subproblems via divide-and-conquer reasoning. Structured approach enables systematic solution assembly and improved long-horizon reasoning compared to end-to-end generation.
Langue du texte source : anglais
Improve multimodal reasoning at test-time through confidence-guided latent optimization without retraining. Iteratively refine learnable latent think tokens via policy gradient using confidence reward. Dynamically select and update relevant image patches…
Langue du texte source : anglais
Generate multiple tokens simultaneously by having late transformer layers directly predict multiple outputs after early layer processing. No auxiliary parameters or verification needed. Achieve up to 2x generation speedup.
Langue du texte source : anglais
Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
Langue du texte source : anglais
Build open-source agents for document question-answering by modeling DocQA as information-seeking with explicit tool utilization. DocDancer uses an exploration-then-synthesis pipeline to generate high-quality training data, addressing the scarcity that limits…
Langue du texte source : anglais
Improve long-context length extrapolation by denoising instabilities in Rotary Position Embeddings (RoPE) through spectral analysis and selective head rewriting—training-free post-hoc intervention for longer context windows.
Langue du texte source : anglais
Compress LLMs by 20-30% in width while preserving functionality through optimal transport-based neuron merging. Instead of discarding neurons, redistribute their signal to retained neurons via learned transport maps. Use when you need to reduce model size…
Langue du texte source : anglais
Diagnose and fix multi-agent system failures through targeted interventions (message edits, plan changes) rather than static log analysis. DoVer recovers 18-28% of failed trials with 30-60% hypothesis validation—essential for autonomous multi-agent…
Langue du texte source : anglais
Enable Diffusion Language Models to achieve 8.5x inference speedup (24-30 steps vs. 256) through certainty-forcing distillation that trains models to achieve simultaneous high confidence across multiple tokens. Use when optimizing inference latency for dLLM…
Langue du texte source : anglais
Replace PPO's heuristic ratio-based clipping with Divergence Proximal Policy Optimization (DPPO) that directly constrains policy divergence using either Total Variation or KL, enabling lightweight approximations (Binary, Top-K) for vocabulary-scale…
Langue du texte source : anglais
Implements DPWriter from arXiv:2601.09609
Langue du texte source : anglais
Use per-layer routers trained with Monte Carlo Tree Search to dynamically skip, execute, or repeat transformer layers for each token. Saves ~5 layers per example while maintaining accuracy on diverse benchmarks.
Langue du texte source : anglais
Enable stable multi-agent reinforcement learning by normalizing advantages per-agent rather than globally, preventing gradient-norm inflation in heterogeneous multi-agent systems.
Langue du texte source : anglais
Dr. Zero: Self-Evolving Search Agents without Training Data. From arXiv:2601.07055
Langue du texte source : anglais