Menu
Skills in this repository
SkillsMP has collected 1,228 skills from ADu2021/skillXiv. Open a skill to review its source and details.
ADu2021/skillXivShowing 40 of 1,228 collected skills.
Achieve up to 1.50x speedup in diffusion language models by computing head-specific sparse attention patterns once during early denoising steps and reusing them across all subsequent iterations, while preserving full attention in critical early phases to…
Accelerates LoRA fine-tuning 2.2× computationally and 1.6× wall-clock by leveraging contextual sparsity to compute gradients only for important weight channels. Uses training-free SVD sparsity estimation without full computation. Apply for efficient…
Discovers that <5% of attention heads process visual information in MLLMs; introduces SparseMM for asymmetric KV-cache allocation achieving 1.38x acceleration and 52% memory reduction.
Accelerates self-attention by reordering tokens to concentrate relevant tokens within fewer blocks. Achieves 2.75x speedup on long-context prefilling by permuting tokens so semantically related information clusters together, enabling aggressive block sparsity…
Align intermediate visual embeddings in vision-language-action models with 3D geometric representations from pretrained foundation models. Improves spatial understanding and enables faster training (3.8x speedup) without explicit 3D inputs.
Improve spatial reasoning in vision-language models through five automatically-formulated pretext tasks (patch shuffling, flipping, inpainting, depth ordering, 3D position) that require zero human annotation, using only RGB/RGB-D images and RL optimization…
Train multimodal LLMs to parse 3D point clouds and generate executable Python code representing structured indoor scene layouts with walls, doors, and objects.
Uses distilled language model (DLM) for KV cache token retrieval, reducing layer-wise retrieval parameters by 90% via head-level attention weights, with asynchronous prefetching and elastic CPU offloading. Deploy for ultra-long-context inference with memory…
Enables continuous self-improvement through corpus-grounded self-play. Challenger mines difficult examples from document corpus for Reasoner to solve. External corpus prevents task stagnation that plagues closed-loop self-play. Achieves 8.9% math, 9.8%…
A self-play framework enabling language models to develop sophisticated reasoning through competitive multi-turn games without human supervision. Achieves 10% improvement on reasoning benchmarks by training models to win against evolving opponents while…
Unify sparse and quantized representations in a single hardware-friendly format for efficient LLM inference. Exploit complementary acceleration properties—high precision for sparse operations, low precision for dense—when W4A8 bottlenecks GPU throughput.
Uses stable rank (variance distribution across semantic dimensions) as annotation-free reward signal in GRPO to incentivize high-dimensional representation structures, eliminating dependency on human preference annotations or learned reward models.
Teach multimodal language models to reflect on their reasoning and improve answers through structured RL training.
Train VLA models for robotic manipulation by using the model's own successful trajectories as self-reference for reward—enable progress-based feedback for failed attempts without external rewards or demonstrations.
Enable unified multimodal models to self-improve by using their understanding module as evaluator for generation. Apply hierarchical dual rewards: global for overall semantics and local for fine-grained details.
Enable LLMs to perform internal knowledge search using structured prompting and rule-based rewards, reducing reliance on external search while maintaining accuracy and reducing hallucination.
Generate code using diffusion-based language models with specialized warmup and noise scheduling, outperforming autoregressive baselines while supporting code editing and low-resource language scenarios. Use when building flexible code generation systems that…
Fixes training instability in Group Relative Policy Optimization for discrete language models by replacing conditional clipping with strict importance ratio bounds and self-normalized advantages. Prevents gradient spikes and policy collapse.
Justifies token-level optimization objectives as first-order approximation to sequence rewards via Routing Replay, which fixes MoE routers during policy optimization to reduce training-inference discrepancy. Use when scaling RL training on large MoE language…
Accelerate visual autoregressive (VAR) image generation 3.4× without retraining by analyzing generation stages. Exploits semantic irrelevance in detail-refinement stages where classifier-free guidance becomes redundant and features exhibit low-rank…
Reveal that Mixture-of-Experts models harbor a 'Standing Committee' of consistent expert coalitions handling majority computational load across domains. Challenges specialization assumptions and suggests training approaches like load-balancing losses may work…
Stabilize RL training in LLMs by detecting and masking gradient contributions from spurious tokens that comprise 0.01% of output but cause disproportionate instability. Identifies tokens with low probability, low entropy, and positive advantage, then…
Learn to implement Transformer Autoregressive Flows for efficient high-resolution image synthesis using latent space normalization and maximum likelihood training.
Generate videos end-to-end using normalizing flows instead of diffusion: leverage invertible transformations for exact likelihood computation, eliminate train-test mismatch, and achieve non-iterative sampling with native support for multiple tasks…
Reframe reasoning tokens as externalized computational state—the sole persistent information carrier across stateless generation cycles—not human-readable explanations. Model LLM generation as recursive token prediction where state diverges from human…
Guide multi-step reasoning through structured action templates that decompose problems into discrete, executable steps. STATe improves exploration efficiency in tree-of-thoughts by constraining action space while maintaining reasoning flexibility.
STATIC converts prefix trees into sparse matrices for vectorized constrained decoding, achieving 948x speedup over CPU and enabling production-scale recommendation systems.
Adapt LLMs efficiently by composing multiple pre-existing semantic steering vectors via Bayesian optimization, balancing adaptation gain and safety without retraining model parameters.
Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
Enable autonomous AI research systems to move beyond web search toward true research through four composable atomic capabilities: planning/decomposition, deep search/synthesis, reflection/verification, and report generation. Progressive training across…
Reduces chain-of-thought verbosity through step entropy metrics that identify and prune low-information reasoning steps while maintaining accuracy.
Train GUI agents through calibrated step-level reasoning anchored to trajectory-level validation. Use trajectory verification rather than step annotation achieving 90% accuracy with 10-100× cost reduction. Implement self-evolving training framework with…
Improve training efficiency under iteration budgets using the Unified Budget-Aware (UBA) schedule, a theoretically grounded learning rate approach governed by a single hyperparameter φ that balances adaptability across network architectures.
Train stepwise judges as generative models that perform meta-reasoning about intermediate steps, combining explainability with improved accuracy over static process reward models
Enables 2.8-5.5x longer sequences during LLM training via linear decomposition of chain rule along sequence dimension, maintaining exact gradients with lower memory cost.
Generate videos in real-time (16 FPS) by streaming frames continuously via modified flow matching with moving buffer mechanism and adaptive time embeddings.
Accelerate video LLMs by 2× with minimal accuracy loss using training-free spatio-temporal token merging that exploits video redundancy through quadtree-based spatial compression and temporal frame similarity, enabling efficient multi-turn reasoning with KV…
Improve long-horizon task success via subgoal decomposition and dense milestone-based rewards, dramatically outperforming sparse-reward RL and standard baselines.
Generate coherent, consistent long-form text through structured planning, hierarchical reflection, and Monte Carlo tree search-guided optimization.
Correct reasoning errors with minimal data collection by using an oracle to surgically fix only erroneous steps in existing trajectories. Use binary classification loss on rectified pairs with implicit KL regularization to prevent knowledge forgetting.