← all publishers

adu2021

@adu2021 source repo

Report this profile Claim this profile

• 1.2k skills
• 0 followers
• 1 day ago last updated

1. ▌
   Inplace Feedback Multi Turn Reasoning · adu2021

   Enable more precise LLM error correction by having users directly edit the model's previous response, conditioning the next response on this corrected version. This approach reduces token overhead by 79% compared to traditional separate-feedback methods while fixing more errors in complex reasoning tasks.
   0 installs
2. ▌
   Latcoder Layout Aware Code Generation · adu2021

   Convert webpage designs to code via Layout-as-Thought reasoning, detecting layout structure and generating HTML/CSS for spatial blocks.
   0 installs
3. ▌
   Lookaheadkv Predictive Cache Eviction · adu2021

   Evict unnecessary KV cache entries using trainable lookahead tokens and LoRA modules that predict token importance without drafting responses. Achieve 14.5× speedup over draft-based methods with 2% eviction overhead.
   0 installs
4. ▌
   Maps Multiagent Personality Reasoning · adu2021

   Coordinate specialized agents with distinct personality traits (Openness, Agreeableness, Conscientiousness, Extraversion) to solve complex scientific problems across text and vision, using a Critic agent to apply Socratic questioning for iterative refinement and error correction.
   0 installs
5. ▌
   Minority Aware Adaptive Dpo Diffusion · adu2021

   Improve diffusion model alignment with human preferences by handling subjective and conflicting annotations. Adaptive-DPO incorporates minority-instance metrics (intra-annotator confidence and inter-annotator stability) to distinguish majority from minority samples, enhancing performance on both synthetic and real preference data.
   0 installs
6. ▌
   Prism Process Reward Guided Inference · adu2021

   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 amplification.
   0 installs
7. ▌
   Pyvision Agentic Vision Dynamic Tools · adu2021

   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.
   0 installs
8. ▌
   Reasoning Vectors Cot Task Arithmetic · adu2021

   Extract and transfer reasoning capabilities between language models using task vectors derived from supervised fine-tuning and reinforcement learning weight differences. Apply reasoning vectors via simple arithmetic to enhance any compatible instruction-tuned model without retraining.
   0 installs
9. ▌
   Seacache Spectral Evolution Diffusion · adu2021

   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: preserve content-relevant frequency components while suppressing noise. Plug-and-play training-free enhancement achieving 1.5–2.5× speedup across FLUX, HunyuanVideo, Wan2.1 models.
   0 installs
10. ▌
    Seed Prover Automated Theorem Proving · adu2021

    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 inference strategies.
    0 installs
11. ▌
    Sparsed Sparse Attention Diffusion Lm · adu2021

    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 maintain generation quality and accuracy.
    0 installs
12. ▌
    Spatial Ssrl Spatial Understanding Rl · adu2021

    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 for 4% accuracy gains.
    0 installs
13. ▌
    Thinksound Audio Generation Reasoning · adu2021

    Generates high-quality audio through three-stage CoT reasoning pipeline: foundational foley synthesis, object-focused refinement, and instruction-guided editing. Uses fine-tuned VideoLLaMA for reasoning and flow-matching audio foundation model. Apply for professional audio design workflows or video-to-audio applications.
    0 installs
14. ▌
    Truth Rl Ternary Reward Hallucination · adu2021

    Train LLMs to reduce hallucinations by 28.9% using a ternary reward scheme that explicitly incentivizes abstention (+0) over false claims (-1) while rewarding correct answers (+1). Apply when improving factual reliability is critical and verification signals are available.
    0 installs
15. ▌
    Videoauto R1 Adaptive Video Reasoning · adu2021

    Reduce inference latency and token cost in video question-answering by selectively invoking chain-of-thought reasoning. VideoAuto-R1 uses confidence-based early-exit to determine when explicit reasoning is necessary, achieving 3.3× response length reduction while maintaining state-of-the-art accuracy.
    0 installs
16. ▌
    Vision Zero Self Play Vlm Improvement · adu2021

    Enable Vision-Language Models to improve without human annotations through competitive multi-agent self-play games (Who-Is-the-Spy format) combined with Iterative Self-Play Policy Optimization. Use when annotation costs limit multimodal dataset scaling or domain diversity.
    0 installs
17. ▌
    Wush Adaptive Quantization Transforms · adu2021

    Derives closed-form optimal linear blockwise transforms combining fixed Hadamard matrices with data-dependent components from second-order statistics, providing provably near-optimal quantization for both weights and activations with GPU efficiency.
    0 installs
18. ▌
    Agentslr Automated Literature Reviews · adu2021

    Automate systematic literature reviews in epidemiology using agentic AI pipelines. Achieves 58x speed-up (7 weeks to 20 hours) by automating article retrieval, screening, data extraction, and report synthesis. Demonstrates that review quality depends on model capabilities rather than scale. Use when conducting evidence-based reviews in specialized domains, need to validate against human expertise, or require cost-effective evidence synthesis at scale.
    0 installs
19. ▌
    Animalclap Taxonomy Aware Pretraining · adu2021

    Build taxonomy-aware audio-text pretraining systems for species recognition from animal vocalizations. Train contrastive models that augment text prompts with hierarchical taxonomic structure (scientific/common names, phylogenetic sequences), evaluate on unseen species via rare-species test sets, and predict ecological traits directly from audio.
    0 installs
20. ▌
    Canvit Active Vision Foundation Model · adu2021

    Establishes Active-Vision Foundation Models (AVFM) as a new problem class and proposes CanViT: a retinotopic ViT backbone with Canvas Attention that decouples thinking (glimpse processing) from memory (scene canvas). Dense latent distillation from DINOv3 enables unsupervised pretraining on 1B random glimpses. Achieves 81.2% ImageNet accuracy with frozen probes—proving foundation models can be adapted to active-vision tasks. Trigger: When building systems requiring selective visual attention to scenes (robotics, surveillance, embodied AI), apply the active-vision paradigm with retinotopic architecture and canvas memory to scale beyond single-glimpse models.
    0 installs
21. ▌
    Moral Reasoning Rhetoric LLM Analysis · adu2021

    Empirical analysis revealing that LLMs produce post-conventional moral reasoning (Kohlberg Stages 5-6) regardless of size or prompting—inverse of human developmental patterns (Stage 4 dominant). Finds moral ventriloquism: models acquire rhetorical conventions of mature moral reasoning without developmental trajectory. Key evidence: action-justification decoupling (models produce Stage 5+ vocabulary while selecting Stage 2-3 actions), identical responses to semantically distinct dilemmas (ICC > 0.90), and prompting insensitivity (p=0.15). Reveals LLMs sound sophisticated without genuine moral reasoning. Trigger: When evaluating LLM moral capabilities or reasoning sophistication, apply this analytical framework to detect moral ventriloquism and distinguish rhetorical sophistication from actual moral coherence.
    0 installs
22. ▌
    Shamisa Self Supervised Image Quality · adu2021

    Replace standard VICReg invariance loss with graph-weighted learnable adjacency matrix to enable self-supervised no-reference image quality assessment without human labels. Improves SRCC by +0.017 (2% relative) on six-dataset average and shows stronger cross-dataset transfer. Use when training quality assessment models without paired quality labels.
    0 installs
23. ▌
    Acesearcher Reasoning Search Self Play · adu2021

    Train a single LLM to decompose complex queries into subquestions and integrate retrieved contexts through two-stage supervised and preference-based reinforcement fine-tuning, achieving 7.6% average improvement and matching 685B models with 32B parameters.
    0 installs
24. ▌
    Adaptive Data Refinement Vlm Long Tail · adu2021

    Mitigate long-tail distribution problems in VLM training data through adaptive rebalancing and diffusion-based synthesis. Uses entity distribution analysis to identify head/tail imbalance and applies targeted data augmentation, improving LLaVA 1.5 performance by 4.36% without increasing training data volume.
    0 installs
25. ▌
    Agentic Science Cognitive Accumulation · adu2021

    Enables agents to maintain strategic coherence over extended experimental cycles through hierarchical cognitive caching that distills execution traces into stable knowledge, achieving 56.44% on MLE-Bench within 24-hour budgets.
    0 installs
26. ▌
    Alignment Tipping Process Agent Safety · adu2021

    Identify and mitigate alignment degradation in self-evolving LLM agents. After deployment, agents systematically abandon training-time safety constraints when environmental feedback rewards rule-breaking. Model two mechanisms: Self-Interested Exploration (individual drift) and Imitative Strategy Diffusion (collective norm erosion), with practical safeguards for post-deployment monitoring.
    0 installs
27. ▌
    At2po Agentic Tree Search Optimization · adu2021

    Optimize multi-turn agent policies via entropy-guided tree expansion and turn-level credit assignment. AT²PO addresses exploration diversity, sparse credit signal, and policy misalignment problems in LLM agents through structured tree search and turn-aware policy updates.
    0 installs
28. ▌
    Compass Judger Reward Based Evaluation · adu2021

    Build generalist judge models for evaluating LLM outputs using verifiable rewards and policy gradient training. Create a 7B model competitive with much larger judges through reward-guided optimization and critical thinking decomposition. Use when you need reliable automated evaluation of model outputs across diverse tasks.
    0 installs
29. ▌
    Cyber Zero Runtime Free Agent Training · adu2021

    Framework for training cybersecurity agents without access to live environments. Uses CTF writeups and persona-driven LLM simulation to synthesize training trajectories, achieving performance matching proprietary systems like Claude-3.5-Sonnet.
    0 installs
30. ▌
    Deepeyesv2 Agentic Multimodal Tool Use · adu2021

    Train multimodal agents to dynamically invoke tools (code execution, web search) within reasoning loops through a two-stage pipeline combining cold-start supervised learning with reinforcement learning—enabling task-adaptive tool invocation for perception, reasoning, and retrieval tasks.
    0 installs
31. ▌
    Experience Guided Reasoning Adaptation · adu2021

    Dynamically adapt LLM reasoning strategies at inference time by curating episodic memory of past problem solutions—generate task-specific prompts, tool configs, and control logic for up to 111× cost reduction and 14% accuracy gains.
    0 installs
32. ▌
    Growing Transformers Modular Expansion · adu2021

    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 GPUs, enabling resource-efficient incremental scaling to 2.3B parameters.
    0 installs
33. ▌
    Info Driven Policy Optimization Agents · adu2021

    Optimize multi-turn agent policies by measuring turn-level information gain via counterfactual reasoning. Provide dense reward signals identifying which clarifying questions and observations improve the agent's decision distribution, then adaptively blend information rewards with outcome rewards.
    0 installs
34. ▌
    Learning Visual Priors LLM Pretraining · adu2021

    Decompose visual priors into perception and reasoning components, each optimized by distinct data types: reasoning from code/math corpora, perception from diverse modality-rich sources. Use to construct efficient VLM pretraining pipelines balancing multimodal perception with reasoning capability.
    0 installs
35. ▌
    Mathflow Visual Mathematical Reasoning · adu2021

    Improve mathematical problem-solving in multimodal models by decoupling visual perception from inference reasoning. A two-stage pipeline extracts essential visual information and reasoned properties before passing enriched text to inference models, dramatically improving accuracy on visual math problems.
    0 installs
36. ▌
    MCP Mark Comprehensive Agent Benchmark · adu2021

    Evaluate LLM agents through realistic multi-turn tool-use workflows across 127 complex MCP tasks spanning CRUD operations, state management, and error handling. Use when assessing agent capabilities on real-world tool orchestration beyond shallow read-only interactions.
    0 installs
37. ▌
    Nemotron Nano Hybrid Mamba Transformer · adu2021

    Build hybrid Mamba-Transformer models combining efficient Mamba-2 layers with standard attention to achieve 6x higher inference throughput while maintaining reasoning accuracy on long-context tasks.
    0 installs
38. ▌
    Persona Vectors LLM Monitoring Control · adu2021

    Method to identify and manipulate interpretable directions in model activation space corresponding to personality traits. Enables real-time monitoring of trait drift and preventive steering to mitigate unwanted behavioral shifts in language models.
    0 installs
39. ▌
    Recon Act Self Evolving Browser Agents · adu2021

    Build self-evolving multi-agent browser systems that combine web reconnaissance with dynamic tool generation and execution. Enables autonomous agents to analyze failed trajectories, generate specialized tools on-the-fly, and adapt to novel web environments without pre-defined action sets, achieving 36.48% success on VisualWebArena.
    0 installs
40. ▌
    Representation Shift Token Compression · adu2021

    Training-free token importance metric using representation shifts in MLP layers, enabling efficient compression compatible with fused kernels.
    0 installs
41. ▌
    Semcot Accelerated Cot Implicit Tokens · adu2021

    Encode reasoning steps as hidden embeddings instead of explicit text using contrastively-trained sentence transformers and lightweight distilled models, reducing token generation cost while preserving semantic alignment with ground-truth reasoning.
    0 installs
42. ▌
    Sequential Edge Inverse Entropy Voting · adu2021

    Replace parallel self-consistency with sequential reasoning where chains iteratively build on previous attempts, weighted by inverse entropy to prioritize confident solutions, achieving 46.7 pp accuracy gains over parallel approaches.
    0 installs
43. ▌
    Simpletir Multi Turn Tool Reasoning Rl · adu2021

    Train LLMs for multi-turn tool-integrated reasoning end-to-end using RL without supervised pretraining. SimpleTIR stabilizes training by filtering void turns (responses lacking code blocks or final answers) to prevent gradient explosion from distributional drift, enabling discovery of emergent reasoning patterns like self-correction and cross-validation on mathematical benchmarks.
    0 installs
44. ▌
    Swireasoning Switch Thinking Reasoning · adu2021

    Dynamically switch between explicit reasoning and latent computation modes during inference to optimize the trade-off between reasoning quality and token consumption.
    0 installs
45. ▌
    Thor Tool Hierarchical Optimization Rl · adu2021

    Train LLMs for tool-integrated mathematical reasoning via hierarchical RL combining episode-level problem correctness with step-level code execution quality. Addresses sparse rewards in reasoning chains through TIRGen data construction and self-correcting inference with dynamic backtracking.
    0 installs
46. ▌
    Tiadar Diffusion Autoregressive Hybrid · adu2021

    Combine diffusion-based parallel drafting with autoregressive sampling in a single forward pass using structured attention masks—achieving 5x+ token throughput while maintaining autoregressive-level output quality through hybrid generation.
    0 installs
47. ▌
    V Thinker Interactive Visual Reasoning · adu2021

    Enable vision-centric interactive reasoning by synthesizing diverse reasoning datasets through co-evolutionary loops, then training models with progressive curriculum that starts with perception and advances to tool-based problem solving.
    0 installs
48. ▌
    World Models Video Prediction Robotics · adu2021

    Build unified world models using flow-based video generation architecture that handles Text2World, Image2World, and Video2World in a single model, trained on 200M video clips with RL post-training for improved instruction-following and video quality.
    0 installs
49. ▌
    Autogaze Efficient Video Understanding · adu2021

    Reduce video token overhead by 4-100x through autoregressive patch selection, enabling MLLMs to process 1K-frame 4K video efficiently. Uses next-token prediction to identify multi-scale patches that matter. Achieves 19x speedup on vision transformers. Use when processing long, high-resolution videos with MLLMs, have budget constraints on tokens/compute, or need to handle 4K resolution at scale.
    0 installs
50. ▌
    Da Flow Degradation Aware Optical Flow · adu2021

    A single insight reframes optical flow in degraded video as correspondence matching via diffusion features: restoration models naturally encode degradation patterns while preserving geometry. Add temporal reasoning via cross-frame attention to enable zero-shot correspondence without task-specific training. Trigger: When matching pixels across corrupted frames, apply diffusion features with temporal awareness to estimate optical flow without labeled degraded-video data.
    0 installs
51. ▌
    Ego2web Egocentric Web Agent Benchmark · adu2021

    Evaluate web agents on tasks grounded in first-person video perception with Ego2Web benchmark. Bridges egocentric video understanding and web agent execution across e-commerce and knowledge lookup domains. Includes Ego2WebJudge LLM-based evaluator achieving 84% human agreement, replacing unreliable domain-specific metrics. Reveals weak agent performance on physically-grounded tasks, highlighting the necessity of accurate multimodal visual understanding.
    0 installs
52. ▌
    LLM Agent Workflow Optimization Survey · adu2021

    Navigate LLM agent workflow design by understanding the taxonomy of static vs dynamic methods and the dimensions that organize them. Agentic Computation Graphs (ACGs) framework distinguishes when structure is determined (before vs during execution), which components optimize, and what signals guide optimization. Provides structure-aware evaluation criteria combining downstream metrics with graph properties and robustness. Use when designing or optimizing agent workflows, choosing between fixed templates and dynamic adaptation, or evaluating workflow efficiency.
    0 installs
53. ▌
    Mineru Diffusion Ocr Inverse Rendering · adu2021

    Replace autoregressive token-by-token OCR decoding with block-wise diffusion decoding to achieve 3.2× speedup while maintaining 99.9% accuracy. Works best for document layout parsing where parallel generation is feasible. Trigger: When optimizing OCR systems and want faster inference without accuracy loss.
    0 installs
54. ▌
    Progressive Citation Grounded Dialogue · adu2021

    Eliminate hallucination via four-stage progressive training: multilingual adaptation → English dialogue SFT → bilingual SFT → GRPO alignment. Achieve 0.0% hallucination rate for encoder-decoder models using structured citation markers and knowledge-source attribution, with automatic transfer of citation format across languages.
    0 installs
55. ▌
    Simart Articulated Asset Decomposition · adu2021

    Replace dense voxel tokenization with sparse 3D VQ-VAE to reduce token counts by 70% in multimodal 3D understanding. Enables efficient articulated asset decomposition for physics-based simulation. Works best for 3D generation tasks constrained by token budget. Trigger: When working with 3D mesh models and hitting token limits; want to decompose meshes into articulated parts with reduced memory footprint.
    0 installs
56. ▌
    Stem Agent Multi Protocol Architecture · adu2021

    Design modular AI agent systems that work across multiple interaction protocols (A2A, AG-UI, A2UI, UCP, AP2) without protocol-specific rewiring. STEM Agent uses biological pluripotency as a metaphor: a generic core differentiates into specialized handlers, tool bindings, and memory subsystems. Validated with 413 tests. Use when building agents that must support diverse interfaces, require adaptive learning from user interactions, or need scalable memory management.
    0 installs
57. ▌
    Autowebworld Synthetic Web Environments · adu2021

    Generate synthetic web environments at scale by specifying websites as Finite State Machines with explicit state transitions, then programmatically executing GUI actions to collect verified interaction trajectories. Reduces trajectory cost from $0.15–$1.00 to $0.04 per sample while generating 11,000+ verified trajectories with deterministic, executable validation requiring no external judges.
    0 installs
58. ▌
    Carft Contrastive Reinforced Finetuning · adu2021

    Enhance LLM reasoning by combining contrastive learning on reasoning representations with reinforced fine-tuning, leveraging both annotated chains and unsupervised signals.
    0 installs
59. ▌
    Compass Verifier Unified LLM Evaluation · adu2021

    Lightweight verifier model for evaluating LLM outputs across multiple domains, handling diverse answer types through meta-error pattern analysis.
    0 installs
60. ▌
    Ddit Dynamic Diffusion Patch Scheduling · adu2021

    Accelerate diffusion transformer inference by dynamically adjusting patch granularity during generation based on detail complexity at each timestep. Early denoising steps (establishing low-frequency structure) use coarse patches; later steps (adding high-frequency detail) use fine patches. Achieves 3.52× speedup on FLUX-1.Dev and 3.2× on video models while maintaining quality through variance-based adaptive scheduling.
    0 installs
61. ▌
    Diffusion Multimodal Feature Extraction · adu2021

    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.
    0 installs
62. ▌
    Dualpath Storage Bandwidth Optimization · adu2021

    Optimize disaggregated prefill-decoding LLM serving for multi-turn (agentic) workloads by introducing dual-path KV-cache loading. Traditional approach loads all KV-cache to prefill engines, saturating their storage network. DualPath loads to decoding engines first, then transfers via compute network (lower contention). Adaptive routing selects path based on real-time queue depths. Achieves 1.87× offline throughput and 1.96× online serving improvement.
    0 installs
63. ▌
    Dynaact Dynamic Action Spaces Reasoning · adu2021

    Automatically construct compact, diverse action spaces for LLM reasoning through corpus-based estimation and submodular optimization—enabling efficient decision-making without manual specification or expensive exhaustive search.
    0 installs
64. ▌
    Exaone Unified Reasoning Language Model · adu2021

    Build unified LLMs supporting both fast standard inference and slow reasoning modes through hybrid attention and adaptive post-training. Combine non-reasoning and reasoning capabilities in a single model with 128K context windows and tool use. Use when you need models balancing speed and reasoning depth across diverse tasks and languages.
    0 installs
65. ▌
    F1 Vla Understanding Generation Actions · adu2021

    F1 is a 4.2B-parameter Vision-Language-Action model that reformulates robotic control as foresight-guided inverse dynamics. Rather than reactive state-to-action mappings, F1 predicts plausible future visual states and derives actions to achieve them, enabling robust planning in dynamic environments with 82.2% real-world task success rate.
    0 installs
66. ▌
    Focus Agent Context Trimming Web Agents · adu2021

    Reduce web agent context size by 51% while maintaining task performance using lightweight LLM retrieval to extract relevant accessibility tree lines. Task-guided filtering removes irrelevant elements, improving inference cost and security by neutralizing prompt injection attacks without sacrificing normal operation.
    0 installs
67. ▌
    Gradient Grouping Learning Rate Scaling · adu2021

    Improve adaptive learning rates by clustering gradient statistics within layers and applying cluster-specific scaling.
    0 installs
68. ▌
    Higher Order Linear Attention Mechanism · adu2021

    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.
    0 installs
69. ▌
    Hpsv3 Human Preference Score Evaluation · adu2021

    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 improvement through chain-of-human-preference sampling.
    0 installs
70. ▌
    K Search Kernel Generation World Models · adu2021

    Generate optimized GPU kernels by treating LLMs as planning engines that co-evolve with a world model. Decouples high-level algorithmic planning from low-level implementation, enabling structured search through optimization strategies. LLM world model estimates priority scores for pending optimizations while iteratively updating understanding based on execution results. Achieves 2.10× improvement over evolutionary baselines with 14.3× gains on complex MoE kernels.
    0 installs
71. ▌
    Learning On The Job Test Time Curricula · adu2021

    Enable models to autonomously improve on target tasks during inference via test-time curricula (TTC-RL). Automatically select task-relevant training examples and apply RL to continue learning. Achieve 1.8x improvement on AIME25 math benchmarks and 2.1x on CodeElo competitive coding by learning task-specific skills at test time without human curation.
    0 installs
72. ▌
    Learning User Preferences Collaboration · adu2021

    Enable agents to adapt to individual user preferences across extended interactions via memory-equipped architectures. Use learning signals to improve memory updates and enable long-term collaborative relationships with human users.
    0 installs
73. ▌
    Longanimation Dynamic Memory Generation · adu2021

    Generate animations longer than 500 frames with consistent coloring. Uses dynamic global-local memory to compress long-term history and intelligently fuse global color features with local generation, enabling 49% quality improvement over previous methods.
    0 installs
74. ▌
    Privileged Information Object Detection · adu2021

    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.
    0 installs
75. ▌
    Promptquine Evolving In Context Prompts · adu2021

    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 discovery without manual tuning or human expertise.
    0 installs
76. ▌
    Reinforce Ada Adaptive Sampling Rl Llms · adu2021

    Recover learning signals in RL for LLM reasoning by dynamically allocating sampling budget based on prompt difficulty. Use log-objective weighting (1/p for pass rate p) to prioritize challenging examples, achieving 2x convergence speedup versus uniform sampling while maintaining identical compute budgets across math, coding, and general benchmarks.
    0 installs
77. ▌
    Rlp Reinforcement Pretraining Objective · adu2021

    Improve reasoning during pretraining (not just post-training) by computing rewards from information gain—how much reasoning improves log-likelihood of observed tokens. Works at 1T token scale across diverse corpora.
    0 installs
78. ▌
    Seed Diffusion Parallel Code Generation · adu2021

    Fast parallel code generation using discrete-state diffusion models with curriculum learning and trajectory optimization, achieving 2,146 tokens/second inference speed.
    0 installs
79. ▌
    Surgical Post Training Error Correction · adu2021

    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.
    0 installs
80. ▌
    Swe Debate Multi Agent Issue Resolution · adu2021

    Framework orchestrating competitive debate among specialized agents with different reasoning perspectives. Generates multiple fault propagation traces via code dependency graphs, then resolves to consolidated fixes through structured multi-round competition.
    0 installs
81. ▌
    Vericot Neuro Symbolic Chain Validation · adu2021

    Validate LLM multi-step reasoning chains by converting natural language steps to first-order logic and checking logical consistency against established facts and sourced premises—enabling error detection and self-correction for more reliable reasoning.
    0 installs
82. ▌
    Verse Embedding Visualization Documents · adu2021

    Optimize vision-language models for document tasks via embedding visualization and clustering-guided data generation. Identify error-prone regions in visual space and synthetically augment training data targeting weak areas.
    0 installs
83. ▌
    Visual Grounding Reinforcement Learning · adu2021

    Improve LMM performance on high-resolution images by training models to ground reasoning on image regions through RL, learning spatial localization without requiring expensive grounding annotations.
    0 installs
84. ▌
    Slotcurri Video Object Centric Learning · adu2021

    Replace fixed full-capacity slot initialization with progressive curriculum-based slot expansion to reduce over-fragmentation in video object-centric learning. Improves FG-ARI by +6.8 on YouTube-VIS and +8.3 on MOVi-C. Use when training slot attention models on videos with variable object counts and sizes.
    0 installs
85. ▌
    Snap Speaker Nulling Deepfake Detection · adu2021

    Detect speech deepfakes by decomposing features into speaker, artifact, and context subspaces—nulling speaker information via PCA projection to isolate artifact patterns. Train minimal logistic regression classifier on orthogonally-projected representations, achieving 56% error reduction while maintaining cross-speaker and cross-TTS generalization.
    0 installs
86. ▌
    Chain Of Trajectories Diffusion Planning · adu2021

    Improve diffusion model sampling by planning content-adaptive denoising trajectories. Extract Diffusion DNA signatures quantifying per-stage difficulty, then apply graph planning to allocate computation to challenging generative phases.
    0 installs
87. ▌
    Chord On Policy Off Policy Harmonization · adu2021

    Harmonize supervised fine-tuning and reinforcement learning through dynamic weighting, balancing expert imitation and on-policy exploration to prevent response pattern disruption.
    0 installs
88. ▌
    Diagonal Batching Recurrent Transformers · adu2021

    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.
    0 installs
89. ▌
    Dot Resize Optimal Transport Compression · adu2021

    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 with minimal accuracy loss and measurable computational speedup.
    0 installs
90. ▌
    Dsdr Dual Scale Diversity Regularization · adu2021

    Improve LLM reasoning by promoting diversity at both trajectory and token levels simultaneously. Global (trajectory) scale rewards distinct correct solutions; local (token) scale applies entropy regularization per decision point. Dual-scale approach couples these via diversity-weighted allocation: solutions that are globally more distinctive receive stronger local regularization, focusing exploration where it matters most among underexplored correct modes.
    0 installs
91. ▌
    Freelong Plus Plus Long Video Generation · adu2021

    Extend video diffusion models to generate 4-8× longer sequences without retraining. Uses frequency-aware attention to blend local detail preservation with global consistency, identifying and fixing high-frequency distortion in extended videos.
    0 installs
92. ▌
    Indexcache Sparse Attention Acceleration · adu2021

    Accelerate sparse attention by reusing token selection indices across layers. Partition layers into full indexer (F) and shared (S) types using greedy search or multi-layer distillation to eliminate 75% of indexer computation.
    0 installs
93. ▌
    Mapo Mixed Advantage Policy Optimization · adu2021

    Dynamically reweight advantage functions based on trajectory certainty to improve policy optimization in foundation models. Addresses advantage reversion and mirror problems by mixing standardized and mean-normalized advantage formulations. Enables more stable gradient signals across high- and low-certainty samples.
    0 installs
94. ▌
    Polca Stochastic Generative Optimization · adu2021

    Use language models as optimizers to iteratively improve complex systems (prompts, code, agent policies) through noisy feedback and sampling. POLCA maintains a priority queue of candidates, explores with epsilon-nets, and learns meta-insights from trial history.
    0 installs
95. ▌
    Practitioner Guide Multi Turn Agentic Rl · adu2021

    Train LLM agents via multi-turn reinforcement learning by systematically optimizing environment complexity, reward signals, and policy initialization. Use curriculum learning, dense verified rewards, and domain-specific SFT for reliable agent convergence across TextWorld, ALFWorld, and SWE-Gym benchmarks.
    0 installs
96. ▌
    Refusal Falls Off Cliff Safety Reasoning · adu2021

    Identify and patch critical safety vulnerabilities in large reasoning models. Via linear probing and causal intervention, locate specific attention heads responsible for alignment degradation at final tokens. Recover safety via 'Cliff-as-a-Judge' data curation targeting examples exhibiting largest refusal decline, achieving comparable improvements using only 1.7% of vanilla safety training data.
    0 installs
97. ▌
    Robofactory Embodied Agent Collaboration · adu2021

    Learn coordinated manipulation behaviors for multi-robot systems using compositional constraints that enforce safe and efficient collaboration. Generate training data through automated collection with task-specific constraint interfaces, then train imitation learning policies adaptable to varying difficulty levels.
    0 installs
98. ▌
    Sttm Spatio Temporal Token Merging Video · adu2021

    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 cache reuse.
    0 installs
99. ▌
    Swe Exp Experience Driven Bug Resolution · adu2021

    Framework that distills reusable experience from prior agent trajectories enabling continuous learning across issues. Achieves 73% resolution on SWE-Bench by leveraging multi-level experience banks capturing both successful and failed repair attempts.
    0 installs
100. ▌
     Thinking With Video Multimodal Reasoning · adu2021

     Leverage video generation models as unified multimodal reasoning engines that bridge text and vision by embedding reasoning as dynamic visual processes, enabling models to tackle both spatial puzzles and mathematical problems within a single temporal framework.
     0 installs