← all publishers

adu2021

@adu2021 source repo

1228 published skills · page 5 of 13

  1. Dcm Dual Expert Consistency · adu2021
    Accelerate video generation through dual-expert consistency distillation, using separate denoisers for semantic layout/motion and detail refinement to resolve conflicting optimization gradients.
    0
    installs
  2. Decoupled Reasoning Proving · adu2021
    Bridge the gap between informal mathematical reasoning (80% accuracy) and formal proof synthesis (8% success) by decoupling them: a general-purpose reasoner generates strategic lemmas, then a specialized prover verifies them formally. First open-source solver of post-2000 IMO problems.
    0
    installs
  3. Diffusionvl Ar To Diffusion · adu2021
    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, causal between blocks. Requires less than 5% of data compared to prior diffusion VLM methods.
    0
    installs
  4. Dive Diverse Task Synthesis · adu2021
    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.
    0
    installs
  5. Driving Gen Video Benchmark · adu2021
    DrivingGen: A Comprehensive Benchmark for Generative Video World Models in Autonomous Driving. From arXiv:2601.01528
    0
    installs
  6. Dvoting Diffusion Consensus · adu2021
    Accelerate test-time scaling for diffusion language models by identifying inconsistent tokens, selectively remask and regenerate only uncertain tokens, and aggregate across samples via voting. Achieve 5.5-22× speedup over standard iterative sampling with 6-8% accuracy gains on reasoning tasks.
    0
    installs
  7. Efficient Dlm Ar Conversion · adu2021
    Systematically convert pretrained autoregressive models into efficient diffusion language models via block-wise attention and position-dependent masking. Efficient-DLM family (1.5B/4B/8B) maintains comparable accuracy to standard AR models while delivering 4.5× higher throughput.
    0
    installs
  8. Entropy Guided Regeneration · adu2021
    Monitor Shannon entropy in LLM token distributions to detect sudden uncertainty spikes, then trigger adaptive prompt consolidation to realign conversation context and improve accuracy by 56.6% and reliability by 35.3%.
    0
    installs
  9. Evolutionary Data Synthesis · adu2021
    Generate verifiable synthetic training data (problems + solutions + verification artifacts) through evolutionary synthesis that works across math, code, and agent tasks without task-specific rules.
    0
    installs
  10. F Grpo Divergence Alignment · adu2021
    Unify LLM alignment methods through f-divergence theory. f-GRPO extends GRPO to handle any divergence measure (KL, Jensen-Shannon, Hellinger), enabling tailored alignment objectives. f-HAL combines on-policy and off-policy preference learning to prevent reward hacking while maintaining safety alignment.
    0
    installs
  11. Flow Map Trajectory Tilting · adu2021
    Uses flow maps as look-ahead operators to enable principled reward-guided diffusion by predicting trajectory endpoints at any denoising step. Deploy when applying rewards or preferences to diffusion trajectories with meaningful gradients throughout generation.
    0
    installs
  12. Gain Rl Angle Concentration · adu2021
    Improve RL training efficiency by 2.5× using angle concentration between token hidden states as a cost-effective data scheduling signal, selecting high-gradient samples dynamically.
    0
    installs
  13. Gem Agentic LLM Environment · adu2021
    A standardized environment framework for training and evaluating LLM agents, providing 24+ tasks with asynchronous vectorized execution, extensible wrappers, and integration examples for five RL frameworks. Enables reproducible agent research and training at scale.
    0
    installs
  14. Golden Goose Task Synthesis · adu2021
    Synthesize unlimited verifiable training tasks from unverifiable text by converting reasoning passages into multiple-choice problems. Creates higher-quality training data for RLHF systems without requiring new human labels.
    0
    installs
  15. Hardtests Code Verification · adu2021
    Generate comprehensive test cases for code problems that reliably detect wrong solutions through LLM-based edge case synthesis and test quality ranking.
    0
    installs
  16. Kling Omni Multimodal Video · adu2021
    Unify video generation, editing, and reasoning through Multimodal Visual Language (MVL) paradigm processing text instructions, reference images, and video contexts in shared embedding space. Employ prompt enhancer, omni-generator with diffusion transformer, and multimodal super-resolution. Support diverse user inputs with in-context generation and reasoning-based editing.
    0
    installs
  17. Kv Cache Steering Reasoning · adu2021
    Guide frozen language models toward multi-step reasoning by modifying cached key-value representations after the prefilling stage. Extract steering vectors from contrastive prompt pairs and apply them to KV cache with scalar coefficients. Improves reasoning on GSM8K, ARC, CommonsenseQA while adding only 10ms overhead per token.
    0
    installs
  18. Learning Discover Test Time · adu2021
    Apply reinforcement learning at test time to enable language models to continue adapting on specific problems, achieving state-of-the-art results across mathematics, GPU optimization, algorithms, and biology. Use when you need models to discover domain-specific problem-solving strategies without retraining.
    0
    installs
  19. Log Linear Sparse Attention · adu2021
    Reduce self-attention complexity from O(N²) to O(N log N) through hierarchical token selection and enrichment. Perform hierarchical Top-K selection progressively adopting sparse Top-K at each level. Implement sparse index transpose algorithm avoiding dense mask construction. Achieve 28.27× faster inference and 6.09× faster training.
    0
    installs
  20. Longcontext Reward Modeling · adu2021
    Train reward models to evaluate long-context responses by introducing faithfulness as a criterion and using consistency-based alignment to maintain judgment-explanation coherence.
    0
    installs
  21. Longllada Diffusion Context · adu2021
    Training-free method extending diffusion LLMs to 6x context length using NTK-based RoPE scaling, exploiting bidirectional attention stability.
    0
    installs
  22. Mact Document Understanding · adu2021
    Decomposes document processing into specialized agents (planning, execution, judgment, answer) with agent-wise adaptive test-time scaling. Achieves 9.9-11.5% performance gain with smaller models while maintaining reasoning.
    0
    installs
  23. Memento Skills Agent Design · adu2021
    Enable agents to autonomously design and refine task-specific agents by evolving externalised behavioral skills and prompts without modifying base LLM parameters.
    0
    installs
  24. Mirror Speculative Decoding · adu2021
    Run draft and target models in parallel across GPU/NPU using bidirectional speculation: draft predicts forward continuations while target speculates correction paths simultaneously. Achieve 2.8x-5.8x wall-time speedup on 14B-66B models.
    0
    installs
  25. Mixture Of Depths Attention · adu2021
    Allow attention heads to reference features from multiple depths by accessing both current-layer and depth key-value pairs. Prevent signal degradation in deep models while maintaining computational efficiency.
    0
    installs
  26. Molecular Thought Reasoning · adu2021
    Improve agent reasoning by designing thought structures that balance deep analysis, self-reflection, and exploratory thinking. Framework discovers that effective long-form reasoning exhibits molecular-like interaction patterns—specific bonds between reasoning components that enable fast entropy convergence. Method synthesizes improved reasoning trajectories using distribution-transfer, improving both model performance and RL training stability.
    0
    installs
  27. Multi Agent Tool Policy Opt · adu2021
    Train planner and worker agent roles within a single LLM via role-specific prompts and RL, avoiding multi-instance overhead while preserving specialization. Trigger: improve tool-use planning robustness to noisy outputs without deploying separate models.
    0
    installs
  28. Opensir Self Play Reasoning · adu2021
    Enable open-ended mathematical learning through co-evolutionary teacher-student self-play where a single model alternates roles to generate progressively harder problems and solve them, creating a virtuous cycle without external data.
    0
    installs
  29. Paced Weighted Distillation · adu2021
    Weight distillation loss by student pass rate using Beta kernel: suppress mastered (high pass rate) and intractable (low pass rate) problems, prioritize intermediate zone where learning signal is richest.
    0
    installs
  30. Paceevolve Evolution Search · adu2021
    Improves LLM-based evolutionary search by addressing context pollution, mode collapse, and weak collaboration through hierarchical context management, momentum-based backtracking, and adaptive sampling policies.
    0
    installs
  31. Pretraining Data Refinement · adu2021
    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.
    0
    installs
  32. Prl Process Reward Learning · adu2021
    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.
    0
    installs
  33. Progressive Residual Warmup · adu2021
    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 reduction.
    0
    installs
  34. Prompt Orchestration Markup · adu2021
    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.
    0
    installs
  35. Query Bandits Hallucination · adu2021
    QueryBandits adaptively learns per-query rewriting strategies to reduce LLM hallucinations, achieving 87.5% improvement without model retraining.
    0
    installs
  36. Reason Rank Passage Ranking · adu2021
    Enhances passage ranking through reasoning capabilities via synthesized training data and multi-stage training combining supervised fine-tuning with reinforcement learning for improved ranking accuracy.
    0
    installs
  37. Recall Hierarchical Merging · adu2021
    Prevents catastrophic forgetting in continual learning by merging models using layer-wise hidden representations as similarity proxies. Shallow layers preserve domain-general features while deep layers enable task-specific adaptation, enabling seamless multi-domain integration without task labels or historical data.
    0
    installs
  38. Rl Plus Capability Boundary · adu2021
    Hybrid-policy optimization combining multiple importance sampling and exploration-based advantage functions. Prevents capability boundary collapse while maintaining verification rewards, enabling LLMs to exceed baseline boundaries.
    0
    installs
  39. Rlad Knowledge Distillation · adu2021
    RLAD performs selective distillation during RL, using trust-region ratio objectives to balance exploration, exploitation, and teacher guidance for efficient reasoning model training.
    0
    installs
  40. Robot R1 Embodied Reasoning · adu2021
    Train vision-language models for robotic manipulation using RL to improve embodied reasoning about spatial relationships and movements, achieving 31% improvement on manipulation benchmarks.
    0
    installs
  41. Sample More Think Less Gfpo · adu2021
    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.
    0
    installs
  42. Segment Policy Optimization · adu2021
    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.
    0
    installs
  43. Sodec Diffusion Compression · adu2021
    Replaces iterative diffusion with single-step decoding for image compression. Combines VAE latents with fidelity guidance and rate annealing training. Achieves 20× decoding speedup with improved perceptual quality.
    0
    installs
  44. Spark Process Aware Rewards · adu2021
    Train process reward models without ground-truth references using synthetic verification data from generators and verifiers. SPARK achieves 67.5 F1 on ProcessBench—ideal when step-level annotations are expensive but verification is available.
    0
    installs
  45. State Over Tokens Reasoning · adu2021
    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 semantic interpretation.
    0
    installs
  46. Static Constrained Decoding · adu2021
    STATIC converts prefix trees into sparse matrices for vectorized constrained decoding, achieving 948x speedup over CPU and enabling production-scale recommendation systems.
    0
    installs
  47. Step Gui Calibrated Rewards · adu2021
    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 generation-refinement cycles and verifiable reward signals.
    0
    installs
  48. Stepwiser Generative Judges · adu2021
    Train stepwise judges as generative models that perform meta-reasoning about intermediate steps, combining explainability with improved accuracy over static process reward models
    0
    installs
  49. Streambp Efficient Backprop · adu2021
    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.
    0
    installs
  50. T3d Trajectory Distillation · adu2021
    Accelerate diffusion language models from many to few generation steps using trajectory self-distillation. Collect clean-noisy state pairs along teacher trajectories, apply reverse-KL mode-seeking distillation, and weight losses by decoding order to prioritize early predictions where cascading errors compound.
    0
    installs
  51. Taoavatar Augmented Reality · adu2021
    Create real-time full-body talking avatars for AR using hybrid parametric-Gaussian representations. Teacher-student distillation transfers pose-dependent deformations from a large network to a compact student model, enabling 90+ FPS rendering on mobile devices.
    0
    installs
  52. Teso Black Box Optimization · adu2021
    Apply tabu search enhanced with short-term tabu lists and long-term elite memory for simulation optimization with expensive, noisy evaluations. Balances exploration (avoiding cycling) and exploitation (leveraging best solutions). Use for multimodal landscapes where function evaluations are costly and multiple function calls per solution are impractical.
    0
    installs
  53. Thinking Map Spatial Agents · adu2021
    Improve agent reasoning for spatial tasks by augmenting LLMs with map tools and parallel test-time exploration. Framework uses reinforcement learning to train agents to iteratively refine hypotheses using map feedback. Parallel exploration enables agents to test multiple candidate locations before committing to answers. Introduces MAPBench benchmark for evaluating spatial reasoning in image geolocalization tasks.
    0
    installs
  54. Tool Orchestra Coordination · adu2021
    8B parameter orchestrator trained with end-to-end RL balancing outcome, efficiency, and user preference rewards to strategically coordinate diverse tools and models. Generate realistic tool-use data via ToolScale synthetic pipeline for verifiable multi-turn scenarios.
    0
    installs
  55. Tool Verification Reasoning · adu2021
    Tool Verification stabilizes self-improving reasoning models by using external tool execution as ground-truth evidence to prevent spurious consensus from becoming reinforced training signals.
    0
    installs
  56. Vcrl Variance Curriculum Rl · adu2021
    VCRL improves LLM mathematical reasoning by dynamically adjusting training sample difficulty based on group reward variance. Uses variance-based sampling with memory replay to focus on moderately-difficult samples where models succeed ~50% of the time, achieving 4.67-point improvement over GRPO on 8B models.
    0
    installs
  57. Videovla Robot Manipulators · adu2021
    Transform video generation models into robot manipulators by jointly predicting actions and future visual outcomes. VideoVLA demonstrates that imagining futures improves action reliability—ideal when you need generalizable robot learning from video imagination.
    0
    installs
  58. Voyager 3d Scene Generation · adu2021
    Generate spatially-coherent 3D point-cloud videos from single images using depth-fused diffusion with efficient world caching for infinite scene exploration.
    0
    installs
  59. Zero Shot Motion Generation · adu2021
    Generate diverse human motions from text descriptions using million-scale datasets and wavelet-enhanced quantization. Achieves state-of-the-art zero-shot generalization on out-of-domain motions, compositional descriptions, and complex choreography through efficient tokenization and scalable transformer-decoder architectures.
    0
    installs
  60. Demystifying Rl Tool Agents · adu2021
    Comprehensive recipe for RL-training tool-using agents spanning reward design, data synthesis, model scaling, and algorithm selection. Seven ranked findings: scale-dependent rewards (curriculum for 1.5B–3B; dense for 7B), semi-sparse 'Macro' rewards balance specialization/transfer, 1K-sample sweet spot with 4:3:3 difficulty mix. Achieves SOTA on TravelPlanner with smaller models than leading proprietary systems.
    0
    installs
  61. Perceptio Spatial Token Vlm · adu2021
    Enhance VLM spatial grounding by enforcing explicit intermediate spatial reasoning before text generation. Generate segmentation and depth tokens as perception pathways, anchoring responses in 2D and 3D geometric reasoning for improved referring expressions and spatial tasks.
    0
    installs
  62. Scalable Prompt Routing Moe · adu2021
    Route queries across frontier models using two-stage system: graph-based task discovery identifies ~332 latent task types via semantic similarity + preference patterns; MoE with task-specific adapters estimates quality for candidate models. Achieves <50% inference cost of strongest single model while exceeding its performance; applies when managing pools of frontier models with narrow capability gaps.
    0
    installs
  63. Scaling Dora Factored Norms · adu2021
    Optimize adapter parameter efficiency at scale by decomposing row-wise norm computation into base/cross/BA components (15× memory reduction) and fusing kernel operations. Achieves 1.5–2.0× inference speedup with 77 GB peak VRAM reduction across 8–32B vision-language models; applies when training adapter-based models with strict memory budgets across hundreds of modules.
    0
    installs
  64. Agent Long Context Benchmark · adu2021
    Build controllable benchmarks for evaluating long-context agents using environment rollouts. Generate diverse multi-step agent tasks that require maintaining context across extended interaction sequences, enabling evaluation of agent reasoning quality in scenarios with long history requirements.
    0
    installs
  65. Agent Reasoning Reward Model · adu2021
    Build multi-faceted reward models for agent trajectories that provide structured feedback on intermediate reasoning quality. Implement explicit reasoning traces, focused critiques with refinement guidance, and overall process scores to train more effective agentic agents without relying solely on sparse outcome rewards.
    0
    installs
  66. Alphaone Test Time Reasoning · adu2021
    Dynamically modulate reasoning depth at test time using alpha moments and Bernoulli scheduling to optimize inference speed-quality tradeoffs without retraining.
    0
    installs
  67. Amft Meta Learning Alignment · adu2021
    Use meta-learning to automatically balance Supervised Fine-Tuning and Reinforcement Learning signals, treating SFT and RL as complementary rewards in a unified single-stage training framework.
    0
    installs
  68. Automatic Triton Programming · adu2021
    Train specialized LLMs to generate optimized Triton GPU kernels using RL with dual rewards for correctness and syntax compliance. 8B model achieves parity with Claude-Sonnet and DeepSeek-R1 by combining supervised fine-tuning on curated code pairs with RL exploration beyond imitation learning ceilings.
    0
    installs
  69. Bifrost Patch Clip Diffusion · adu2021
    Connects multimodal language models with diffusion models using patch-level CLIP embeddings as shared latent variables, enabling controllable image generation with minimal training overhead.
    0
    installs
  70. Bro Rl Broad Rollout Scaling · adu2021
    Overcome reasoning model training plateaus by increasing rollouts per prompt (N=512) rather than training steps, addressing unsampled coupling that destabilizes learning. Theoretical analysis shows broad exploration eliminates plateau bottleneck.
    0
    installs
  71. Camels Computer Use Security · adu2021
    Protects computer use agents from prompt injection by using single-shot execution planning that generates complete control flow graphs before UI observation, preventing instruction hijacking while maintaining 57% performance on frontier models.
    0
    installs
  72. Comorag Cognitive Memory RAG · adu2021
    Iteratively retrieve and reason over long narratives using a dynamic memory workspace that integrates retrieved facts into a shared context for complex multi-hop reasoning.
    0
    installs
  73. Cross Domain Agent Knowledge · adu2021
    Create a universal memory infrastructure enabling agents across different frameworks to share experience trajectories without retraining. Improve agent performance by retrieving workflows from related domains and applying diagnostic fixes.
    0
    installs
  74. Deep Research Eval Framework · adu2021
    DeepResearchEval: An Automated Framework for Deep Research Task Construction and Agentic Evaluation. From arXiv:2601.09688
    0
    installs
  75. Depth Anything Any Condition · adu2021
    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 real-world scenarios.
    0
    installs
  76. Divide And Conquer Reasoning · adu2021
    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.
    0
    installs
  77. Dr Zero Self Evolving Search · adu2021
    Dr. Zero: Self-Evolving Search Agents without Training Data. From arXiv:2601.07055
    0
    installs
  78. Efficient Machine Unlearning · adu2021
    Framework for efficient machine unlearning that reformulates forgetting as inverse learning. Achieves significant computational speedup by replacing expensive Hessian operations with gradient-based optimization, enabling privacy-preserving model updates.
    0
    installs
  79. Encoder Pretraining Strategy · adu2021
    Choose optimal pretraining strategy for text encoders: pure MLM, pure CLM, or biphasic CLM-then-MLM training, with empirical guidance on performance across downstream tasks.
    0
    installs
  80. Exevrm Video Reward Modeling · adu2021
    Build robust reward models for computer-use agents by processing execution videos with spatiotemporal token pruning (STP+TTP) to remove redundancy while preserving task-relevant UI details. Achieve 84.7% accuracy with adversarial instruction translation.
    0
    installs
  81. Flowblending Video Inference · adu2021
    Accelerate video generation by allocating smaller models to intermediate diffusion timesteps and larger models to capacity-critical early and late stages. Achieves 1.65x speedup and 57% FLOP reduction while maintaining visual quality. Use when video generation latency or computational cost is critical and you have multiple model sizes available.
    0
    installs
  82. Focus Agent Context Trimming · adu2021
    Use a lightweight LLM to filter accessibility tree observations by task relevance, reducing agent context size by 50-80% while maintaining equivalent task performance.
    0
    installs
  83. Env Scaler Synthesis · adu2021
    Automatically generate diverse, scalable tool-interactive training environments for LLM agents without manual sandbox creation. Uses topic mining and logic modeling to create varied environment architectures with task scenarios, enabling agents to learn complex multi-turn, multi-tool interactions. Synthesis framework tested on 191 environments with ~7,000 scenarios, improving Qwen3 model performance on knowledge-intensive and search tasks.
    0
    installs
  84. G2rl Gradient Guided · adu2021
    Guide LLM exploration through the model's own gradient geometry rather than external signals. Extract sequence-level gradient features measuring how tokens would reshape output distributions. Reward responses introducing novel gradient directions while deemphasizing redundant ones. Achieve orthogonal gradient directions and improved accuracy.
    0
    installs
  85. Illusion Of Thinking · adu2021
    Evaluate reasoning model capabilities by analyzing three complexity-dependent behavioral regimes and identifying fundamental limitations in symbolic manipulation rather than computational budgets.
    0
    installs
  86. Justrl Simple Recipe · adu2021
    Demonstrate that simple single-stage RL with fixed hyperparameters matches complex multi-stage approaches for training small LLMs on mathematical reasoning. Use basic setup: GRPO algorithm, rule-based verification, 16K token context, standard training data without difficulty filtering. Train two 1.5B models to competitive performance using 2× less compute than sophisticated approaches.
    0
    installs
  87. Kv Cache Compression · adu2021
    Enables 8x KV cache compression with minimal training overhead to improve reasoning accuracy by allowing more token generation within computational budgets.
    0
    installs
  88. Low Prob Exploration · adu2021
    Preserve reasoning diversity in RL by protecting low-probability tokens that represent novel thinking paths. Trigger: maintain exploration capability during RL training while avoiding entropy collapse on reasoning tasks.
    0
    installs
  89. Meta Aware Reasoning · adu2021
    Improve reasoning models by aligning their meta-predictions with actual rollouts through self-generated training signals. Trigger: accelerate reasoning model training while maintaining performance through better meta-cognitive awareness.
    0
    installs
  90. Omni Reward Modeling · adu2021
    Builds generalist reward models evaluating outputs across 5 modalities (text, image, video, audio, 3D) using free-form preference data. Combines discriminative and generative reward modeling approaches. Covers 9 tasks with 317K preference examples, enabling preference-based alignment for diverse output modalities.
    0
    installs
  91. Open Vision Reasoner · adu2021
    Transfer reasoning behaviors learned in language models to visual domains through two-stage training: cold-start linguistic fine-tuning followed by multimodal RL. Open-Vision-Reasoner achieves 95.3% on MATH500 and 54.6% on MathVerse by learning visual analogs of backtracking, verification, and subgoal decomposition using rule-based rewards.
    0
    installs
  92. Opv Process Verifier · adu2021
    Verify long reasoning chains by combining outcome and process verification through active learning. OPV achieves 83.1 F1 on verification—crucial when both final answers and reasoning paths must be validated with minimal annotation.
    0
    installs
  93. Pear Sft Preparation · adu2021
    Improve post-RL performance by reweighting SFT loss using importance sampling. Prioritize training examples that match the target policy distribution, not the behavior policy. Achieves 14.6% Pass@8 gains on AIME.
    0
    installs
  94. Peft Rlvr Evaluation · adu2021
    Comprehensively evaluate 12+ parameter-efficient fine-tuning methods for RL with Verifiable Rewards (RLVR). Show DoRA/AdaLoRA outperform LoRA, SVD-based methods fail on RL, extreme reduction creates bottlenecks—providing empirical evidence that geometric-aware adapters align better with RL's off-principal update dynamics.
    0
    installs
  95. Prophet Diffusion Lm · adu2021
    Detect when diffusion language models converge on correct answers before completing refinement steps using confidence gap monitoring, achieving 3.4x decoding speedup
    0
    installs
  96. Repro Process Reward · adu2021
    Dual-scoring mechanism (Magnitude and Stability scores) enabling process-level rewards that penalize inefficient reasoning patterns like overthinking and backtracking without trained reward models. Improve reasoning efficiency in RL without additional supervision.
    0
    installs
  97. Sage Self Hinting Rl · adu2021
    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 refresh mechanism.
    0
    installs
  98. Self Evolving Agents · adu2021
    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.
    0
    installs
  99. Smolvla Robotics Vla · adu2021
    Deploy compact vision-language-action models that run on consumer GPUs for natural language robot control.
    0
    installs
  100. Superwriter Longform · adu2021
    Generate coherent, consistent long-form text through structured planning, hierarchical reflection, and Monte Carlo tree search-guided optimization.
    0
    installs