adu2021

@adu2021 source repo

1.2k skills
0 followers
1 day ago last updated

▌
Neural Thickets Dense Task Experts · adu2021
Discover that large pretrained models have dense neighborhoods of task-specific experts—random weight perturbations improve performance. Use RandOpt: sample perturbations, select top performers, ensemble via voting for multi-task adaptation.

0
installs
▌
Parallel Probe Efficient Reasoning · adu2021
Monitor parallel reasoning branches via 2D probing—periodically extracting intermediate answers to detect consensus and prune divergent branches. Reduces token cost by 25%+ while maintaining accuracy without requiring model retraining.

0
installs
▌
Pvchat Personalized Video One Shot · adu2021
Enable identity-aware video question answering with one-shot learning using Mixture-of-Heads enhanced ViLLM. Learns subject-specific features from single video through synthetic augmentation and progressive image-to-video training, enabling recognition of individuals in medical, smart home, and entertainment contexts.

0
installs
▌
Re Bottleneck Latent Restructuring · adu2021
Restructure latent representations in pretrained audio autoencoders without full retraining. Apply three variants—ordered, semantic, and equivariant—to enforce structure like channel ordering, semantic alignment, or filter correspondence. Achieves 20-60% semantic gains in under 48 GPU hours versus 14.5K hours for full retraining.

0
installs
▌
Retroagent Dual Intrinsic Feedback · adu2021
Train LLM agents to evolve by generating dual intrinsic feedback signals—numerical rewards for capability progress and language lessons for future reuse. Combines hindsight reflection with memory-augmented RL using SimUtil-UCB retrieval.

0
installs
▌
Scaling Laws Optimal Data Mixtures · adu2021
Predict optimal data mixture proportions for multi-domain LLM training using scaling laws that require only 10-20 small experiments. Determine which domains should contribute how much data across model sizes (186M-7B), reducing computational waste in foundation model pretraining.

0
installs
▌
Sculptor Active Context Management · adu2021
Framework enabling LLMs to actively manage internal working memory and context through tools for fragmentation, summarization, and semantic search. Mitigates proactive interference and improves reasoning at extended context lengths.

0
installs
▌
Seagent Self Evolving Computer Use · adu2021
Vision-based computer use agent that self-improves through experiential learning, curriculum generation, and reward-based RL on diverse software.

0
installs
▌
Sensenovamars Multimodal Reasoning · adu2021
Build vision-language agents that seamlessly integrate visual reasoning with dynamic tool manipulation (search, cropping) through reinforcement learning, achieving state-of-the-art performance on fine-grained visual understanding tasks—surpassing proprietary models like Gemini and GPT.

0
installs
▌
Subgoal Driven Long Horizon Agents · adu2021
Improve long-horizon task success via subgoal decomposition and dense milestone-based rewards, dramatically outperforming sparse-reward RL and standard baselines.

0
installs
▌
Tape Tool Guided Adaptive Planning · adu2021
Improve LLM agents operating under strict feasibility constraints (budget limits, tool usage caps) by separating planning from execution. Generate multiple candidate plans, merge into plan graph, then use external solver (ILP) to find optimal feasible path. Constrained decoding forces execution of planned actions, eliminating sampling errors. Adaptive replanning handles observation surprises. Achieves 21+ pp improvements on constrained tasks vs. ReAct.

0
installs
▌
Think At Hard Selective Refinement · adu2021
Reduce reasoning model inference cost by selectively applying latent iterations—use a lightweight neural decider to identify hard tokens and refine only those with focused LoRA modules, achieving 94% token exemption with 8-11% accuracy gain.

0
installs
▌
Thinkdial Reasoning Effort Control · adu2021
Control LLM reasoning effort through discrete modes (High/Medium/Low) using budget-aware supervised fine-tuning and adaptive reward shaping, enabling compression-performance tradeoffs.

0
installs
▌
Toolrm Outcome Reward Tool Calling · adu2021
Specialized outcome reward models (1.7B-14B) for evaluating tool-calling performance in LLMs. Addresses the critical gap in reward modeling where general-purpose models miss key signals of effective tool use. Enables better Best-of-N sampling, data filtering, and RL-based policy training through FC-RewardBench evaluation framework.

0
installs
▌
Toucan Tool Agentic Data Synthesis · adu2021
Synthesize large-scale training data for LLM agents by generating diverse tool-use trajectories from real Model Context Protocol (MCP) environments. A 1.5M-example dataset created via multi-stage pipeline: query generation, quality filtering, trajectory creation with real tool execution, validation, and diversification.

0
installs
▌
Uncertainty Metric Control Signals · adu2021
Transform uncertainty quantification in LLMs from passive reliability measurement into active control signals for reasoning optimization, autonomous agent decision-making, and reinforcement learning. Use when building systems where uncertainty drives real-time behavior modification and improved reliability.

0
installs
▌
Versatile Controls Video Diffusion · adu2021
Enable flexible control over video diffusion models through multi-modal control signals (edges, masks, poses) without retraining. Apply lightweight Transformer-based auxiliary modules to add Canny edge, segmentation, and pose constraints to frozen pre-trained generators.

0
installs
▌
Vision Language Reasoning Transfer · adu2021
Transfer reasoning capabilities from text LLMs to visual domains using reinforcement learning, achieving human-level visual reasoning on complex benchmarks. 38B parameters match closed-source VLMs by optimizing cross-modal connector alignment and entropy-based reasoning signals.

0
installs
▌
F4splat Feed Forward Densification · adu2021
Enable spatially adaptive Gaussian allocation in 3D Gaussian Splatting without iterative optimization. Reduces primitive count by 70-90% while maintaining reconstruction quality through learned densification score prediction.

0
installs
▌
Pearl Personalized Streaming Video · adu2021
Defines Personalized Streaming Video Understanding (PSVU) as a new task bridging static image personalization and video intelligence. PEARL-Bench (132 videos, 2173 annotations) establishes founding experiments. PEARL framework provides training-free plug-and-play strategy using dual-grained memory (concept + streaming) and concept-aware retrieval. Trigger: When building AI assistants that interact with streaming video and personalize on user-defined concepts, apply this problem definition and framework to support real-time multi-turn reasoning.

0
installs
▌
Pepo Token Level Multimodal Policy · adu2021
Replace uniform token-level advantages with perception-exploration gating that weights tokens by visual grounding strength. Adds 3.67 points to geometry reasoning and 5.32 to few-shot classification with <1% compute overhead. Works best for multimodal CoT where visual grounding anchors reasoning. Trigger: When using token-level RL on VLMs and want to emphasize visually-grounded reasoning steps.

0
installs
▌
Pivotrl Efficient Agentic Training · adu2021
Achieve high-accuracy agentic post-training with 4x fewer rollout turns and 5.5x less wall-clock time than end-to-end RL. Identify informative intermediate turns via pivot filtering and use verifier-based functional rewards for precise credit assignment.

0
installs
▌
Alignguard Lora Safety Preservation · adu2021
Preserve LLM safety alignment during LoRA fine-tuning via Fisher information regularization and collision-aware geometric constraints.

0
installs
▌
Chain Of Mindset Adaptive Reasoning · adu2021
Enable language models to dynamically switch between four cognitive modes (spatial, convergent, divergent, algorithmic) during problem-solving. Meta-agent observes state and selects optimal mode per step, improving reasoning across math, coding, and spatial tasks without requiring model training.

0
installs
▌
Cola Test Time Depth Adaptation LLM · adu2021
Dynamically adapt LLM depth per input at test time by skipping, repeating, or reordering layers using MCTS search, correcting 60% of initially wrong predictions and processing 75% of correct predictions with shorter architectures without retraining.

0
installs
▌
Cov Chain Of View Spatial Reasoning · adu2021
Enable vision-language models to perform embodied question answering in 3D environments through active camera exploration. CoV uses training-free test-time reasoning to iteratively select relevant viewpoints and adjust camera angles until sufficient context is gathered, achieving 11-13% accuracy improvements across spatial reasoning benchmarks.

0
installs
▌
Dc Video Gen Compression Adaptation · adu2021
Accelerate video generation by 14.8x through deep compression autoencoder (32x-64x spatial, 4x temporal compression) combined with lightweight adapter-based model adaptation. Use when deploying video diffusion models under compute or latency constraints.

0
installs
▌
Deer Diffusion Speculative Decoding · adu2021
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 proposals. Achieve 5.54× speedup on HumanEval vs. 2.41× for AR-based methods.

0
installs
▌
Discrete Diffusion Faster Inference · adu2021
Enables diffusion LLMs to achieve 2.5× faster inference than autoregressive models through block-wise generation with parallel inter-block decoding.

0
installs
▌
Evolution Strategies LLM Finetuning · adu2021
Scale Evolution Strategies to billion-parameter LLMs without backpropagation for superior robustness and stability across diverse models, reward horizons, and evaluation tasks. Outperforms RL methods while eliminating gradient computation overhead.

0
installs
▌
Experiential Reinforcement Learning · adu2021
Improve RL training efficiency by embedding explicit reflection and revision loops. Models generate initial responses, receive feedback, produce self-reflections describing improvements, revise their attempts, and distill successful corrections into the base policy. Achieves up to 81% improvement on complex tasks through structured behavioral change.

0
installs
▌
Flowrl Reward Distribution Matching · adu2021
Train LLMs with distribution-matching rewards instead of reward maximization to achieve 10% improvement on math reasoning while improving solution diversity by matching the full reward distribution via flow balancing, addressing mode collapse in long chain-of-thought tasks.

0
installs
▌
Fusionroute Token LLM Collaboration · adu2021
Combine multiple specialized language models at token-level granularity without joint training or architectural compatibility. FusionRoute performs expert selection and generates complementary logits to overcome routing limitations, achieving superior cross-domain performance at inference time.

0
installs
▌
Genie Envisioner Robotic Foundation · adu2021
Unified platform combining instruction-conditioned video diffusion, flow-matching action decoder, and action-conditioned simulator. Enables scalable robot learning without extensive labeled demonstrations.

0
installs
▌
Geometry Forcing Video Diffusion 3d · adu2021
Improve video diffusion consistency by aligning intermediate diffusion features with 3D geometric representations from pretrained foundation models, enabling spatially coherent and temporally stable video generation through angular and scale alignment losses.

0
installs
▌
Learning On Job Self Evolving Agent · adu2021
Build autonomous agents that accumulate structured knowledge from task execution into hierarchical memory (strategic, procedural, tool) without human annotation, enabling knowledge transfer to unseen tasks.

0
installs
▌
Longvie Multimodal Video Generation · adu2021
Generate consistent ultra-long videos (up to one minute) using depth and trajectory controls with autoregressive segment generation and unified noise initialization.

0
installs
▌
Memskill Evolving Memory Operations · adu2021
Build agents that evolve their own memory operations by learning a skill bank of memory transformations and periodically discovering new skills from challenging cases, enabling adaptive memory management that improves with scale.

0
installs
▌
Muon Optimizer Tail Memory Learning · adu2021
Improve LLM training efficiency by selectively applying Muon optimizer to Value-Output attention weights and FFN layers, which function as associative memories. Use when training data exhibits heavy-tailed distributions requiring robust rare-fact learning.

0
installs
▌
Pi Rl Flow Matching Vla Fine Tuning · adu2021
Apply reinforcement learning to flow-based VLA models by modeling denoising as an MDP and converting ODEs to SDEs, enabling efficient policy optimization for robotic control without expensive supervised fine-tuning data collection.

0
installs
▌
Policy Discriminative Reward Models · adu2021
Learn generalizable reward models via unsupervised policy discrimination: pretraining models to distinguish between different policies enables efficient adaptation to human preferences and strong RLHF performance.

0
installs
▌
Re Align Reasoning Image Generation · adu2021
Bridge image understanding-generation gap via In-Context Chain-of-Thought reasoning and RL training with surrogate rewards. Improve faithful execution of mixed image-text prompts in generation and editing tasks.

0
installs
▌
Robomemory Multi Memory Embodied AI · adu2021
Brain-inspired framework integrating spatial, temporal, episodic, and semantic memory systems for embodied agents. Achieves 26.5% performance improvement via dynamic spatial knowledge graphs and closed-loop planning with critic modules.

0
installs
▌
Sliding Window Attention Adaptation · adu2021
Adapt full-attention language models to sliding window attention without expensive retraining. Combine five synergistic strategies (full decode, sink tokens, interleaved layers, chain-of-thought, fine-tuning) achieving 30-100% speedups while maintaining 90-100% accuracy.

0
installs
▌
Temporal In Context Video Diffusion · adu2021
Adapt pretrained video diffusion models to conditional tasks using only 10-30 samples without architectural changes.

0
installs
▌
Token Sparse Attention Long Context · adu2021
Dynamically select important tokens at the attention head level, performing dense attention only on selected tokens and scattering results back. Achieves 3.23x attention speedup at 128K context with 1% accuracy loss through layer-wise representation stability analysis.

0
installs
▌
Training Free Group Relative Policy · adu2021
Improve LLM performance at test time through in-context learning and experience libraries, eliminating the need for parameter updates while maintaining competitive results.

0
installs
▌
Veriguard Agent Safety Verification · adu2021
Generate safety policies as executable code with formal verification, enabling provably-correct agent behavior through offline policy generation and online runtime enforcement.

0
installs
▌
Vision Language Vision Distillation · adu2021
Build efficient vision-language models by distilling knowledge from frozen diffusion decoders and vision encoders. Achieve GPT-4o-level captioning with <$1000 training cost by leveraging pre-trained components. Use when you need high-quality vision-language understanding without expensive end-to-end training.

0
installs
▌
Watch And Learn Computer Use Videos · adu2021
Convert internet tutorial videos into executable UI action trajectories using inverse dynamics models trained on 600K synthetic screen transitions. Generate 53K high-quality demonstrations spanning 69 applications, achieving state-of-the-art 7B agent performance on WindowsAgentArena via both in-context learning and supervised fine-tuning.

0
installs
▌
World Models Calibrated Uncertainty · adu2021
Quantify confidence in video generation by estimating latent-space uncertainty and mapping to interpretable heatmaps. Detect untrustworthy regions in generated frames for robot planning and policy evaluation—critical when video hallucinations could cause real-world failures.

0
installs
▌
Chanrg Rna Structure Generalization · adu2021
Overturn the assumption that scaling foundation models improves RNA structure prediction by understanding why they fail out-of-distribution. Includes structure-aware deduplication revealing 33-fold residual redundancy in prior benchmarks, out-of-distribution test regimes (GenA, GenC, GenF), and root cause analysis showing coverage and wiring failures. Foundation models achieving 67.3% on held-out test drop to 18.0% OOD (26.7% retention), while structured decoders retain 92.3%, enabling practitioners to recognize when scaling fails and when inductive structure matters.

0
installs
▌
Group3d Semantic Grouping Detection · adu2021
Enforce semantic compatibility constraints directly into instance construction for 3D object detection. Uses MLLM-driven semantic grouping to partition object vocabulary into plausible cross-view category equivalence clusters, preventing geometry-driven over-merging.

0
installs
▌
Ovie Monocular Novel View Synthesis · adu2021
A single insight eliminates multi-view requirements for novel-view synthesis: monocular depth acts as a training-time geometric scaffold to generate synthetic view pairs from unpaired internet images, but can be discarded at inference. This reframes the problem from needing paired multi-view data to leveraging abundant 2D internet imagery. Trigger: When limited to monocular video or single-image novel-view synthesis, use depth as training scaffold on unpaired data—the model learns geometry without needing it at inference.

0
installs
▌
Roboalign Language Action Reasoning · adu2021
Bridge the modality gap between language reasoning and low-level robot actions through two-stage alignment training. Use RL with action-accuracy rewards to ground linguistic reasoning in precise motor control, achieving 17.5%-106.6% improvements over SFT baselines.

0
installs
▌
Session Risk Memory Temporal Safety · adu2021
Add trajectory-level temporal authorization to stateless execution gates by maintaining a semantic centroid of agent behavioral profiles with exponential moving average risk accumulation. Improves F1 from 0.9756 to 1.0 and reduces false positives from 5% to 0% on slow-burn security violations. Use when deploying multi-turn agents and need to detect gradual privilege escalation and data exfiltration.

0
installs
▌
Sparse Critical Rlvr Token Analysis · adu2021
Analyzes how Reinforcement Learning from Verification Rewards (RLVR) improves reasoning by examining token-level probability distributions. Finds that >83% of token positions exhibit near-zero divergence—RL operates through sparse, targeted refinements. Cross-sampling experiments show 1.5-7.8% RL-selected tokens recover full gains, while reverting 5-10% of RL tokens collapses performance. Reveals that RL primarily reallocates probability within existing candidates (80% overlap in top-k tokens), not inventing novel tokens. Trigger: When analyzing LLM reasoning improvements, apply token-level divergence analysis and cross-sampling to identify which positions drive gains and whether changes are sparse or distributed.

0
installs
▌
Thinkjepa Dual Temporal World Model · adu2021
Replace single-pathway JEPA with a dual-temporal architecture combining dense frame sampling (fine-grained dynamics) and uniformly-sampled VLM guidance (semantic coherence) to improve egocentric trajectory prediction by 14-27% on ADE/accuracy metrics. Effective when predicting hand-object interactions where both low-level dynamics and high-level semantic context matter, and long-horizon predictions benefit from hierarchical visual representations.

0
installs
▌
Derf Normalization Free Transformers · adu2021
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.

0
installs
▌
Dissecting Tool Integrated Reasoning · adu2021
Evaluate and optimize tool-integrated reasoning in LLMs through empirical benchmarking, performance-cost metrics (PAC, AUC-PCC), and measurement frameworks for diverse reasoning tasks.

0
installs
▌
Evolving Programmatic Skill Networks · adu2021
System for evolving and composing programmatic skills through learning, enabling agents to discover and develop new behavioral capabilities dynamically.

0
installs
▌
Flexibility Trap Diffusion Reasoning · adu2021
Understand how token generation flexibility in diffusion LMs paradoxically constrains reasoning, as models exploit ordering flexibility to avoid uncertain tokens, and apply simplified approaches that preserve parallel decoding benefits. Use when optimizing diffusion-based language models for reasoning tasks.

0
installs
▌
Freemorph Tuning Free Image Morphing · adu2021
Generate smooth morphing sequences between images without fine-tuning or alignment. Uses guidance-aware spherical interpolation and step-oriented attention blending to handle diverse semantic and layout scenarios, completing morphs 50× faster than fine-tuning methods.

0
installs
▌
Goedel Prover Formal Theorem Proving · adu2021
Train language models for formal theorem proving via expert iteration with verifier-guided self-correction and checkpoint merging.

0
installs
▌
Graph Optimization Test Time Compute · adu2021
Optimize test-time computation through graph-based collaborative architecture where nodes represent models, edges represent information flow, and topology itself is optimizable via reinforcement learning to discover ideal model assignments and configurations.

0
installs
▌
Ifdecorator Instruction Following Rl · adu2021
Enhances RLVR through cooperative-adversarial flywheel, intent verification, and trap instructions. Detects reward hacking and improves training efficiency, achieving 87.43% on IFEval.

0
installs
▌
Industrial Defect Multimodal Dataset · adu2021
Introduce IMDD-1M dataset with 1M aligned image-text pairs spanning 60+ material categories and 400+ defect types. Train diffusion-based vision-language models requiring 5% task-specific data vs. dedicated expert models for manufacturing quality control.

0
installs
▌
Matchtir Fine Grained Tool Reasoning · adu2021
Improves tool-integrated reasoning by using bipartite matching to assign dense turn-level rewards, enabling credit assignment for individual tool interactions in multi-turn tasks where 4B models outperform 8B competitors.

0
installs
▌
Medical Multimodal Foundation Models · adu2021
Build efficient medical vision-language models that reason about medical images and text simultaneously, achieving competitive performance with much larger models while maintaining 500× lower computational cost.

0
installs
▌
Mixture Of Experts Adaptive Capacity · adu2021
Optimize Mixture-of-Experts efficiency by decoupling token-level expert activation from layer architecture. Use dynamic threshold routing where expert count per token varies by input complexity, and apply layer-wise capacity scheduling to match representational diversity patterns.

0
installs
▌
Multimodal Video Document Embeddings · adu2021
Generate unified embeddings for videos, images, and visual documents enabling semantic similarity, retrieval, and clustering across heterogeneous visual content types.

0
installs
▌
Nemotron Elastic Efficient Reasoning · adu2021
Deploy multiple reasoning model sizes efficiently by embedding nested submodels within a single parent—use end-to-end trained routers to select submodels at inference, achieving 360× cost reduction vs training families separately.

0
installs
▌
Os Sentinel Safety Mobile Gui Agents · adu2021
Detect safety violations in mobile GUI agents using dual mechanisms: formal verification rules for system-level risks (privilege escalation, credential exposure) and VLM-based contextual judges evaluating action appropriateness, achieving 10-30% safety improvement over baselines.

0
installs
▌
Pairwise Self Verification Reasoning · adu2021
Verify solution quality through pairwise comparison rather than pointwise scoring. Implement topology coverage and Swiss refinement to allocate verification compute to uncertain pairs, improving calibration and reducing verification overhead.

0
installs
▌
Perception Aware Policy Optimization · adu2021
Optimize multimodal LLMs by directly targeting perception errors using KL-divergence based perception loss, improving visual reasoning by 8-19% on vision-dependent tasks. Integrates perception-aware signals into policy gradients without relying solely on reward modifications.

0
installs
▌
Perk Long Context Test Time Learning · adu2021
Enable language models to reason over extremely long contexts (128K tokens) by encoding context into lightweight LoRA adapters during test time, achieving 20% performance improvements without full model retraining.

0
installs
▌
Pets Trajectory Allocation Framework · adu2021
Optimize sampling budget allocation for self-consistency inference by treating trajectory allocation as a crowdsourcing problem. Introduce self-consistency rate (agreement with infinite-budget consensus) as optimization target. Offline setting uses Optimistic Knowledge Gradient for Bayesian optimization; online setting uses difficulty grid discretization and greedy allocation. Achieves up to 75% budget reduction (offline) and 55% (online) while maintaining accuracy.

0
installs
▌
Pretraining Midtraining Rl Interplay · adu2021
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 training.

0
installs
▌
Reinforcement Learning Vision Models · adu2021
Train multimodal LLMs with RL (PIVOT) instead of SFT to produce stronger, precisely-localized visual representations in vision encoders using <1% of standard pretraining cost.

0
installs
▌
Sequential Diffusion Language Models · adu2021
Implement adaptive parallel decoding for language models using diffusion-based next-sequence prediction. Enable dynamic block-based token generation with confidence thresholds to achieve 2x+ speedups while maintaining competitive performance. Retrofit existing autoregressive models with minimal additional training data.

0
installs
▌
Action Quantization Behavior Cloning · adu2021
Establish regret bounds for behavior cloning with discretized actions combining statistical error and quantization error terms. Prove smoothness requirements for safe quantizer design, show that learning-based quantizers fail these requirements, and propose model-based augmentation to reduce error dependence from H² to H.

0
installs
▌
Camouflage Attacks Vehicle Detection · adu2021
Craft stealthy vehicle appearances that fool detectors by formulating attacks as conditional image editing. Apply image-level stylization (match surroundings) and scene-level strategies (match semantic concepts), achieve 38% AP50 reduction with 85%+ human perceptual success, and transfer to black-box detectors.

0
installs
▌
Foreagent Predict Execute · adu2021
Replace expensive test-based execution loops with learned prediction models that forecast agent action outcomes before commitment. Framework uses internalized execution priors and structured analysis reports to achieve 6x faster convergence and 6% higher performance compared to execute-first baselines. Applicable to scientific discovery, optimization, and hypothesis testing where verification is computationally or financially expensive.

0
installs
▌
Gnosis LLM Self Awareness · adu2021
Enable frozen LLMs to predict their own correctness by decoding signals from internal hidden states and attention patterns, achieving reliable self-verification without external judges—adding only 5M parameters while reducing inference cost and improving calibration.

0
installs
▌
Gorl Generative Online Rl · adu2021
Separates generative policy optimization through latent encoder (standard RL algorithms) and conditional decoder (frozen then refined), using two-timescale alternating schedule to eliminate gradient instability from direct generative policy optimization.

0
installs
▌
Learned 4bit Quantization · adu2021
Compress LLM weights to 4-bit precision using learned per-row lookup tables that minimize output activation error, achieving better accuracy than fixed formats (int4, fp4, nf4) while maintaining inference speed.

0
installs
▌
Light If Preview Checking · adu2021
Multi-stage training approach using entropy-preserving SFT and token-wise entropy-adaptive RL to improve instruction adherence. Combines data curation with reward-guided reasoning, outperforming larger models on IFEval.

0
installs
▌
LLM Local Linear Mappings · adu2021
Interpret LLM behavior as locally linear mappings between hidden representations, enabling mechanistic understanding of computation without examining individual weights or attention patterns.

0
installs
▌
Long Context Rl Reasoning · adu2021
Train LLMs for long-context reasoning using KeyChain synthesis: convert short multi-hop QA into long-context tasks by embedding UUID chains in distractor documents, enabling 16K→128K generalization.

0
installs
▌
Lsrif Logic Structured Rl · adu2021
Improves instruction-following by using differentiated reward mechanisms that recognize logical dependencies (sequential, conditional, parallel) in instructions, enabling better model reasoning about task structure.

0
installs
▌
Memory LLM Ffn Decoupling · adu2021
Decouple feed-forward networks from self-attention by training FFNs on context-free token embeddings instead of residual streams. Enables pre-computation of FFN outputs as static lookup tables for inference efficiency and improved interpretability.

0
installs
▌
Mint Cot Visual Reasoning · adu2021
Integrates fine-grained visual tokens into mathematical reasoning via Interleave Tokens that dynamically select relevant image regions for each reasoning step.

0
installs
▌
Mmgr Multimodal Reasoning · adu2021
Evaluate whether generative models possess foundational reasoning capabilities. Develop five-ability framework (physical, logical, 3D spatial, 2D spatial, temporal reasoning) across abstract reasoning, embodied navigation, and physical commonsense benchmarks. Use structured rubric requiring simultaneous satisfaction of all sub-metrics.

0
installs
▌
Multi Agent Memory System · adu2021
Build persistent, structured memory systems for LLM agents that remember user context across sessions, organize information semantically, and retrieve relevant knowledge automatically before responding. Achieves 35% accuracy gains over RAG baselines with 99.9% smaller storage overhead.

0
installs
▌
Openthoughts Data Recipes · adu2021
Design data generation pipelines for reasoning models through systematic experimentation with answer sampling, teacher selection, and source quality optimization.

0
installs
▌
Orion Language Of Thought · adu2021
Compact Mentalese symbolic format trained via SFT, then refined with SLPO (Shorter Length Preference Optimization) to reward concise correct solutions without over-penalizing necessarily long reasoning. Compress reasoning while maintaining accuracy.

0
installs
▌
Pacore Parallel Reasoning · adu2021
PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning. From arXiv:2601.05593

0
installs
▌
Pensieve Stateful Context · adu2021
Enable language models to actively manage their context using memory tools (read, index, note-take, delete). Models receive agency to self-engineer context rather than passively consume pre-staged information, maintaining compact high-quality reasoning states through strategic deletion of irrelevant information while preserving distilled notes.

0
installs
▌
Poss Speculative Decoding · adu2021
Improve speculative decoding throughput by employing position-specialized draft layers that handle position-specific error accumulation patterns.

0
installs
▌
Proactive Self Refinement · adu2021
Enable models to refine outputs dynamically during generation based on internal signals, reducing token consumption by 41.6% while improving accuracy by 8.2%.

0
installs

1228 published skills · page 3 of 13