adu2021
- 1.2k skills
- 0 followers
- 17 hours ago last updated
- ▌ Finally Outshining The Random Baseline A Simple · adu2021Active learning (AL) has the potential to drastically reduce annotation costs in 3D biomedical image segmentation, where expert labeling of volumetric data is both time-consuming and expensive. Yet, existing AL methods are unable to consistently outperform improved random sampling baselines adapted to 3D data, leaving the field without a reliable solution. We introduce Class-stratified Scheduled Power Predictive Entropy (ClaSP PE), a simple and effective query strategy that addresses two key lim...
- ▌ Finvault Benchmarking Financial Agent Safety In · adu2021Financial agents powered by large language models (LLMs) are increasingly deployed for investment analysis, risk assessment, and automated decision-making, where their abilities to plan, invoke tools, and manipulate mutable state introduce new security risks in high-stakes and highly regulated financial environments. However, existing safety evaluations largely focus on language-model-level content compliance or abstract agent settings, failing to capture execution-grounded risks arising from re...
- ▌ Locality Aware Parallel Autoregressive Decoding · adu2021Accelerate image generation by generating multiple patches in parallel instead of sequentially. Uses locality-aware ordering that generates spatially-close tokens while keeping concurrent tokens far apart, reducing steps from 256 to 20 (12× speedup) while maintaining quality.
- ▌ Mmdeepresearch Bench A Benchmark For Multimodal · adu2021Deep Research Agents (DRAs) generate citation-rich reports via multi-step search and synthesis, yet existing benchmarks mainly target text-only settings or short-form multimodal QA, missing end-to-end multimodal evidence use. We introduce MMDeepResearch-Bench (MMDR-Bench), a benchmark of 140 expert-crafted tasks across 21 domains, where each task provides an image-text bundle to evaluate multimodal understanding and citation-grounded report generation. Compared to prior setups, MMDR-Bench emphas...
- ▌ Towards Efficient And Robust Linguistic Emotion · adu2021Linguistic expressions of emotions such as depression, anxiety, and trauma-related states are pervasive in clinical notes, counseling dialogues, and online mental health communities, and accurate recognition of these emotions is essential for clinical triage, risk assessment, and timely intervention. Although large language models (LLMs) have demonstrated strong generalization ability in emotion analysis tasks, their diagnostic reliability in high-stakes, context-intensive medical settings remai...
- ▌ Typhoon Ocr Open Vision Language Model For Thai · adu2021Document extraction is a core component of digital workflows, yet existing vision-language models (VLMs) predominantly favor high-resource languages. Thai presents additional challenges due to script complexity from non-latin letters, the absence of explicit word boundaries, and the prevalence of highly unstructured real-world documents, limiting the effectiveness of current open-source models. This paper presents Typhoon OCR, an open VLM for document extraction tailored for Thai and English. Th...
- ▌ A Bertology View Of LLM Orchestrations Token And · adu2021Production LLM systems often rely on separate models for safety and other classification-heavy steps, increasing latency, VRAM footprint, and operational complexity. We instead reuse computation already paid for by the serving LLM: we train lightweight probes on its hidden states and predict labels in the same forward pass used for generation. We frame classification as representation selection over the full token-layer hidden-state tensor, rather than committing to a fixed token or fixed layer ...
- ▌ Abc Bench Benchmarking Agentic Backend Coding In · adu2021The evolution of Large Language Models (LLMs) into autonomous agents has expanded the scope of AI coding from localized code generation to complex, repository-level, and execution-driven problem solving. However, current benchmarks predominantly evaluate code logic in static contexts, neglecting the dynamic, full-process requirements of real-world engineering, particularly in backend development which demands rigorous environment configuration and service deployment. To address this gap, we intr...
- ▌ Kage Bench Fast Known Axis Visual Generalization · adu2021Pixel-based reinforcement learning agents often fail under purely visual distribution shift even when latent dynamics and rewards are unchanged, but existing benchmarks entangle multiple sources of shift and hinder systematic analysis. We introduce KAGE-Env, a JAX-native 2D platformer that factorizes the observation process into independently controllable visual axes while keeping the underlying control problem fixed. By construction, varying a visual axis affects performance only through the in...
- ▌ Memoryrewardbench Benchmarking Reward Models For · adu2021Existing works increasingly adopt memory-centric mechanisms to process long contexts in a segment manner, and effective memory management is one of the key capabilities that enables large language models to effectively propagate information across the entire sequence. Therefore, leveraging reward models (RMs) to automatically and reliably evaluate memory quality is critical. In this work, we introduce MemoryRewardBench, the first benchmark to systematically study the ability of RMs to evaluate l...
- ▌ Rethinking Thinking Tokens Improvement Operators · adu2021Enable longer reasoning within bounded context through iterative refinement: generate solution → verify → compress → refine. Achieves +11% on AIME with lower per-call latency than naive chain-of-thought.
- ▌ Sin Bench Tracing Native Evidence Chains In Long · adu2021Evaluating whether multimodal large language models truly understand long-form scientific papers remains challenging: answer-only metrics and synthetic 'Needle-In-A-Haystack' tests often reward answer matching without requiring a causal, evidence-linked reasoning trace in the document. We propose the 'Fish-in-the-Ocean' (FITO) paradigm, which requires models to construct explicit cross-modal evidence chains within native scientific documents. To operationalize FITO, we build SIN-Data, a scientif...
- ▌ The Responsibility Vacuum Organizational Failure · adu2021Modern CI/CD pipelines integrating agent-generated code exhibit a structural failure in responsibility attribution. Decisions are executed through formally correct approval processes, yet no entity possesses both the authority to approve those decisions and the epistemic capacity to meaningfully understand their basis. We define this condition as responsibility vacuum: a state in which decisions occur, but responsibility cannot be attributed because authority and verification capacity do not coi...
- ▌ Toward Efficient Agents Memory Tool Learning And · adu2021Recent years have witnessed increasing interest in extending large language models into agentic systems. While the effectiveness of agents has continued to improve, efficiency, which is crucial for real-world deployment, has often been overlooked. This paper therefore investigates efficiency from three core components of agents: memory, tool learning, and planning, considering costs such as latency, tokens, steps, etc. Aimed at conducting comprehensive research addressing the efficiency of the a...
- ▌ Agentic R Learning To Retrieve For Agentic Search · adu2021Agentic search has recently emerged as a powerful paradigm, where an agent interleaves multi-step reasoning with on-demand retrieval to solve complex questions. Despite its success, how to design a retriever for agentic search remains largely underexplored. Existing search agents typically rely on similarity-based retrievers, while similar passages are not always useful for final answer generation. In this paper, we propose a novel retriever training framework tailored for agentic search. Unlike...
- ▌ Cogflow Bridging Perception And Reasoning Through · adu2021Advanced reasoning approach for optimizing inference efficiency through meta-cognitive planning, enabling agents to make better decisions with reduced computational overhead.
- ▌ Digital Twin AI Opportunities And Challenges From · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Doc Pp Document Policy Preservation Benchmark For · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Gametalk Training Llms For Strategic Conversation · adu2021Implement techniques from GameTalk: Training LLMs for Strategic Conversation. Strategic decision-making in multi-agent settings is a key challenge for large language models (LLMs), particularly when coordination and negotiation must unfold over extended conversations
- ▌ Infinidepth Arbitrary Resolution And Fine Grained · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Mdagent2 Large Language Model For Code Generation · adu2021Generative approach for creating complex dynamic scenes and content, supporting agent capabilities in understanding and reasoning about multi-agent environments.
- ▌ Roi Reasoning Rational Optimization For Inference · adu2021Advanced reasoning approach for optimizing inference efficiency through meta-cognitive planning, enabling agents to make better decisions with reduced computational overhead.
- ▌ Unicorn Towards Self Improving Unified Multimodal · adu2021Advanced computer vision technique for robust spatial understanding in complex scenes, supporting agent navigation and environmental reasoning capabilities.
- ▌ Aceff A State Of The Art Machine Learning Potentia · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Agentic Rubrics As Contextual Verifiers For Swe Ag · adu2021Framework for enabling autonomous agents to self-verify code generation and reasoning quality through structured evaluation, supporting software engineering agent deployment with built-in correctness checking mechanisms.
- ▌ Atlas Orchestrating Heterogeneous Models And Tools · adu2021Adaptive framework for dynamically selecting optimal model-tool combinations in multi-domain reasoning, using cluster-based routing and reinforcement learning for improved agent reasoning across diverse tasks.
- ▌ Benchmark2 Systematic Evaluation Of LLM Benchmarks · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Can Llms Clean Up Your Mess A Survey Of Applicatio · adu2021Implement techniques from Can LLMs Clean Up Your Mess? A Survey of Application-Ready Data Preparation with LLMs. Data preparation aims to denoise raw datasets, uncover cross-dataset relationships, and extract valuable insights from them, which is essential for a wide range of data-centric applications
- ▌ Cgpt Cluster Guided Partial Tables With LLM Genera · adu2021Implement techniques from CGPT: Cluster-Guided Partial Tables with LLM-Generated Supervision for Table Retrieval. General-purpose embedding models have demonstrated strong performance in text retrieval but remain suboptimal for table retrieval, where highly structured content leads to semantic compression and query-table mismatch
- ▌ Chartverse Scaling Chart Reasoning Via Reliable Pr · adu2021Implement techniques from ChartVerse: Scaling Chart Reasoning via Reliable Programmatic Synthesis from Scratch. Chart reasoning is a critical capability for Vision Language Models (VLMs)
- ▌ Cooperbench Why Coding Agents Cannot Be Your Teamm · adu2021Implement techniques from CooperBench: Why Coding Agents Cannot be Your Teammates Yet. Resolving team conflicts requires not only task-specific competence, but also social intelligence to find common ground and build consensus
- ▌ Dreamstyle A Unified Framework For Video Stylizati · adu2021Unified system for multi-task video generation combining audio and visual synthesis, demonstrating scalable approaches for content generation that can enhance agent communication capabilities.
- ▌ Drpg Decompose Retrieve Plan Generate An Agentic F · adu2021Implement techniques from DRPG (Decompose, Retrieve, Plan, Generate): An Agentic Framework for Academic Rebuttal. Despite the growing adoption of large language models (LLMs) in scientific research workflows, automated support for academic rebuttal, a crucial step in academic communication and peer review, remains largely underexplored
- ▌ Dsgym A Holistic Framework For Evaluating And Trai · adu2021Implement techniques from DSGym: A Holistic Framework for Evaluating and Training Data Science Agents. Data science agents promise to accelerate discovery and insight-generation by turning data into executable analyses and findings
- ▌ E Grpo High Entropy Steps Drive Effective Reinforc · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Endless Terminals Scaling Rl Environments For Term · adu2021Implement techniques from Endless Terminals: Scaling RL Environments for Terminal Agents. Environments are the bottleneck for self-improving agents
- ▌ Enhancing Linguistic Competence Of Language Models · adu2021Enhanced language model pre-training methodology improving linguistic competence across languages, strengthening foundational capabilities for multilingual agent systems.
- ▌ Entropy Adaptive Fine Tuning Resolving Confident C · adu2021Technique for efficient model adaptation that mitigates catastrophic forgetting during fine-tuning, enabling agents to learn new tasks while preserving existing capabilities.
- ▌ Epiqal Benchmarking Large Language Models In Epide · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Exposeanyone Personalized Audio To Expression Diff · adu2021Multimodal system for joint audio-video understanding and generation, enhancing agent perception and communication in environments with both audio and visual information.
- ▌ Ffp 300k Scaling First Frame Propagation For Gener · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Gen3r 3d Scene Generation Meets Feed Forward Recon · adu2021Generative approach for creating complex dynamic scenes and content, supporting agent capabilities in understanding and reasoning about multi-agent environments.
- ▌ Guidelines To Prompt Large Language Models For Cod · adu2021Implement techniques from Guidelines to Prompt Large Language Models for Code Generation: An Empirical Characterization. Large Language Models (LLMs) are nowadays extensively used for various types of software engineering tasks, primarily code generation
- ▌ Inference Time Scaling Of Verification Self Evolvi · adu2021Implement techniques from Inference-Time Scaling of Verification: Self-Evolving Deep Research Agents via Test-Time Rubric-Guided Verification. While the majority of existing efforts focus on enhancing policy capabilities via post-training, we propose an alternative paradigm: self-evolving the agent's ability by iteratively verifying the policy model's outputs, guided by meticulously crafted rubrics
- ▌ Jet Rl Enabling On Policy Fp8 Reinforcement Learni · adu2021Implement techniques from Jet-RL: Enabling On-Policy FP8 Reinforcement Learning with Unified Training and Rollout Precision Flow. Reinforcement learning (RL) is essential for enhancing the complex reasoning capabilities of large language models (LLMs)
- ▌ Klear Unified Multi Task Audio Video Joint Generat · adu2021Generative approach for creating complex dynamic scenes and content, supporting agent capabilities in understanding and reasoning about multi-agent environments.
- ▌ Large Reasoning Models Are Not Yet Multilingual La · adu2021Advanced reasoning approach for optimizing inference efficiency through meta-cognitive planning, enabling agents to make better decisions with reduced computational overhead.
- ▌ Ltx 2 Efficient Joint Audio Visual Foundation Mode · adu2021Multimodal system for joint audio-video understanding and generation, enhancing agent perception and communication in environments with both audio and visual information.
- ▌ Magma A Multi Graph Based Agentic Memory Architect · adu2021Memory management architecture using multi-graph representations for autonomous agents, enabling efficient knowledge organization, contextual retrieval, and dynamic memory expansion to support complex agent decision-making.
- ▌ Mecellem Models Turkish Models Trained From Scratc · adu2021Implement techniques from Mecellem Models: Turkish Models Trained from Scratch and Continually Pre-trained for the Legal Domain. This paper presents Mecellem models, a framework for developing specialized language models for the Turkish legal domain through domain adaptation strategies
- ▌ Mechanistic Interpretability Of Large Scale Counti · adu2021Analysis framework for understanding how large language models perform complex cognitive tasks, revealing internal reasoning mechanisms that inform agent architecture and capability assessment.
- ▌ Meeplelm A Virtual Playtester Simulating Diverse S · adu2021Implement techniques from MeepleLM: A Virtual Playtester Simulating Diverse Subjective Experiences. Recent advancements have expanded the role of Large Language Models in board games from playing agents to creative co-designers
- ▌ Mindwatcher Toward Smarter Multimodal Tool Integra · adu2021System enabling intelligent tool integration and selection for enhanced multimodal reasoning, allowing agents to leverage heterogeneous capabilities for complex problem-solving across domains.
- ▌ Moss Transcribe Diarize Accurate Transcription Wit · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Muses Designing Composing Generating Nonexistent F · adu2021Enhanced language model pre-training methodology improving linguistic competence across languages, strengthening foundational capabilities for multilingual agent systems.
- ▌ Nitrogen An Open Foundation Model For Generalist G · adu2021Open-source foundation model trained for diverse agent interactions across gaming and interactive environments, supporting generalist agent development with robust performance foundations.
- ▌ Openrt An Open Source Red Teaming Framework For Mu · adu2021Comprehensive evaluation dataset for systematic vulnerability testing of language models, enabling identification and mitigation of failure modes before agent deployment.
- ▌ Parallel Latent Reasoning For Sequential Recommend · adu2021Advanced reasoning approach for optimizing inference efficiency through meta-cognitive planning, enabling agents to make better decisions with reduced computational overhead.
- ▌ Pearmut Human Evaluation Of Translation Made Trivi · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Rebuttalagent Strategic Persuasion In Academic Reb · adu2021Implement techniques from RebuttalAgent: Strategic Persuasion in Academic Rebuttal via Theory of Mind. Although artificial intelligence (AI) has become deeply integrated into various stages of the research workflow and achieved remarkable advancements, academic rebuttal remains a significant and underexplored challenge
- ▌ Redbench A Universal Dataset For Comprehensive Red · adu2021Comprehensive evaluation dataset for systematic vulnerability testing of language models, enabling identification and mitigation of failure modes before agent deployment.
- ▌ Restok Learning Hierarchical Residuals In 1d Visua · adu2021Generative approach for creating complex dynamic scenes and content, supporting agent capabilities in understanding and reasoning about multi-agent environments.
- ▌ Rgs Slam Robust Gaussian Splatting Slam With One S · adu2021Advanced computer vision technique for robust spatial understanding in complex scenes, supporting agent navigation and environmental reasoning capabilities.
- ▌ Salad Achieve High Sparsity Attention Via Efficien · adu2021Implement techniques from SALAD: Achieve High-Sparsity Attention via Efficient Linear Attention Tuning for Video Diffusion Transformer. Diffusion Transformers have recently demonstrated remarkable performance in video generation
- ▌ Scientific Image Synthesis Benchmarking Methodolog · adu2021Implement techniques from Scientific Image Synthesis: Benchmarking, Methodologies, and Downstream Utility. While synthetic data has proven effective for improving scientific reasoning in the text domain, multimodal reasoning remains constrained by the difficulty of synthesizing scientifically rigorous images
- ▌ Scievalkit An Open Source Evaluation Toolkit For S · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Sop A Scalable Online Post Training System For Vis · adu2021Enhanced language model pre-training methodology improving linguistic competence across languages, strengthening foundational capabilities for multilingual agent systems.
- ▌ Steerability Of Instrumental Convergence Tendencie · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Swe Pruner Self Adaptive Context Pruning For Codin · adu2021Implement techniques from SWE-Pruner: Self-Adaptive Context Pruning for Coding Agents. LLM agents have demonstrated remarkable capabilities in software development, but their performance is hampered by long interaction contexts, which incur high API costs and latency
- ▌ The Sonar Moment Benchmarking Audio Language Model · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Thinkrl Edit Thinking In Reinforcement Learning Fo · adu2021Advanced reasoning approach for optimizing inference efficiency through meta-cognitive planning, enabling agents to make better decisions with reduced computational overhead.
- ▌ Towards Pixel Level Vlm Perception Via Simple Poin · adu2021Implement techniques from Towards Pixel-Level VLM Perception via Simple Points Prediction. We present SimpleSeg, a strikingly simple yet highly effective approach to endow Multimodal Large Language Models (MLLMs) with native pixel-level perception
- ▌ Twinbrainvla Unleashing The Potential Of Generalis · adu2021Implement techniques from TwinBrainVLA: Unleashing the Potential of Generalist VLMs for Embodied Tasks via Asymmetric Mixture-of-Transformers. The fundamental premise of Vision-Language-Action (VLA) models is to harness the extensive general capabilities of pre-trained Vision-Language Models (VLMs) for generalized embodied intelligence
- ▌ U Net Like Spiking Neural Networks For Single Imag · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ Unified Thinker A General Reasoning Modular Core F · adu2021Advanced reasoning approach for optimizing inference efficiency through meta-cognitive planning, enabling agents to make better decisions with reduced computational overhead.
- ▌ Video Rts Reinforcement Learning Test Time Scaling · adu2021Train video QA models on 6K examples using pure RL instead of costly supervised fine-tuning, then adaptively sample video frames at inference by monitoring answer consensus across multiple reasoning trajectories.
- ▌ Visgym Diverse Customizable Scalable Environments · adu2021Implement techniques from VisGym: Diverse, Customizable, Scalable Environments for Multimodal Agents. The suite spans symbolic puzzles, real-image understanding, navigation, and manipulation, and provides flexible controls over difficulty, input representation, planning horizon, and feedback
- ▌ Vista Path An Interactive Foundation Model For Pat · adu2021Implement techniques from VISTA-PATH: An interactive foundation model for pathology image segmentation and quantitative analysis in computational pathology. Accurate semantic segmentation for histopathology image is crucial for quantitative tissue analysis and downstream clinical modeling
- ▌ Webgym Scaling Training Environments For Visual We · adu2021Training platform with thousands of realistic web navigation tasks for developing visual web agents, supporting scalable training and evaluation of agents that can generalize to arbitrary websites.
- ▌ Why Llms Arent Scientists Yet Lessons From Four Au · adu2021Research contribution advancing agent and reasoning capabilities through novel approaches to model development, training, and evaluation.
- ▌ X Mutest Multilingual Benchmark For Explainable Ha · adu2021Systematic evaluation toolkit for assessing large language models across multiple dimensions, enabling comprehensive benchmarking of agent capabilities and comparative analysis of model performance.
- ▌ Wildworld Action Conditioned World Modeling Dataset · adu2021Build action-conditioned world models with explicit state tracking using WildWorld's 108M+ frames from Monster Hunter: Wilds. Includes data acquisition protocol with skeleton and world state annotations, quality filtering pipeline removing temporal discontinuities and cutscenes, and WildBench evaluation metrics (video quality, camera control, action following, state alignment) for assessing long-horizon consistency and state-aware predictions.
- ▌ Video Generation Latent Rewards · adu2021Use pre-trained video generation models (VGMs) as efficient reward models by evaluating video quality directly in latent space at any denoising timestep, enabling process-reward learning across the entire generation trajectory without expensive VAE decoding to RGB.
- ▌ Vision Language Action Dreaming · adu2021Predict robot actions through a perception-prediction-action loop: forecast dynamic regions, depth, and semantic features from visual observations, then generate action sequences via diffusion transformers.
- ▌ Web Agent Declarative Framework · adu2021Enable safe and efficient AI agent interactions with websites via declarative HTML extensions—define agent-executable tools and context using <tool> and <context> tags instead of relying on brittle UI inference.
- ▌ Web Agent Training Optimization · adu2021Optimize open-source LLM web agent training through systematic analysis of supervised fine-tuning vs. reinforcement learning trade-offs. Achieve 45% lower compute cost by branching into RL at strategic SFT checkpoints.
- ▌ Z Image Single Stream Diffusion · adu2021Unified text-image token processing in a compact 6B-parameter transformer enabling sub-second inference on enterprise GPUs through hierarchical distillation and single-stream architecture. Use when generating high-quality images with tight latency budgets or limited GPU memory.
- ▌ Agent World Model Env Generation · adu2021Automatically synthesize executable RL training environments with database backends, Python tools, and task descriptions. Generate 1000+ diverse domains with 10K+ tasks enabling data-efficient tool-use agent training without manual scenario design.
- ▌ Agentfly Memory Augmented Agents · adu2021Enable agent learning through episodic memory and neural case selection without fine-tuning the underlying LLM, achieving efficient continual adaptation via policy updates in memory space.
- ▌ Block Ffn Sparse Mixture Experts · adu2021Deploy efficient MoE models on resource-constrained edge devices by learning chunk-level activation sparsity that achieves 3.67× speedup. Use when you need to compress LLMs for on-device inference while maintaining reasoning quality and supporting speculative decoding acceleration.
- ▌ Deepseek Math V2 Self Verifiable · adu2021Synergistic verifier-generator training loop enabling LLMs to identify logical issues in mathematical proofs without reference solutions, improving reasoning rigor through meta-verification. Apply when you need to scale mathematical reasoning without hand-labeled proof annotations.
- ▌ Diffthinker Multimodal Reasoning · adu2021Apply diffusion models as native generative agents for vision-centric reasoning tasks (sequential planning, constraint satisfaction, spatial configuration) instead of text-based LLM chains. Achieves 3x+ improvements over GPT-5 and Gemini-3 on visual reasoning. Use when image-to-image generation better captures the reasoning constraints than text-based problem decomposition.
- ▌ Dope Denoising Rotary Embeddings · adu2021Improve long-context length extrapolation by denoising instabilities in Rotary Position Embeddings (RoPE) through spectral analysis and selective head rewriting—training-free post-hoc intervention for longer context windows.
- ▌ Dreamgym Experience Synthesis Rl · adu2021Scale agent learning by synthesizing diverse experiences using reasoning-based models instead of costly real-world rollouts, maintaining replay buffers with both real and synthetic interactions while using adaptive curriculum to focus on challenging tasks.
- ▌ Dynamic Visual Token Pruning Vlm · adu2021Dynamic token pruning framework for VLMs that adapts compression to scene complexity through single-pass selection. Removes 92.6% of visual tokens while maintaining performance and enabling superior fine-tuning.
- ▌ Efficient Long Context Reasoning · adu2021Achieve 10× higher decoding throughput on long prompts by replacing 50% of cross-attention layers with gated memory units (GMUs) combining SSMs and attention. Maintains reasoning capability while reducing memory I/O bottleneck from O(d_kv·N) to O(d_h).
- ▌ Empo2 Memory Augmented LLM Agent · adu2021Improve exploration in LLM-based agents through external memory-augmented RL with hybrid on/off-policy training. Agents generate exploration 'tips' (self-reflections) after trajectories, storing them in memory. During rollouts, policy samples between standard execution and memory-conditioned execution. Off-policy updates distill memory-guided behaviors into base policy via reward-guided knowledge distillation. Achieves 128.6% improvement on ScienceWorld and 11.3% on WebShop vs. GRPO.
- ▌ Entropy Ratio Clipping Stable Rl · adu2021Stabilize LLM post-training by constraining global distributional shifts in policy exploration. Entropy Ratio Clipping supplements local clipping mechanisms with global entropy constraints—essential when PPO alone produces unstable gradients and distribution shifts.
- ▌ F Grpo Focal Policy Optimization · adu2021Prevent policy collapse onto common solutions during reinforcement learning by applying difficulty-aware focal weighting to gradient contributions, maintaining diversity across solution modes while preserving performance.
- ▌ Fairy2i Complex LLM Quantization · adu2021Enable extreme 1-2bit quantization of pre-trained LLMs by leveraging complex-valued arithmetic. Convert real-valued linear layers to complex domain losslessly, quantize to fourth roots of unity {±1, ±i}, and apply recursive residual error quantization for near full-precision performance.