Uncategorized

It’s MCP projects in production, not specification elegance or market buzz, that will determine if MCP (or something else) stays on top.Read More

Handling extremely long documents remains a persistent challenge for large language models (LLMs). Even with techniques such as length extrapolation and sparse attention, models often suffer from performance degradation and high computational costs. To address this, researchers from ByteDance Seed and Tsinghua University introduce MemAgent, a reinforcement learning-based memory agent designed to enable long-context processing with linear complexity and minimal performance loss. Limitations of Existing Approaches Current solutions for long-context modeling fall into three main categories: Length Extrapolation Methods (e.g., NTK, PI, YaRN, DCA): Extend the context window via positional embedding manipulations. However, they often face performance degradation and scaling issues. Sparse and Linear Attention Mechanisms: Reduce attention complexity to O(n) but typically require retraining from scratch and rely on fixed patterns or human-defined rules. Context Compression: Use token-level or external memory modules to condense long inputs but often disrupt standard generation and struggle with extrapolation. These approaches fail to deliver all three critical attributes: arbitrary input length support, consistent accuracy, and efficient linear complexity. MemAgent: Human-Like Memory Strategy Inspired by how humans summarize key information while ignoring noise, MemAgent processes input as a stream of evidence. At each step, it reads a document chunk and an internal memory, overwriting the latter with updated, compressed context. Key innovations: Fixed-Length Token-Based Memory: Compresses essential information while maintaining model compatibility. Segment-Wise Overwrite Mechanism: Supports infinite text lengths without growing memory. Linear Complexity: Memory update and decoding cost remain constant per chunk. Multi-Conv RL Training with GRPO MemAgent treats each document chunk interaction as an independent dialogue. It is trained via Group Relative Policy Optimization (GRPO) within a multi-conversation RL pipeline called DAPO, enabling reward-driven memory update. Key elements include: Rule-Based Verifier: Calculates outcome rewards by comparing model answers with multiple ground truths. Token-Level RL Signal: Applied uniformly across conversations stemming from a sample. This setup encourages memory compression focused on answer-relevant information and discards distractors. Performance Evaluation Using the RULER benchmark and synthetic datasets from HotpotQA and SQuAD, MemAgent was trained with an 8K context window and extrapolated up to 3.5 million tokens. Model 224K 896K 3.5M Qwen2.5-Instruct-14B-1M 37.5% 0.0% N/A QwenLong-L1-32B 17.2% 11.7% N/A RL-MemAgent-14B 81.3% 77.3% 78.1% MemAgent maintained over 95% accuracy on RULER benchmarks (8K to 512K tokens) and consistently outperformed long-context and distillation-based baselines. Case Study: Multi-Hop QA Given the query “The director of the romantic comedy ‘Big Stone Gap’ is based in what New York city?”, MemAgent progressively tracked relevant content across 3 chunks: Recognized unrelated content but retained location information. Maintained memory against irrelevant chunks. Correctly updated memory upon encountering Adriana Trigiani’s biography. Final answer: Greenwich Village, New York City. Theoretical Foundation and Complexity MemAgent reformulates the autoregressive model using latent memory variables (m₁…mₖ): p(x₁:N) = ∑ₘ₁:ₖ ∏ₖ p(cₖ | mₖ₋₁) * p(mₖ | cₖ, mₖ₋₁) This enables O(N) compute cost and human-readable intermediate memory—unlike attention-based feature compression. RL is essential, as memory updates are discrete and can’t be learned via backpropagation. Conclusion MemAgent offers a scalable and efficient solution to the long-context trilemma: unlimited input length, near-lossless accuracy, and linear complexity. Its RL-based overwrite memory mechanism allows LLMs to read, abstract, and generate over multi-million-token inputs without architectural modification. FAQs Q1: What is MemAgent?MemAgent is a reinforcement learning-based framework that equips LLMs with memory tokens to handle extremely long contexts efficiently. Q2: How is it different from attention or extrapolation methods?Unlike attention-based scaling or extrapolation techniques, MemAgent uses token-based memory updated via reinforcement learning. Q3: What models can MemAgent be applied to?Any Transformer-based LLM. No changes to the model architecture are required. Q4: How does it scale with input size?It maintains linear computational complexity regardless of input length by fixing the memory size. Q5: What are the applications of MemAgent?Long-document QA, agent memory systems, legal document review, scientific literature analysis, and real-time decision-making with large evidence bases. Check out the Paper. All credit for this research goes to the researchers of this project. Sponsorship Opportunity: Reach the most influential AI developers in US and Europe. 1M+ monthly readers, 500K+ community builders, infinite possibilities. [Explore Sponsorship] The post MemAgent: A Reinforcement Learning Framework Redefining Long-Context Processing in LLMs appeared first on MarkTechPost.

NVIDIA AI has introduced OpenReasoning-Nemotron, a family of large language models (LLMs) designed to excel in complex reasoning tasks across mathematics, science, and code. This model suite—comprising 1.5B, 7B, 14B, and 32B parameter versions—has been distilled from the 671B DeepSeek R1 0528 model, capturing its high-level reasoning capabilities in significantly smaller and more efficient models. The release positions NVIDIA as a leading contributor to the open-source LLM ecosystem, delivering models that push state-of-the-art (SOTA) performance while remaining commercially permissive and widely accessible via Hugging Face. Model Overview and Architecture Distillation from DeepSeek R1 0528 (671B) At the heart of OpenReasoning-Nemotron lies a distillation strategy that transfers reasoning ability from DeepSeek R1—a massive 671B parameter model—into smaller architectures. The process prioritizes reasoning generalization over raw token prediction, enabling compact models to perform effectively on structured, high-cognition tasks. The distillation dataset emphasizes mathematics, science, and programming languages, aligning model capabilities with key reasoning domains. Model Variants and Specs Model Name Parameters Intended Use Hugging Face Page OpenReasoning-Nemotron-1.5B 1.5B Entry-level reasoning and inference Link OpenReasoning-Nemotron-7B 7B Mid-scale reasoning, good for code/math Link OpenReasoning-Nemotron-14B 14B Advanced reasoning capabilities Link OpenReasoning-Nemotron-32B 32B Near frontier-model performance in logic-intensive tasks Link All models are compatible with transformer architectures, support FP16/INT8 quantization, and are optimized for NVIDIA GPUs and NeMo frameworks. Performance Benchmarks These models set new state-of-the-art pass@1 scores for their size class across multiple reasoning benchmarks: Model GPQA MMLU‑PRO HLE LiveCodeBench SciCode AIME24 AIME25 HMMT Feb 2025 1.5B 31.6 47.5 5.5 28.6 2.2 55.5 45.6 31.5 7B 61.1 71.9 8.3 63.3 16.2 84.7 78.2 63.5 14B 71.6 77.5 10.1 67.8 23.5 87.8 82.0 71.2 32B 73.1 80.0 11.9 70.2 28.5 89.2 84.0 73.8 All quoted scores are pass@1 without GenSelect. GenSelect (Heavy Mode) Using Generative Selection with 64 candidates (“GenSelect”), performance further improves, especially at 32B: 32B achieves: AIME24 89.2 → 93.3, AIME25 84.0 → 90.0, HMMT 73.8 → 96.7, LiveCodeBench 70.2 → 75.3. This demonstrates strong emergent reasoning performance at scale. Training Data and Reasoning Specialization The training corpus is a distilled, high-quality subset of the DeepSeek R1 0528 dataset. Key features include: Heavily curated reasoning data from math, science, and CS disciplines. Prompt-engineered fine-tuning designed to reinforce multi-step thought chains. Emphasis on logical consistency, constraint satisfaction, and symbolic reasoning. This deliberate curation ensures strong alignment with real-world reasoning problems found in both academia and applied ML domains. Open and Ecosystem Integration All four OpenReasoning-Nemotron models are released under an open and commercially permissive license, with model cards, evaluation scripts, and inference-ready weights available on Hugging Face: OpenReasoning-Nemotron-1.5B OpenReasoning-Nemotron-7B OpenReasoning-Nemotron-14B OpenReasoning-Nemotron-32B These models are designed to plug into the NVIDIA NeMo framework, and support TensorRT-LLM, ONNX, and Hugging Face Transformers toolchains, facilitating rapid deployment in production and research settings. Key Use Cases Math tutors and theorem solvers Scientific QA agents and medical reasoning systems Code generation and debugging assistants Chain-of-thought multi-hop question answering Synthetic data generation for structured domains Conclusion NVIDIA’s OpenReasoning-Nemotron models offer a pragmatic, open-source path toward scaling reasoning ability without frontier-scale compute costs. By distilling from the 671B DeepSeek R1 and targeting high-leverage reasoning domains, these models deliver a powerful balance of accuracy, efficiency, and accessibility. For developers, researchers, and enterprises working on logic-intensive AI applications, OpenReasoning-Nemotron provides a compelling foundation—free from the trade-offs that often accompany proprietary or overgeneralized models. Frequently Asked Questions (FAQs) Q1. What benchmarks are supported?GPQA, MMLU-PRO, HLE, LiveCodeBench, SciCode, AIME 2024/25, HMMT Feb 2025 (pass@1). Q2. How much data was used?A distillation corpus of 5 million reasoning log examples across domains, generated by DeepSeek‑R1‑0528. Q3. Is reinforcement learning used?No—models are trained purely via SFT, preserving efficiency while enabling future RL research. Q4. Can I scale reasoning with GenSelect?Yes. Using GenSelect significantly boosts performance—32B jumps from 73.8 to 96.7 on HMMT with 64 candidates. Check out the Technical details. All credit for this research goes to the researchers of this project. Sponsorship Opportunity: Reach the most influential AI developers in US and Europe. 1M+ monthly readers, 500K+ community builders, infinite possibilities. [Explore Sponsorship] The post NVIDIA AI Releases OpenReasoning-Nemotron: A Suite of Reasoning-Enhanced LLMs Distilled from DeepSeek R1 0528 appeared first on MarkTechPost.

Discover OpenAI’s red team blueprint: How 110 coordinated attacks and 7 exploit fixes created ChatGPT Agent’s revolutionary 95% security defense system.Read More

The Growing Threat Landscape for LLMs LLMs are key targets for fast-evolving attacks, including prompt injection, jailbreaking, and sensitive data exfiltration. It is necessary to adapt defense mechanisms that move beyond static safeguards because of the fluid nature of these threats. Current LLM security techniques suffer due to their reliance on static, training-time interventions. Static filters and guardrails are fragile against minor adversarial tweaks, while training-time adjustments fail to generalize to unseen attacks after deployment. Machine unlearning often fails to erase knowledge completely, leaving sensitive information vulnerable to resurfacing. Current safety and security scaling mainly focuses on training-time methods, with limited exploration of test-time and system-level safety. Why Existing LLM Security Methods Are Insufficient RLHF and safety fine-tuning methods attempt to align models during training but show limited effectiveness against novel post-deployment attacks. System-level guardrails and red-teaming strategies provide additional protection layers, yet prove brittle against adversarial perturbations. Unlearning unsafe behaviors shows promise in specific scenarios, yet fails to achieve complete knowledge suppression. Multi-agent architectures are effective in distributing complex tasks, but their direct application to LLM security remains unexplored. Agentic optimization methods like TEXTGRAD and OPTO utilize structured feedback for iterative refinement, and DSPy facilitates prompt optimization for multi-stage pipelines. However, they are not applied systematically to security enhancement at inference time. AegisLLM: An Adaptive Inference-Time Security Framework Researchers from the University of Maryland, Lawrence Livermore National Laboratory, and Capital One have proposed AegisLLM (Adaptive Agentic Guardrails for LLM Security), a framework to improve LLM security through a cooperative, inference-time multi-agent system. It utilizes a structured agentic system of LLM-powered autonomous agents that continuously monitor, analyze, and reduce adversarial threats. The key components of AegisLLM are Orchestrator, Deflector, Responder, and Evaluator. Through automated prompt optimization and Bayesian learning, the system refines its defense capabilities without model retraining. This architecture allows real-time adaptation to evolving attack strategies, providing scalable, inference-time security while preserving the model’s utility. Coordinated Agent Pipeline and Prompt Optimization AegisLLM operates through a coordinated pipeline of specialized agents, each responsible for distinct functions while working in concert to ensure output safety. All agents are guided by carefully designed system prompts and user input at test time. Each agent is governed by a system prompt that encodes its specialized role and behavior, but manually crafted prompts typically fall short of optimal performance in high-stakes security scenarios. Therefore, the system automatically optimizes each agent’s system prompt to maximize effectiveness through an iterative optimization process. At each iteration, the system samples a batch of queries and evaluates them using candidate prompt configurations for specific agents. Benchmarking AegisLLM: WMDP, TOFU, and Jailbreaking Defense On the WMDP benchmark using Llama-3-8B, AegisLLM achieves the lowest accuracy on restricted topics among all methods, with WMDP-Cyber and WMDP-Bio accuracies approaching to 25% theoretical minimum. On the TOFU benchmark, it achieves near-perfect flagging accuracy across Llama-3-8B, Qwen2.5-72B, and DeepSeek-R1 models, with Qwen2.5-72B almost 100% accuracy on all subsets. In jailbreaking defense, results show strong performance against attack attempts while maintaining appropriate responses to legitimate queries on StrongREJECT and PHTest. AegisLLM achieves a 0.038 StrongREJECT score, competitive with state-of-the-art methods, and an 88.5% compliance rate without requiring extensive training, enhancing defense capabilities. Conclusion: Reframing LLM Security as Agentic Inference-Time Coordination In conclusion, researchers introduced AegisLLM, a framework that reframes LLM security as a dynamic, multi-agent system operating at inference time. AegisLLM’s success highlights that one should approach security as an emergent behavior from coordinated, specialized agents, rather than a static model characteristic. This shift from static, training-time interventions to adaptive, inference-time defense mechanisms solves the limitations of current methods while providing real-time adaptability against evolving threats. Frameworks like AegisLLM that enable dynamic, scalable security will become increasingly important for responsible AI deployment as language models continue to advance in capability. Check out the Paper and GitHub Page. All credit for this research goes to the researchers of this project. Sponsorship Opportunity Reach the most influential AI developers worldwide. 1M+ monthly readers, 500K+ community builders, infinite possibilities. [Explore Sponsorship] The post AegisLLM: Scaling LLM Security Through Adaptive Multi-Agent Systems at Inference Time appeared first on MarkTechPost.