Explore tools, prompts, skills, news and resources related to MIT 第2页.
Latest articles, prompts, use cases and editorial content about MIT 第2页
FIRME is a registry-anchored memory graph designed to reconcile recurring investor observations into persistent canonical relationships while retaining source, time, and context. It uses rule-based within-period fusion t...
An arXiv preprint reports that multi-turn reinforcement learning in simulation improved a language model’s structured psychiatric intake performance compared with its identically prompted base model. In 120 encounters wi...
Researchers present a model-agnostic neurosymbolic framework for interpreting the final hidden layer of audio classifiers through symbolic concept induction. The method is evaluated on music genre classification with GTZ...
The paper identifies a sequence-confidence misalignment in reinforcement learning with verifiable rewards: because the advantage is blind to a policy's confidence, a response the policy barely found and one it produ...
This arXiv paper examines role spoofing, an attack in which malicious text—such as content from a tool output or webpage—exploits LLM role confusion to impersonate the user. The authors show that models track speakers us...
Microsoft Research describes an Industry Evaluation Data Generation Platform that creates fully synthetic enterprise tenants for evaluating Microsoft Copilot without touching customer data. Environments are grounded in p...
This arXiv study examines looping in dense decoder-only transformers, where weights are reused to make models deeper without increasing parameter count. It finds that looping the full transformer gives the strongest qual...
The paper measures the energy cost of speculative decoding under production serving conditions on H100 and H200 GPUs using SGLang with CUDA graphs, at batch sizes up to 64 across five corpora. It finds that on dense mode...
TaskFrame is a hierarchical planner-controller framework that pairs a vision-language task planner with a causal Transformer controller that emits keyboard and mouse actions at every game step. The planner decomposes a h...
The paper analyzes advantage estimation in GRPO and factors it into sample-level baselining and token-level credit assignment, noting that existing estimators typically handle only one stage. TACE replaces linear standar...
Researchers present Bridge-LP, a cross-modal framework that unifies graph and language representations for link prediction on text-attributed graphs. The design combines textual-proximity jumping connections, compressed...
DynaRev introduces 181 tasks that evaluate how web agents respond when users revise requirements during execution, measuring correctness, responsiveness and efficiency. Evaluations across five models show that agents oft...
A new arXiv paper addresses offline goal-conditioned reinforcement learning in stochastic long-horizon settings, where compounding value-estimation errors make reliable goal reaching difficult. The authors identify a fai...
The paper proposes Domain-Specialized Upcycling (DS-Upcycling), which expands a dense Transformer layer with additional experts and a router rather than attaching per-domain LoRA adapters. A new domain-aware supervision...
论文指出 AI 代理现在承担创建 PR、选择依赖、重构代码库、解释测试失败以及执行基础设施变更等任务，但会话一旦结束，它们的推理、决策和工具调用记录就会消失；而 git log 只显示改了什么，不显示哪个模型、调用了哪些工具、谁对结果负责。作者提出 Agentic Process Evidence (APE)，这是一种签名、防篡改、机器可读的代理过程证明，把代理会话与其结果绑定。APE 支持 in-toto 风格证明框架，并支持 Pol...
This paper analyzes clustered optimal-transport activation steering for toxicity mitigation in Gemma-2-2B and Granite-3.1-2B. It proves that cluster displacements averaged by cluster size equal the standard difference-of...
A compute-matched study reports that synthetic pre-pretraining (PPT) never outperformed pure natural-language training across four curricula (k-Dyck, neural cellular automata, declarative reference corpora and code) and...
Researchers introduced EPC-RawDC, a conditional 3D latent diffusion framework for CBCT-to-CT image synthesis that incorporates scanner-specific acquisition physics rather than relying on image data alone. The method comb...
NVIDIA and MIT researchers published model weights for Open-H3-Regenerate-2K on Hugging Face, a video enhancement project built on MiniMax's open-source H3 video enhancement approach. The release targets high-resolu...
Discover tools, products and platforms related to MIT 第2页
