Install
RedditVote46FlipShareTweet46 SharesRedditVote46FlipShareTweet46 Shares
- 1,173articles · 365d
- 4+ hour agolatest article
- Sep 13, 2025earliest in window
- 97%with images · 26 videos
- 339avg words
- Science & Technology 1,152
- Software Dev. 999
- Computers & Electronics 896
- Science & Nature 147
- Jobs & Education 101
- STEM 83
- News 33
- Business & Industrial 17
Please confirm you are human
This browser or connection looks automated. Press and continuously hold the control for 3 seconds to enable Google-hosted web results and, when separately allowed, AI-assisted answers.
A successful check enables 100 search requests. Interactive access does not authorize scraping, systematic collection, or reuse of search output.
News
Stanford Researchers Introduce TRACE: A Capability-Targeted Agentic Training System That Turns Recurrent Agent Failures Into Synthetic RL Environment
1+ mon, 4+ week ago (676+ words) Agentic LLMs often fail the same way, again and again. A Stanford research team traced this to missing, reusable capabilities. Their system, TRACE, diagnoses those gaps and trains for them directly. TRACE stands for Turning Recurrent Agent failures into Capability-targeted…...
Prime Intellect Releases prime-rl 0.6.0 to Train Trillion-Parameter MoE Models on Agentic RL Workloads
2+ mon, 2+ week ago (550+ words) The research team trained GLM-5 on SWE tasks at up to 131k sequence length. Step times stayed under five minutes. The batch size was 256 rollouts. The run used only 28 H200 nodes. prime-rl is an open framework for asynchronous reinforcement learning. It post-trains…...
VibeThinker-3B: A 3B Dense Reasoning Model Built on Qwen2.5-Coder-3B With the Spectrum-to-Signal Post-Training Pipeline
2+ mon, 3+ week ago (749+ words) While recent breakthroughs in AI reasoning have largely been driven by massive scale, pouring in billions of parameters to cross complex cognitive thresholds—VibeThinker-3B is charting a completely different path. Created by researchers from Sina Weibo Inc (China), this 3-billion-parameter…...
Parallax: A Parameterized Local Linear Attention That Keeps Softmax and Adds a Learned Covariance Correction Branch
3+ mon, 1+ week ago (1040+ words) The Transformer’s attention mechanism has barely changed since 2017. Most efficiency work has tried to replace softmax attention outright. A new paper takes a different route. It keeps softmax attention and bolts on a correction branch. A team of researchers from…...
Trajectory Releases a Concurrent Multi-LoRA Training Stack for Continual Learning, Reporting a 2.81× Experiment-Throughput Gain
3+ mon, 1+ week ago (962+ words) Most language models improve in discontinuous jumps. A team collects data, trains, and ships a new version. This takes months and produces remarkable or catastrophic behavior for users. Trajectory wants to replace that cycle with continual learning. The Trajectory team…...
Nous Research Proposes Lighthouse Attention: A Training-Only Selection-Based Hierarchical Attention That Delivers 1.4–1.7× Pretraining Speedup at Long Context
3+ mon, 3+ week ago (1602+ words) Lighthouse takes a different approach on both design decisions. It pools queries, keys, and values symmetrically across a multi-level pyramid, and it places selection entirely outside the attention kernel. After selection, the system gathers the chosen entries into a contiguous,…...
Google DeepMind Introduces Decoupled DiLoCo: An Asynchronous Training Architecture Achieving 88% Goodput Under High Hardware Failure Rates
4+ mon, 2+ week ago (649+ words) MarkTechPost Training frontier AI models is, at its core, a coordination problem. Thousands of chips must communicate with each other continuously, synchronizing every gradient update across the network. When one chip fails or even slows down, the entire training run…...
Hugging Face Releases ml-intern: An Open-Source AI Agent that Automates the LLM Post-Training Workflow
4+ mon, 3+ week ago (883+ words) Hugging Face has released ml-intern, an open-source AI agent designed to automate end-to-end post-training workflows for large language models (LLMs). Built on the company’s smolagents framework, the tool can autonomously perform literature review, dataset discovery, training script execution, and iterative…...
Hugging Face Releases TRL v1.0: A Unified Post-Training Stack for SFT, Reward Modeling, DPO, and GRPO Workflows
5+ mon, 1+ week ago (223+ words) In the early stages of the LLM boom, post-training was often treated as an experimental ‘dark art.’ TRL v1.0 aims to change that by providing a consistent developer experience built on three core pillars: a dedicated Command Line Interface (CLI), a…...
Liquid AI Released LFM2.5-350M: A Compact 350M Parameter Model Trained on 28T Tokens with Scaled Reinforcement Learning
5+ mon, 1+ week ago (334+ words) In the current landscape of generative AI, the ‘scaling laws’ have generally dictated that more parameters equal more intelligence. However, Liquid AI is challenging this convention with the release of LFM2.5-350M. This model is actually a technical case study in intelligence…...