Recent
1744 →
llama.cpp b9785 Release with Hardened Caps Check and Multi-Platform Binaries
Prompt Format Inquiry for Training Unsloth/Phi-3.5-mini-instruct
Space-Efficient Language Generation in the Limit
Argus Benchmark Evaluates Uncertainty Quantification Stability Across Vision-Language Models and GUI Grounding Datasets
SARA: Unlocking Multilingual Knowledge in Mixture-of-Experts via Semantically Anchored Routing Alignment
+1739 more
Inference efficiency
173 →
llama.cpp b9785 Release with Hardened Caps Check and Multi-Platform Binaries
Gemma4-26B-A4B & 31B-QAT Uncensored Balanced Released with MTP Speed Boosts
GLM-5.2 on 4x DGX Spark: Reconstructing Missing Build Steps for MTP Speculative Decode
+170 more
Training methods
219 →
Space-Efficient Language Generation in the Limit
SARA: Unlocking Multilingual Knowledge in Mixture-of-Experts via Semantically Anchored Routing Alignment
Gefen: A Drop-in Replacement for AdamW with Claimed 8x Memory Reduction
+216 more
Research paper
270 →
Space-Efficient Language Generation in the Limit
SARA: Unlocking Multilingual Knowledge in Mixture-of-Experts via Semantically Anchored Routing Alignment
Colony: An Educational Simulation of LLM Attention Mechanisms Using Agent-Based Analogies
+267 more
Retrieval & RAG
38 →
How Large Language Models Source Brand Reputation Across Languages and Markets
Ontological Inversion: Flipping LLM Emotional Concepts via Negative Gain
Hi-Seg: Human-AI Collaboration for Pulmonary Nodule Segmentation
+35 more
AI agents
378 →
ToolBench-X: Benchmarking Tool-Using Agents Under Unreliable Environments
Colony: An Educational Simulation of LLM Attention Mechanisms Using Agent-Based Analogies
Claude Code v2.1.191 Release Notes
+375 more
Evaluation & benchmarks
817 →
ToolBench-X: Benchmarking Tool-Using Agents Under Unreliable Environments
How Large Language Models Source Brand Reputation Across Languages and Markets
Community Inquiry on Model Benchmarking Methods
+814 more
Open weights
207 →
Gemma4-26B-A4B & 31B-QAT Uncensored Balanced Released with MTP Speed Boosts
GLM-5.2 on 4x DGX Spark: Reconstructing Missing Build Steps for MTP Speculative Decode
SDXL Running Locally in Browser on WebGPU, Open-Source
+204 more
API & product launches
53 →
Simon Willison converts MDN browser compatibility data into a SQLite database
User Reports Step 3.7 Flash Model Tool Access Failure on HuggingChat
Claude Code v2.1.191 Release Notes
+50 more
Hardware & chips
34 →
Query on Clustering Nvidia DGX Spark and AMD Ryzen AI Max 395 for Unified Memory Inference
MINISFORUM DEG1 Oculink eGPU Dock Refurbished Available for $59
7 Chinese companies shipping H100/H200-class AI chips, most IPO'd in last 6 months
+31 more
Safety & alignment
191 →
User Observes Cloud Chatbots Appear Less Intelligent Than Local Models
Swiss Federal Supreme Court evaluates Heretic for internal use
Governance Decay in Long-Horizon LLM Agents
+188 more
Image generation
51 →
SDXL Running Locally in Browser on WebGPU, Open-Source
Atomistic Language Models Understand and Generate Materials
Unlimited-OCR is now on ModelScope
+48 more
Code generation
270 →
I reverse engineered Windows Copilot into a free OpenAI-compatible API
Gemini 3.5 Flash Adds Computer Use Capability
Has anyone else found vLLM outputs worse than llama.cpp?
+267 more
Voice & audio
37 →
Introducing the FFASR Leaderboard: Benchmarking ASR in the Real World
CN-NewsTTS Bench v0.1 Released
Poster: Exploring Audio-Based Scam Detection in Turkish
+34 more
Multimodal
152 →
Efficient Multimodal Models for Pulmonary Embolism Risk Assessment
MMGist: A Comprehensive Multimodal Benchmark for 2027
Deep Learning Pipeline for Sign Language Recognition and Translation to Indian Vernaculars
+149 more
Reasoning models
684 →
LLMs Use Difference-Making Logic to Learn Causal Structure
Gazer: Training-Free Semantic Correction for Autoregressive Visual Models
Multimodal Chain-of-Thought: Capabilities and Limitations
+681 more
Robotics
16 →
Reward-Petri-Net Interpretation of Temporal Behavior Trees
A Taxonomy of Conceptual Alignment in Human-Robot Dialogue
NVIDIA Launches Halos for Robotics: Full-Stack Functional Safety System
+13 more
Policy & regulation
20 →
Bill to Mandate AI Chip Location Tracking Gains Industry Support
Machine Whistleblowing: A Normative and Principled Approach
OpenAI Builds Shared AI Standards via Appia Foundation
+17 more
Training data
69 →
UD_Czech-PDTC: A Large and Genre-Rich Treebank in Universal Dependencies
Koshur Pixel: First Large-Scale Synthetic OCR Dataset for Kashmiri
Enable Real-Time AI for High-Speed Data Acquisition with DAQIRI
+66 more