Hugging Face Blog
LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillation
LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillation
// espacio-latente.com / radar
Píldora diaria de lo que se mueve en IA: laboratorios, papers, blogs de referencia y releases relevantes, resumido y con link al original. Se actualiza dos veces al día. · ← volver a espacio-latente.com · archivo · RSS
LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillation
v0.125.0
v0.124.0
v0.123.0
Offering Zero Data Retention for frontier models
Replit expands access to software creation with GPT-5.6 Luna
smolmachines / smolvm as a sandbox for untrusted Python & JavaScript
Quoting Jeremy Morrell
Conceptual integrity and counting lines of code
Flight attendants freaked out that Google is buying tons of Spirit employee data
Meta ran ads for an app promising to nudify female politicians
LongNovel: A Multi-Scale Benchmark for Hallucination Detection in Long-Context Novel Summarization
Entity tracking emerges in sub-billion parameter language models and exceeds human performance in naturalistic narratives
Compiler-Guided Adaptive Proof Search with Cross-Model Synergy on Context-Dependent Theorem Proving
Persona-Guided LLM Agents for Task-Oriented Dialogue
SuTRA : Structurally-Unified Tokenization with Root Awareness
Latent Space Refusal Anchoring for Low-Resource African Languages: Mechanistic Safety Recovery Without Retraining
Nine Emotion Centroids: A Label-Free Valence Axis That Transfers Across Four Modalities
Self- and Other-Labels Induce Bidirectional Bias in LLM Judges
Abliteration Mitigation via Refusal Aliases
NE-BERT: A Multilingual Language Model for Nine Northeast Indian Languages
Backdoor Learning in Language Models and Vision-Language Models
MAVEN: A Macro-Societal Value Evaluation Framework of Multimodal Content with Compact Aligned Evaluators
FrenchNews-7: Benchmarking Cross-Publisher French News Editorial Desk Classification
Fractional Decay KV-Cache: Ownership-Aware Memory Management for Improved Inference Relevancy in Dialog Systems
Computational Orientalism: Measuring Structural Discourse Bias in Large Language Models Using the Middle East Cultural Sensitivity Score (MECSS)
BERTilda: Explainable Topic Lifecycle Tracking with Split/Merge Detection via Similarity-and-Flow Temporal Graphs
Stability-Aware Feature Design for Robust Watermark Detection in Machine-Generated Text
DeepTCM1.0: A Multi-Expert AI Agent for Deciphering Mechanisms of Chinese Herbal Formulae Based on General Large Language Models
StocksTalk: A Voice-Enabled Conversational Agent for Structured Query Generation over Web Data
Different Facets of Verbalised Overconfidence: an Interpretability Study
[AINews] Death of Params: Z.ai CEO Jie Tang on GLM 5.3 and the new Post-training Scaling Law
It’s Greg Brockman’s OpenAI now
Welcome to the AI crisis in math
Slack is launching collaborative vibe-coding channels
Google Gemini is getting a dedicated student hub
Linkdaze’s smart calendar is built to run a household, not just track a schedule
Grok keeps sending gibberish responses to users
A third of web pages published since ChatGPT’s launch show signs of AI authorship, study finds
Ramp launches its own AI model router, called Router
Meta brings Pocket, an app that lets you vibe-code and share games, to US users
Inertia Enterprises finds a way to make its fusion fuel fast
Meta AI’s new Mac app wants you to talk to your apps
Binance now lets AI agents trade, but keeping them in check is largely up to users
Stripe didn’t really buy OpenRouter because of the ‘singularity’
OpenAI seeks to one-up Anthropic with new customer privacy protections
Cognition CEO denies report that SpaceX tried to acquire the startup
AI was supposed to win people over by now — it hasn’t
Google packs Search and Gemini with new AI study tools
Proactive Road Safety Intervention in Australia: Predicting Risky Driving Hotspots from Connected Vehicle Data
Detecting and Discriminating Operator Misspecification in Hybrid PDE-Parameter Learning: a Reference-Free Instrument, with Discrimination Bounded In Sample
Data-DPO: Direct Preference Optimization for Target Model Data Selection in LLM Post-Training
AliExpress runs silent WebAudio fingerprinting that breaks Bluetooth multipoint
I should have loved biology
HTML Can Do That
Linux 7.2
CIA funding helped keep NeXT afloat in the 80s
Show HN: I trained a 125M model to autocomplete piano on-device
How to compromise your system with a job interview
Show HN: We chased a weather balloon across Montana and never found it
Sixtyfour (YC P25) Is Hiring
Malicious Rust crate Arrayref runs a build-time payload
DiffusionGemma Technical Report
Construir este día costó 58.110 tokens de entrada y 0 de salida en 63 llamadas a modelos — unos 0,0581 $.