How I Backed Into AI Research by Trying to Poison My Own Model
Travis-ML
• NEW Articles from Team or Enterprise organizations will get promoted to the main section.
Taking Alpamayo to New Heights with Driving Foundation Models and Closed-Loop Training
drmapavone
• • 2A Deep Neural Network that turns Any Image into a Playable Game! All on consumer GPUs, NOT DATACENTER
lucidml
• • 2From Data Repositories to Production Data Pipelines: Bridging Hugging Face Datasets and Dagster with dagster-hf-datasets
AINovice2005
• • 2ClawHub Security Signals: Large Corpus Multi-Scanner Dataset for Agent Skill Security Research
OpenClaw
• • 14Block Diffusion on Apple Silicon with 3.7× Speedup for Qwopus 3.6 27B
junafinity
• Reading the Voids: Topological Contribution Signals in Frozen Geometric Codebooks
AbstractPhil
• Objective Projection v7.2: Closing the Pattern F Gap and Extending the Hard Negatives
leventbulut
• Targeted Refusal Modification (TRM): Precision Separation of Safety and Harm in Large Language Models
senaro
• Sparse Participation Fine-Tuning: Training Small Models Without Moving the Whole Beast
reaperdoesntknow
• • 1Flash Load: Smart Lazy Loading for 200GB Step 3.7 Flash Q8 MLX on <100GB RAM
junafinity
• Summarization Bias: Why Language Models Re-Label the Emotions You Tried to Hide
leventbulut
• • 1Does Depth Actually Help Reasoning? A Tiny Experiment on 2× T4
wop
• • 2Causal Expert Editing in Laguna XS.2 for Cyber Triage
sourishsharma17
• • 1Agents on AWS SageMaker AI with Strands Agents
alvarobartt
• Can Predicted Dynamics Exist in the Physical World?
barakor
• • 1Objective Projection: Engineering Emotion in Text with Physical Parameters Instead of Emotion Labels
leventbulut
• Agentic RL: Token-In, Token-Out Done Right
huggingface
• • 10The geolip-svae-transformer
AbstractPhil
• Why One Bug Fix Made My Model the Most Downloaded Japanese LLM in MLX Format
tocchitocchi
•