| tech.blog / monetary.live / ivan.digital
February 03, 2026 at 19:12 UTC
🤗 HuggingFace Daily Papers
SLIME: Stabilized Likelihood Implicit Margin Enforcement for Preference Optimization
▲ 26 🏢 Floating Point Sigma Lab
  • SLIME is a novel reference-free alignment objective for large language models that decouples preference learning from generation quality through a thr
Rethinking Selective Knowledge Distillation
▲ 17 🎓 Tel Aviv University
  • Selective knowledge distillation in autoregressive language models using student-entropy-guided position selection improves accuracy and efficiency wh
Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics
▲ 10 🎓 Zhejiang University
  • Large language model control methods are unified under a dynamic weight update framework, revealing a preference-utility trade-off and enabling improv
Generative Visual Code Mobile World Models
▲ 10 🏢 Trillion Labs
  • Visual world models for mobile GUI agents are improved through renderable code generation using vision-language models, achieving better performance w
SPARKLING: Balancing Signal Preservation and Symmetry Breaking for Width-Progressive Learn
▲ 9 🎓 Peking University
  • SPARKLING is a framework for mid-stage width expansion in deep learning models that maintains signal preservation and breaks symmetry to stabilize tra
Large Language Models 6Direct Preference Optimization 1Reinforcement Learning From Human Feedback 1Rlhf 1Llms 1Implicit Reward Functions 1Objective Mismatch 1Relative Margin 1Absolute Likelihood 1Unlearning 1Formatting Collapse 1Reference-Free Alignment 1
🏛️ Top Research Institutions
ConStruM: A Structure-Guided LLM Framework for Context-Aware Schema Matching
🏛️ University of Michigan Systems & Infrastructure
  • Problem: The rapid expansion of autonomous vehicle technology necessitates a robust operating system to manage complex vehicle co
  • Solution: DAVOS, an autonomous vehicle operating system designed to transform vehicles into mobile computing platforms that suppor
  • Application: Enhancing the design and deployment of autonomous vehicles, enabling advanced functionalities and improved user experien
Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve
🏛️ Google Software & Programming
  • Problem: Modern compilers struggle to adapt to the complexity of software and hardware due to reliance on hand-crafted heuristics
  • Solution: Magellan, a framework that autonomously discovers novel compiler optimization heuristics using AlphaEvolve.
  • Application: Enhancing compiler performance and adaptability, leading to more efficient software optimization processes.
AEGIS: White-Box Attack Path Generation using LLMs and Training Effectiveness Evaluation f
🏛️ Cyber Defence Test and Evaluation Centre (CyTEC) Applications
  • Problem: The growing prevalence of misinformation generated by large language models poses significant risks to digital ecosystem
  • Solution: A comprehensive framework for understanding and countering the effects of LLM-generated misinformation on various platfo
  • Application: Developing effective strategies for mitigating misinformation in digital communication, enhancing trust in online conten
AI & Machine Learning
384 papers 7 cats
Systems & Infrastructure
23 papers 6 cats
Software & Programming
11 papers 4 cats
Theory & Algorithms
21 papers 5 cats
Applications
76 papers 7 cats
Other CS
19 papers 8 cats
Quantitative Finance
8 papers 9 cats
Economics
6 papers 3 cats