Episodes (Page 12)
✨
Introduced BAT, a benchmark for auto-bidding algorithms in online advertising.
✨
Highlighted Amazon's AI, ML, and robotics research.
Human Feedback Improvements
✨
Introduced T2I-R1 for text-to-image generation using RL and bi-level CoT.
✨
Introduces formula predicting LLM shifts to harmful content.
✨
Proposes pretraining for estimating heterogeneous treatment effects (HTE).
✨
Yo’Chameleon personalizes Large Multimodal Models (LMMs).
✨
AI integration in finance: benefits and risks for stability.
✨
Addresses reward hacking in LLM reasoning tasks with PRMs.
✨
Uses language models for automated patient record linkage.
✨
Surveys Parameter-Efficient Continual Fine-Tuning (PECFT).
✨
Building a code-editing agent requires LLM, loop, and tokens.
✨
Novel two-stage framework for lung tumor segmentation.
✨
Introduces Complex-Edit benchmark for image editing models.
✨
Pre-computes LLM inferences offline for faster query responses.
✨
Miras framework designs sequence models like associative memory.
✨
RUKA is an affordable, open-source humanoid robotic hand.
✨
GenEAva generates expressive cartoon avatars via diffusion.
✨
VCR-Bench evaluates video Chain-of-Thought reasoning for LVLMs.
✨
HalluMeasure automates hallucination detection in LLM outputs.
✨
Meta AI releases Llama 4 family of multimodal models.