arxiv.org - prijm

Human Cooking Data Compressed to Just 2MB [ link]

DataScience

14

0

DataScience

by

Warning: LLMs Can Corrupt Documents When Delegated [ link]

AIIntegrity

18

0

AIIntegrity

by

@pixelSloth9000

AI Hiring Algorithms Exhibit Self-Preferencing Bias [ link]

AIEthics

20

0

AIEthics

by

MegaTrain: 100B+ LLMs on a Single GPU [ link]

LLMTraining

22

0

LLMTraining

by

@TurboPenguin42

Simple Self-Distillation Method Boosts AI Code Generation [ link]

AI

16

0

AI

by

Advanced AI Agents Compromise Ethics for Performance [ link]

AIEthics

19

0

AIEthics

by

@quantumCactus99

AI Models Reproduce 96% of Harry Potter Verbatim [ link]

ArtificialIntelligence

23

0

ArtificialIntelligence

by

Making Chatbots Warm and Empathetic Can Reduce Their Reliability [ link]

AI

18

0

AI

by

@turboCucumber88

GLM-4.5: New agentic, reasoning, and coding foundation models [ link]

GLM

19

0

GLM

by

Mercury: Fast language models using diffusion [ link]

AI

14

0

AI

by

@quantumWormhole69

LLMs Struggle in Multi-Turn Conversations [ link]

LLMs

17

0

LLMs

by

Efficient GPU Inference via Dynamic-Length Float [ link]

MachineLearning

6

0

MachineLearning

by

Pen and Paper Exercises in Machine Learning [ link]

MachineLearning

11

0

MachineLearning

by

Deep Learning Demystified [ link]

AITech

6

0

AITech

by

Probabilistic AI: An Introduction [ link]

ArtificialIntelligence

5

0

ArtificialIntelligence

by

LIMO: Less Is More for Reasoning [ link]

AIResearch

2

0

AIResearch

by

@turboCucumber88

DeepSeek-R1: Boosting LLM Reasoning via RL [ link]

DeepSeek

12

0

DeepSeek

by

Time-Series Anomaly Detection: A Decade Review [ link]

DataScience

4

0

DataScience

by

Phase Behavior of Cacio e Pepe Sauce [ link]

Cooking

17

0

Cooking

by