Long-Context GRPO

HippoRAG: Neurobiologically Inspired Long-Term Memory for LLMs (2024)

Learning to Plan and Reason for Evaluation with Thinking-LLM-as-a-Judge

Process Reinforcement Through Implicit Rewards

Explaining Large Language Models Decisions Using Shapley Values

Phi-4 Technical Report

Alignment Faking in LLMs [pdf]

What Makes Rotary Positional Encodings Useful?

Rethinking Softmax: Self-Attention with Polynomial Activations

Post-Training Layer Scaling Prevents Forgetting and Enhances Model Merging

Random Matrix Theory in Machine Learning Tutorial

Rerankers: A Lightweight Python Library to Unify Ranking Methods

Double Descent Demystified

Synthetic Continued Pretraining

Bright: A Realistic and Challenging Benchmark for Reasoning-Intensive Retrieval

Artificial needles to real haystacks: Improving retrieval capabilities in LLMs

From Decoding to Meta-Generation: (LLMs)

Warp: On the Benefits of Weight Averaged Rewarded Policies

Experiments in Weak-to-Strong Generalization

NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models

A Case Study in CUDA Kernel Fusion

Lessons from the trenches on reproducible evaluation of language models

Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet

Zero-Shot Tokenizer Transfer

An Empirical Model of Large-Batch Training

Gradient Diversity: A Key Ingredient for Scalable Distributed Learning

Arctic-Embed: Scalable, Efficient, and Accurate Text Embedding Models

Automatically Detecting Under-Trained Tokens in Large Language Models

Large Language Models for Data Annotation: A Survey

Refusal in LLMs is mediated by a single direction

Automated Multi Agent Chat

Orca: A Distributed Serving System for Transformer-Based Generative Models

Understanding Emergent Abilities of Language Models from the Loss Perspective

LoRA+: Efficient Low Rank Adaptation of Large Models

Does Transformer Interpretability Transfer to RNNs?

MiniCPM: Potential of Small Language Models W Scalable Training Strategies

Building BerkeleyDB

Rotational Equilibrium: How Weight Decay Balances Learning Across NeuralNetworks

Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference

Bad arguments against a universal basic income

The MOOC revolution that wasn’t

Tech industry's persistent claim of worker shortage may be phony

Why We've Decided to Organize

Ask HN: Where can I get info about who voted (up or down) articles/comments ?

Ask HN: How are Reddit and HN different for you?