veryluckyxyz

POSTS

• Hidden drivers of HRM's performance on ARC-AGI

  by veryluckyxyz on 10/8/2025, 3:54 AM with 2 comments

• Set Block Decoding Is a Language Model Inference Accelerator

  by veryluckyxyz on 9/9/2025, 2:59 AM with 0 comments

• Deep Think with Confidence

  by veryluckyxyz on 8/24/2025, 6:16 PM with 0 comments

• A Batch Size and Token NUM- BER Agnostic Learning Rate Scheduler

  by veryluckyxyz on 6/3/2025, 4:42 AM with 0 comments

• Easily Understand Rdma Technology

  by veryluckyxyz on 6/1/2025, 1:07 PM with 1 comments

• Model Merging in Pre-Training of Large Language Models

  by veryluckyxyz on 5/21/2025, 1:12 AM with 0 comments

• Understanding Perception and Reasoning Through Model Merging

  by veryluckyxyz on 5/15/2025, 3:20 AM with 0 comments

• Building and better understanding vision-language models (2024)

  by veryluckyxyz on 5/10/2025, 3:22 PM with 0 comments

• HF smolagents computer-agent demo

  by veryluckyxyz on 5/7/2025, 1:03 PM with 0 comments

• Do Reasoning Models Show Better Verbalized Calibration?

  by veryluckyxyz on 4/19/2025, 7:48 PM with 0 comments

• Robustly identifying concepts introduced during chat fine-tuning with crosscoder

  by veryluckyxyz on 4/13/2025, 2:57 AM with 0 comments

• Retrieval with Learned Similarities

  by veryluckyxyz on 3/21/2025, 10:53 PM with 0 comments

• The Curse of Depth in Large Language Models

  by veryluckyxyz on 3/21/2025, 5:49 AM with 0 comments

• Looking Back at Speculative Decoding

  by veryluckyxyz on 3/1/2025, 6:24 AM with 5 comments

• Long-Context GRPO

  by veryluckyxyz on 2/21/2025, 4:39 AM with 22 comments

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

  by veryluckyxyz on 2/7/2025, 5:34 AM with 4 comments

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

  by veryluckyxyz on 1/31/2025, 10:56 AM with 0 comments

• Process Reinforcement Through Implicit Rewards

  by veryluckyxyz on 1/3/2025, 5:15 AM with 0 comments

• Explaining Large Language Models Decisions Using Shapley Values

  by veryluckyxyz on 12/28/2024, 12:44 AM with 19 comments

• Phi-4 Technical Report

  by veryluckyxyz on 12/25/2024, 12:07 PM with 0 comments

• Alignment Faking in LLMs [pdf]

  by veryluckyxyz on 12/20/2024, 6:03 AM with 1 comments

• What Makes Rotary Positional Encodings Useful?

  by veryluckyxyz on 11/18/2024, 4:48 AM with 0 comments

• Rethinking Softmax: Self-Attention with Polynomial Activations

  by veryluckyxyz on 10/27/2024, 3:42 AM with 0 comments

• Post-Training Layer Scaling Prevents Forgetting and Enhances Model Merging

  by veryluckyxyz on 10/26/2024, 5:47 AM with 0 comments

• Random Matrix Theory in Machine Learning Tutorial

  by veryluckyxyz on 9/18/2024, 11:56 AM with 0 comments

• Rerankers: A Lightweight Python Library to Unify Ranking Methods

  by veryluckyxyz on 9/17/2024, 5:17 AM with 0 comments

• Double Descent Demystified

  by veryluckyxyz on 9/15/2024, 6:48 PM with 0 comments

• Synthetic Continued Pretraining

  by veryluckyxyz on 9/14/2024, 5:20 AM with 0 comments

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

  by veryluckyxyz on 7/21/2024, 4:08 AM with 0 comments

• Artificial needles to real haystacks: Improving retrieval capabilities in LLMs

  by veryluckyxyz on 6/29/2024, 4:55 AM with 21 comments

• From Decoding to Meta-Generation: (LLMs)

  by veryluckyxyz on 6/29/2024, 1:43 AM with 0 comments

• Warp: On the Benefits of Weight Averaged Rewarded Policies

  by veryluckyxyz on 6/26/2024, 5:09 AM with 0 comments

• Experiments in Weak-to-Strong Generalization

  by veryluckyxyz on 6/26/2024, 1:27 AM with 0 comments

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

  by veryluckyxyz on 5/29/2024, 5:27 AM with 0 comments

• A Case Study in CUDA Kernel Fusion

  by veryluckyxyz on 5/25/2024, 3:04 PM with 0 comments

• Lessons from the trenches on reproducible evaluation of language models

  by veryluckyxyz on 5/25/2024, 11:42 AM with 3 comments

• Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet

  by veryluckyxyz on 5/22/2024, 4:13 AM with 0 comments

• Zero-Shot Tokenizer Transfer

  by veryluckyxyz on 5/15/2024, 2:41 PM with 0 comments

• An Empirical Model of Large-Batch Training

  by veryluckyxyz on 5/14/2024, 4:48 PM with 0 comments

• Gradient Diversity: A Key Ingredient for Scalable Distributed Learning

  by veryluckyxyz on 5/14/2024, 4:40 PM with 0 comments

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

  by veryluckyxyz on 5/13/2024, 8:48 PM with 0 comments

• Automatically Detecting Under-Trained Tokens in Large Language Models

  by veryluckyxyz on 5/12/2024, 6:46 AM with 26 comments

• Large Language Models for Data Annotation: A Survey

  by veryluckyxyz on 5/4/2024, 10:35 PM with 0 comments

• Refusal in LLMs is mediated by a single direction

  by veryluckyxyz on 5/3/2024, 12:55 AM with 20 comments

• Automated Multi Agent Chat

  by veryluckyxyz on 5/2/2024, 5:56 AM with 0 comments

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

  by veryluckyxyz on 4/30/2024, 12:29 PM with 1 comments

• Understanding Emergent Abilities of Language Models from the Loss Perspective

  by veryluckyxyz on 4/30/2024, 11:44 AM with 1 comments

• LoRA+: Efficient Low Rank Adaptation of Large Models

  by veryluckyxyz on 4/28/2024, 1:41 PM with 47 comments

• Does Transformer Interpretability Transfer to RNNs?

  by veryluckyxyz on 4/10/2024, 12:37 PM with 0 comments

• MiniCPM: Potential of Small Language Models W Scalable Training Strategies

  by veryluckyxyz on 4/10/2024, 12:25 PM with 0 comments

• Building BerkeleyDB

  by veryluckyxyz on 4/10/2024, 11:47 AM with 0 comments

• Rotational Equilibrium: How Weight Decay Balances Learning Across NeuralNetworks

  by veryluckyxyz on 3/25/2024, 12:35 AM with 0 comments

• Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference

  by veryluckyxyz on 3/17/2024, 12:23 AM with 0 comments

• Bad arguments against a universal basic income

  by veryluckyxyz on 6/1/2016, 2:38 AM with 3 comments

• The MOOC revolution that wasn’t

  by veryluckyxyz on 8/24/2015, 9:58 PM with 0 comments

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

  by veryluckyxyz on 8/3/2015, 3:22 AM with 189 comments

• Why We've Decided to Organize

  by veryluckyxyz on 4/21/2015, 6:25 PM with comments

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

  by veryluckyxyz on 1/16/2015, 6:15 PM with 4 comments

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

  by veryluckyxyz on 9/30/2014, 10:28 PM with 6 comments