alkinun

AtAndDev

·

alkinun

AI & ML interests

decentralize

Recent Activity

liked a model about 1 hour ago

s3nh/Tess-4-27B-abliterated

reacted to s3nh's post with 👀 about 1 hour ago

Existing methods — GPTQ, AWQ, llama.cpp's k-quants — minimize empirical loss heuristically. None of them prove they are optimal in any information-theoretic sense. ICRB-Q builds a quantization scheme that is provably optimal via the Cramér-Rao lower bound (CRB): no unbiased estimator of a weight can have lower variance than [F(θ)]⁻¹, where F is the Fisher information matrix.

reacted to s3nh's post with ❤️ about 1 hour ago

Existing methods — GPTQ, AWQ, llama.cpp's k-quants — minimize empirical loss heuristically. None of them prove they are optimal in any information-theoretic sense. ICRB-Q builds a quantization scheme that is provably optimal via the Cramér-Rao lower bound (CRB): no unbiased estimator of a weight can have lower variance than [F(θ)]⁻¹, where F is the Fisher information matrix.

View all activity

Organizations

upvoted a paper about 9 hours ago

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

Paper • 2508.06471 • Published Aug 8, 2025 • 213

upvoted a collection 10 days ago

🤏 Smol-Data

Tried and tested mixes for strong pretraining. Inspired by https://huggingface.co/blog/codelion/optimal-dataset-mixing • 14 items • Updated Mar 2 • 14

upvoted a collection 19 days ago

SmolLM3 pretraining datasets

datasets used in SmolLM3 pretraining • 15 items • Updated Aug 12, 2025 • 54

upvoted a paper about 2 months ago

Embarrassingly Simple Self-Distillation Improves Code Generation

Paper • 2604.01193 • Published Apr 1 • 56

upvoted 2 articles 2 months ago

Article

Model2Vec: Distill a Small Fast Model from any Sentence Transformer

Pringled

•

Oct 14, 2024

• 105

Article

LLM Inference on Edge: A Fun and Easy Guide to run LLMs via React Native on your Phone!

medmekk, marcsun13

•

Mar 7, 2025

• 98

upvoted an article 3 months ago

Article

Granite 4.1 LLMs: How They’re Built

ibm-granite

•

Apr 29

• 83

upvoted 2 articles 4 months ago

Article

Welcome Gemma 4: Frontier multimodal intelligence on device

+5

merve, pcuenq, sergiopaniego, burtenshaw, Steveeeeeeen, alvarobartt, SaylorTwift

•

Apr 2

• 916

Article

Cosmopedia: how to create large-scale synthetic data for pre-training Large Language Models

+1

loubnabnl, anton-l, davanstrien

•

Mar 20, 2024

• 115

upvoted 2 papers 5 months ago

SWE-rebench V2: Language-Agnostic SWE Task Collection at Scale

Paper • 2602.23866 • Published Feb 27 • 92

Z1: Efficient Test-time Scaling with Code

Paper • 2504.00810 • Published Apr 1, 2025 • 27

upvoted a collection 5 months ago

Qwen3.5

21 items • Updated Mar 9 • 1.72k

upvoted an article 5 months ago

Article

Train AI models with Unsloth and Hugging Face Jobs for FREE

+4

burtenshaw, danielhanchen, shimmyshimmer, mlabonne, davanstrien, evalstate

•

Feb 20

• 103

upvoted 2 articles 6 months ago

Article

Why Your AI Strategy Needs Hugging Face Storage

AdrianLepers

•

Jan 26

• 13

Article

One Year Since the “DeepSeek Moment”

huggingface

•

Jan 20

• 62

upvoted a changelog 6 months ago

Hugging Face Changelog

Sort Models by Parameter Size

Jan 22

• 39

upvoted a collection 9 months ago

Moshi v0.1 Release

MLX, Candle & PyTorch model checkpoints released as part of the Moshi release from Kyutai. Run inference via: https://github.com/kyutai-labs/moshi • 16 items • Updated 2 days ago • 244

upvoted an article 10 months ago

Article

GRPO for GUI Grounding Done Right

HelloKKMe

•

Jun 11, 2025

• 37

upvoted a collection 10 months ago

Kimi-VL-A3B

Moonshot's efficient MoE VLMs, exceptional on agent, long-context, and thinking • 6 items • Updated Mar 2 • 82

upvoted a changelog 10 months ago

Hugging Face Changelog

Emoji Autocomplete in Discussions and Posts

Sep 11, 2025

• 69