🦥 Hey!

Daniel (Unsloth) PRO

danielhanchen

unsloth

·

https://unsloth.ai/

AI & ML interests

None yet

Recent Activity

new activity about 1 hour ago

unsloth/Inkling-Small-GGUF:Does it support image input?

posted an update about 1 hour ago

We compared 1-bit Kimi K3 to Claude Opus 5 and GPT 5.6. 🤯 We gave 4 models the same prompt: Create a glass aquarium whose side panel develops a visible crack and then bursts... 1-bit Kimi K3 GGUF ran locally on 4x B200s at 36 tok/s. GGUF: https://huggingface.co/unsloth/Kimi-K3-GGUF GitHub repo: https://github.com/unslothai/unsloth

repliedto their post about 1 hour ago

Kimi K3 can now be run locally! ✨ The 1-bit model retains ~78.9% accuracy after we shrunk it from 1.56TB to 594GB (-62% size). Run on a Mac Studio connected with 128GB RAM device. Kimi K3 is the strongest open model to date. GGUF: https://huggingface.co/unsloth/Kimi-K3-GGUF Guide: https://unsloth.ai/docs/models/kimi-k3

View all activity

Organizations

upvoted a collection 21 days ago

NVFP4

Dynamic Unsloth NVFP4 Quants. Faster and more accurate. • 10 items • Updated 13 days ago • 50

upvoted an article about 2 months ago

Article

The Open Source Community is backing OpenEnv for Agentic RL

+18

burtenshaw, spisakjo, lysandre, darktex, willcb, qjoy, pawalt, cwing-nv, danielhanchen, andrewzhou, thegovind, shimmyshimmer, Hamid-Nazeri, Sanyam, zkwentz, emre0, lewtun, sergiopaniego, banghua, unseenmars

•

Jun 8

• 109

upvoted a collection about 2 months ago

Gemma 4 QAT

Gemma 4 QAT (Quantization-Aware Training) for 3x less memory use and near original accuracy. • 16 items • Updated 13 days ago • 114

upvoted an article 4 months ago

Article

Safetensors is Joining the PyTorch Foundation

mcpotato, lysandre

•

Apr 8

• 41

upvoted a collection 4 months ago

Gemma 4

Gemma 4 is Google's new model family including including E2B, E4B, 26B-A4B, and 31B. • 43 items • Updated 13 days ago • 250

upvoted an article 4 months ago

Article

Welcome Gemma 4: Frontier multimodal intelligence on device

+5

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

•

Apr 2

• 918

upvoted a collection 5 months ago

Agentic RL Hackathon (SF) 2026

131 items • Updated 16 days ago • 12

upvoted 2 articles 5 months ago

Article

Community Evals: Because we're done trusting black-box leaderboards over the community

+5

burtenshaw, SaylorTwift, kramp, merve, davanstrien, nielsr, julien-c

•

Feb 4

• 90

Article

Mixture of Experts (MoEs) in Transformers

+5

ariG23498, pcuenq, merve, IlyasMoutawwakil, ArthurZ, sergiopaniego, Molbap

•

Feb 26

• 173

upvoted a collection 5 months ago

Qwen3.5

Qwen3.5 is Qwen's new model family including Qwen3.5 Small: 0.8B, 2B, 4B, 9B and Qwen3.5 Medium: 35B-A3B, 27B, 122B-A10B and 397B-A17B. • 25 items • Updated 13 days ago • 165

upvoted 2 articles 5 months ago

Article

GGML and llama.cpp join HF to ensure the long-term progress of Local AI

+4

ggerganov, ngxson, allozaur, lysandre, victor, julien-c

•

Feb 20

• 508

Article

Train AI models with Unsloth and Hugging Face Jobs for FREE

+4

burtenshaw, danielhanchen, shimmyshimmer, mlabonne, davanstrien, evalstate

•

Feb 20

• 104

upvoted an article 6 months ago

Article

Transformers v5: Simple model definitions powering the AI ecosystem

+2

lysandre, ArthurZ, cyrilvallez, reach-vb

•

Dec 1, 2025

• 313

upvoted a collection 7 months ago

Unsloth Diffusion GGUFs

Find GGUFs and other variants of diffusion based models like Qwen-Image and FLUX. • 20 items • Updated 13 days ago • 98

upvoted 2 collections 8 months ago

Magic Quant

MagicQuant is a benchmark-driven GGUF evaluation and hybrid-discovery system. https://github.com/magiccodingman/MagicQuant-Wiki • 5 items • Updated May 26 • 33

Ministral 3

Mistral Ministral 3: new multimodal models in Base, Instruct, and Reasoning variants, available in 3B, 8B, and 14B sizes. • 36 items • Updated 13 days ago • 35

upvoted an article 8 months ago

Article

Introducing Cogito v2.1

deepcogito

•

Nov 19, 2025

• 17

upvoted a paper 9 months ago

Efficient Long-context Language Model Training by Core Attention Disaggregation

Paper • 2510.18121 • Published Oct 20, 2025 • 124

upvoted a paper about 1 year ago

Speechless: Speech Instruction Training Without Speech for Low Resource Languages

Paper • 2505.17417 • Published May 23, 2025 • 14

upvoted a collection about 1 year ago

TorchAO Quantized Phi-4-mini-instruct

TorchAO quantized Phi-4-mini-instruct models from PyTorch team, runnable in A100, H100 through vLLM and in mobile devices through ExecuTorch • 7 items • Updated Dec 16, 2025 • 3