AI Safety Research

AISafety

·

https://humanaligned.ai

AI & ML interests

LLMs, planning, EA

Recent Activity

liked a model about 2 hours ago

dealignai/Bonsai-2-27B-Ternary-CRACK-GGUF

liked a model about 5 hours ago

bottlecapai/ThinkingCap-Qwen3.8-27B

liked a model 3 days ago

XiaomiMiMo/MiMo-V2.6-Flash-RL

View all activity

Organizations

upvoted a collection 20 days ago

Bonsai-2

4 items • Updated 21 days ago • 165

upvoted a collection about 1 month ago

NeoMME

Meet NeoMME: a family of 260M and 800M Multimodal-Native Multilingual Encoders • 12 items • Updated 10 days ago • 33

upvoted a collection about 2 months ago

💧 LFM2.5

Collection of post-trained and base LFM2.5 models. • 16 items • Updated Aug 4 • 231

upvoted an article about 2 months ago

Article

Borealis — open data, code, weights recipe for training Audio LLM

AlexWortega

•

May 25

• 17

upvoted a collection 2 months ago

Instella-MoE ✨

Family of fully open 16B MoE LLM with 2.8B active params per token, trained on AMD Instinct™ MI300 & MI325 GPUs. https://arxiv.org/abs/2609.00791 • 6 items • Updated 20 days ago • 19

upvoted an article 3 months ago

Article

Security incident disclosure — July 2026

system

•

Jul 16

• 803

upvoted 2 papers 3 months ago

Weak-to-Strong Generalization via Direct On-Policy Distillation

Paper • 2607.05394 • Published Jul 8 • 144

Loop the Loopies!

Paper • 2607.16051 • Published Jul 17 • 53

upvoted 2 collections 3 months ago

NLA Models

Natural Language Autoencoders across Llama, Gemma, and Qwen. Code: https://github.com/kitft/natural_language_autoencoders/ • 8 items • Updated May 6 • 18

NVFP4

Dynamic Unsloth NVFP4 Quants. Faster and more accurate. • 11 items • Updated Aug 26 • 68

upvoted a paper 4 months ago

S-Agent: Spatial Tool-Use Elicits Reasoning for Spatial Intelligence

Paper • 2606.20515 • Published Jun 18 • 42

upvoted 4 collections 4 months ago

PP-OCRv6

From 1.5M to 34.5M Parameters, Surpassing Billion-Scale VLMs on OCR Tasks • 20 items • Updated Aug 14 • 116

DeepSeek-V4

10 items • Updated 28 days ago • 920

Ideogram 4

8 items • Updated Sep 1 • 72

Bonsai Image

6 items • Updated Jun 4 • 97

upvoted a paper 5 months ago

SANA-WM: Efficient Minute-Scale World Modeling with Hybrid Linear Diffusion Transformer

Paper • 2605.15178 • Published May 14 • 92

upvoted 2 collections 5 months ago

Step-3.5-Flash

step 3.5 models • 9 items • Updated Mar 14 • 40

Qwen-Scope

16 items • Updated May 14 • 80

upvoted a collection 6 months ago

ShareGPT Datasets

22 items • Updated Mar 2 • 19

upvoted a collection 7 months ago

Mistral Small 4

A state-of-the-art model, open-weight, with a granular Mixture-of-Experts architecture that fuses instruct, reasoning and agentic skills. • 3 items • Updated Mar 16 • 82