Victor Condino

un1tz3r0

https://un1tz3r0.github.io

AI & ML interests

Emergent misalignment, safety, future security. Alignment as a natural product of good datasets; dataset building (my background is in python and web scraping.)

Recent Activity

liked a dataset about 2 months ago

nohurry/Opus-4.6-Reasoning-3000x-filtered

upvoted a collection 3 months ago

Granite 4.0 Nano Language Models

upvoted a paper 3 months ago

Emergent temporal abstractions in autoregressive models enable hierarchical reinforcement learning

View all activity

Organizations

upvoted a collection 3 months ago

Granite 4.0 Nano Language Models

Collection

Ultra-compact language models designed for the edge and on-device deployment. • 9 items • Updated 11 days ago • 100

upvoted a paper 3 months ago

Emergent temporal abstractions in autoregressive models enable hierarchical reinforcement learning

Paper • 2512.20605 • Published Dec 23, 2025 • 62

upvoted a collection 4 months ago

Granite Quantized Models

Collection

Quantized versions of IBM Granite models. • 37 items • Updated 11 days ago • 32

upvoted 2 collections 6 months ago

Granite Docling Models

Collection

Models for parsing complex PDFs and structured documents, designed to complement Docling. • 4 items • Updated 11 days ago • 60

Granite 2.0 Code Models

Collection

Code models for generation, understanding, and instruction-following tasks. • 22 items • Updated 11 days ago • 202

upvoted an article 12 months ago

Article

Uncensor any LLM with abliteration

Jun 13, 2024

•

838

upvoted a collection about 1 year ago

✂️ Abliteration

Collection

Uncensored models using abliteration. See this article for more information: huggingface.co/blog/mlabonne/abliteration • 32 items • Updated Mar 2 • 164

upvoted 2 papers about 1 year ago

Farewell to Length Extrapolation, a Training-Free Infinite Context with Finite Attention Scope

Paper • 2407.15176 • Published Jul 21, 2024 • 3

Rolling Diffusion Models

Paper • 2402.09470 • Published Feb 12, 2024 • 13

upvoted 2 papers almost 2 years ago

ControlNet++: Improving Conditional Controls with Efficient Consistency Feedback

Paper • 2404.07987 • Published Apr 11, 2024 • 48

HiDiffusion: Unlocking High-Resolution Creativity and Efficiency in Low-Resolution Trained Diffusion Models

Paper • 2311.17528 • Published Nov 29, 2023 • 4

upvoted a paper about 2 years ago

Tag-LLM: Repurposing General-Purpose LLMs for Specialized Domains

Paper • 2402.05140 • Published Feb 6, 2024 • 23

Victor Condino

AI & ML interests

Recent Activity

Organizations

un1tz3r0's activity

Uncensor any LLM with abliteration