Fast360 – Posts

Join the conversation

Join the community of Machine Learners and AI enthusiasts.

All HF Hub posts

danielhanchen

posted an update 2 days ago

Post

3616

Run DeepSeek-V3.1 locally on 170GB RAM with Dynamic 1-bit GGUFs!🐋
GGUFs: unsloth/DeepSeek-V3.1-GGUF

The 715GB model gets reduced to 170GB (-80% size) by smartly quantizing layers.

The 1-bit GGUF passes all our code tests & we fixed the chat template for llama.cpp supported backends.

Guide: https://docs.unsloth.ai/basics/deepseek-v3.1

Xenova

posted an update 2 days ago

Post

1355

Okay this is insane... WebGPU-accelerated semantic video tracking, powered by DINOv3 and Transformers.js! 🤯
Demo (+ source code): webml-community/DINOv3-video-tracking

This will revolutionize AI-powered video editors... which can now run 100% locally in your browser, no server inference required (costs $0)! 😍

How does it work? 🤔
1️⃣ Generate and cache image features for each frame
2️⃣ Create a list of embeddings for selected patch(es)
3️⃣ Compute cosine similarity between each patch and the selected patch(es)
4️⃣ Highlight those whose score is above some threshold

... et voilà! 🥳

You can also make selections across frames to improve temporal consistency! This is super useful if the object changes its appearance slightly throughout the video.

Excited to see what the community builds with it!

ProCreations

posted an update 1 day ago

Post

1671

If @clem comments on this post within the week with a task for me to classify with text and a paramater size for the model, within 48 hours I will create a new dataset, train the model, and post it all. Paramater size must be bellow 50 million params, and task can be text only, and genuinely possible.

if I complete it, Clem must tell everyone on Twitter/x to follow me on huggingface and link it

let's see if he comments

1 reply

daqc

posted an update 1 day ago

Post

1454

Just applied for HF Community Grant for “Hugging Research” — a lightweight CodeAgent‑based research assistant built on Hugging Face’s Open Deep Research project for the Hugging Face Hub (models, datasets, Spaces, users, collections, papers). It gathers links via dedicated tools and organizes them for easy review.

As this is for the community, comments and suggestions are appreciated: daqc/hugging-research#1

MonsterMMORPG

posted an update about 20 hours ago

Post

1412

Qwen Image Edit Full Tutorial: 26 Different Demo Cases, Prompts & Images, Pwns FLUX Kontext Dev

Tutorial Link
https://youtu.be/gLCMhbsICEQ

Extra Info
I tested newly arrived Qwen-Image-Edit-Lightning-8steps-V1.0 (arrived after tutorial recorded) and definitely our existing preset which uses Qwen-Image-Lightning-8steps-V1.1 is better than it, so this tutorial and presets are still 100% best quality and up-to-date

Info
Qwen Image Edit just has been published and since then I have been experimenting to prepare you this amazing tutorial. I have literally shown 26 unique cases and provided demo images and prompts. After watching this tutorial your image editing skills will move to next level i promise you that. Also this tutorial will give you a lot of ideas.

Reubencf

posted an update 3 days ago

Post

2153

Releasing Version 1 of the Konkani LLM

Konkani is a low resource Indian language spoken by 2.5 Million people

Model:https://huggingface.co/Reubencf/gemma3-konkani

space: Reubencf/Gemma3-konkani

2 replies

kanaria007

posted an update 1 day ago

Post

1340

✅ New Article: *Psychology as Structured Mind*

Title:
🧩 Psychology: Identity Constructs and Constraint-Based Emotion
🔗 https://huggingface.co/blog/kanaria007/structured-psychology

---

Summary:
Psychology is often framed as *feelings, behavior, and personality*.
Structured Intelligence reframes the mind as *recursive identity architecture*:

* Emotion as *state feedback in cognitive loops*
* Behavior as *jump execution under constraints*
* Personality as *long‑term stability of structural patterns*

> The psyche isn’t a mystery —
> *it’s the architecture of self in motion.*

---

Why It Matters:
• Explains *how emotion, memory, and decision interact structurally*
• Supports *reflective AI that models affect without mimicry*
• Bridges *clinical insight, cognitive science, and structured reasoning*

---

What’s Inside:
• Psyche as *identity‑bound jump and feedback system*
• *Emotional signal processing* as structural state update
• *Memory loops* enabling personality and self‑coherence
• Implications for *therapy, AI companions, and cognitive modeling*

---

📖 Article 25 of the Structured Intelligence Series

Where Article 24 explored *natural law as structural constraint*,
Article 25 frames *psychology as structured mind* —
showing that *emotion and behavior emerge from recursive architecture*.

---

Next: Relationships as Structural Reciprocity
The next article examines *how interpersonal connection works as multi‑agent structural alignment*,
revealing *love, trust, and betrayal as signals of recursive constraint dynamics*.

> From self to others,
> *structure decides how minds meet without collapse.*

5 replies

ZacMasa5000

posted an update 3 days ago

Post

2503

New Hugging Face Dataset is LIVE 🤗
In this dataset, you’ll find 20,000+ rows of the top trending posts on X over the past several weeks.

🔥Topics Include:
🔹Bitcoin
🔹GPT-5
🔹Grok
🔹Cursor
🔹AI, Agents + Prompts
🔹Chainlink $LINK

📂Download dataset!
Easily, plug it into your AI agents, LLMs or apps.
MasaFoundation/X_Twitter_Trending_Topics_August2025

You can start query fresh data in our app and request API key here: https://bit.ly/4mrqbO2

LPX55

posted an update 4 days ago

Post

577

Hey all, just a friendly PSA heads-up to those using our POC space made for showcasing the various novel possibilities that @qwen-llm 's Image Editing model can be used for:

- We do not collect or store any data of inferences by users. BUT (pun not intended) there's been a minor logic error in how we store session data for custom prompting for our presets feature.
- While we can't deny that we all like nice buttocks as natural human beings, for the time being please reset the prompts to the default state after use, just as a courtesy for the next user.
- As long as you are staying within the law and HF's terms of service, we do not judge for curious minds and the inevitable occasional ... urges driven by humanly hormones, which, our future AGI overlords will never have.

The session storage logic error is completely on me, and will be fixed tomorrow, after I get some sleep as I am severely sleep-deprived (unrelated reasons) and my brain is pretty much mush. But thank you for the chuckles, as it was admittedly a pretty funny way to be made aware of my mistake.

To the anon that left the custom prompts at this state, I hope the "buttocks" were satisfactory. On a more serious note, please do not take this light-hearted note and mistake it as complacency or disregard for obvious lines that should not be crossed. Buttocks, I suppose, are okay. But let's be responsible and keep everything ethical and legal. As long as common sense is used (i.e. you are not harming anyone or breaking any TOS or laws), enjoy all the AI generated buttocks as your heart desires.

The attached image is what everyone sees at the moment when a user tries out the preset batch prompting feature. Though, as mentioned, we also generally like a nice butt, we are clarifying to fellow users that may have come across this that... we are not into butts of Russian nuns. Butt we don't judge. Original preset prompts can be seen in presets.py but possibilities are endless.

1 reply

codelion

posted an update 20 days ago

Post

3608

🧬 We just published our comprehensive analysis of OpenEvolve - an open-source evolutionary coding agent that automatically optimizes algorithms using LLMs!

Our key findings from 29 experiments across 10 models:

- Gemini Flash 2.5 achieved 2.04x speedup across 30 benchmark tasks
- Open models like Gemma 3 27B (1.63x) and Qwen3-Coder 480B (1.41x) rivaled proprietary models
- The system discovered entirely new algorithms - not just code optimizations!
- One task evolved from DFS to BFS to Union-Find approaches
- Specialized coding models outperformed much larger general models 200 iterations beat 100 iterations by 24%
- Ensembles surprisingly failed due to conflicting optimization strategies

Most fascinating: watching models evolve code step-by-step, like transforming matrix operations from basic eigendecomposition to vectorized one-liners with 32x speedup.

Our systematic experimental approach reveals that open-source evolutionary coding is becoming seriously competitive with proprietary solutions. We tested everything from temperature settings to evolution strategies to find optimal configurations.

This research shows automated code optimization is ready for real-world applications. The methodology we developed can guide anyone building evolutionary coding systems.

Full paper with code examples, detailed methodology, and all experimental results: https://huggingface.co/blog/driaforall/towards-open-evolutionary-agents

What optimization challenges could benefit from evolutionary approaches in your work?

Recently active users