Victor Mustar's picture

Victor Mustar PRO

victor

AI & ML interests

Building the UX of this website

Recent Activity

Organizations

Hugging Face's profile picture Google's profile picture Safetensors's profile picture Competitions's profile picture 21 RNN's profile picture Spaces-explorers's profile picture Text Generation Inference's profile picture Spaces Examples's profile picture CVPR Demo Track's profile picture Hugging Chat's profile picture Webhooks Explorers (BETA)'s profile picture lora concepts library's profile picture Huggingface Projects's profile picture Scanned Tokens's profile picture hf admins's profile picture Hugging Face OSS Metrics's profile picture Stable Diffusion Dreambooth Concepts Library's profile picture Core ML Projects's profile picture temp-org's profile picture Blog-explorers's profile picture Mustarz's profile picture Open LLM Leaderboard's profile picture Enterprise Explorers's profile picture The Collectionists's profile picture ZeroGPU Explorers's profile picture Hugging Face Tools's profile picture TstOrg141's profile picture Stable Video benchmark's profile picture Social Post Explorers's profile picture Dev Mode Explorers's profile picture LLHF's profile picture SLLHF's profile picture Self-serve FTW's profile picture Inference Explorers's profile picture

victor's activity

reacted to nicolay-r's post with ๐Ÿ‘€ about 13 hours ago
view post
Post
1548
๐Ÿ“ข For those who start to work with LLM streaming in web, here is a minimalistic example in JS for accessing server hosted by FastAPI via REST:
https://gist.github.com/nicolay-r/840425749cf6d3e397da3d329e894d59

The code above is a revised verison for accessing Replicate API posted earlier
https://huggingface.co./posts/nicolay-r/390307941200307

The key difference from Replicate API:
- using only POST for passing a body with parameters and fetching the reader.
reacted to clem's post with ๐Ÿ”ฅ about 13 hours ago
view post
Post
1881
We crossed 1B+ tokens routed to inference providers partners on HF, that we released just a few days ago.

Just getting started of course but early users seem to like it & always happy to be able to partner with cool startups in the ecosystem.

Have you been using any integration and how can we make it better?

https://huggingface.co./blog/inference-providers
reacted to sayakpaul's post with ๐Ÿ”ฅ about 13 hours ago
view post
Post
1379
Inference-time scaling meets Flux.1-Dev (and others) ๐Ÿ”ฅ

Presenting a simple re-implementation of "Inference-time scaling diffusion models beyond denoising steps" by Ma et al.

I did the simplest random search strategy, but results can potentially be improved with better-guided search methods.

Supports Gemini 2 Flash & Qwen2.5 as verifiers for "LLMGrading" ๐Ÿค—

The steps are simple:

For each round:

1> Starting by sampling 2 starting noises with different seeds.
2> Score the generations w.r.t a metric.
3> Obtain the best generation from the current round.

If you have more compute budget, go to the next search round. Scale the noise pool (2 ** search_round) and repeat 1 - 3.

This constitutes the random search method as done in the paper by Google DeepMind.

Code, more results, and a bunch of other stuff are in the repository. Check it out here: https://github.com/sayakpaul/tt-scale-flux/ ๐Ÿค—
reacted to Quazim0t0's post with ๐Ÿ‘ about 13 hours ago
view post
Post
1131
My first attempt at using SmolAgents:
Quazim0t0/CSVAgent

The video attached was an example for this space.

Based on ZennyKenny's SqlAgent:
ZennyKenny/sqlAgent

You can upload a CSV file and it will automatically populate the table, then you can ask questions about the data.

Grab a sample CSV file here: https://github.com/datablist/sample-csv-files

The questions that can be asked may be limited.

_______________________
Second: Quazim0t0/TXTAgent
Created an Agent that converts a .txt file into a CSV file, then you can ask about the data and also download the CSV file that was generated.

_______________________
Third: Quazim0t0/ReportAgent
Upload Multiple TXT/DOC files to then generate a report from those files.

_______________________
Lastly: Quazim0t0/qResearch
A Research tool that uses DuckDuckGo for Web Searches, Wikipedia and tries to refine the answers in MLA Format.

reacted to prithivMLmods's post with ๐Ÿš€ about 13 hours ago
view post
Post
4127
The last week of Impression Craft Arts and sketches from strangerzonehf๐ŸŽจ๐Ÿง‘๐Ÿปโ€๐ŸŽจ

- Collection : strangerzonehf/Flux-Ultimate-LoRA-Collection

Adapters:
+ Ld-Art : strangerzonehf/Ld-Art
+ Animeopix-Flux : strangerzonehf/Animeopix-Flux
+ Flux-Super-Paint-LoRA : strangerzonehf/Flux-Super-Paint-LoRA
+ CinematicShot-Pics-Flux : strangerzonehf/cinematicShot-Pics-Flux
+ Oil-Wall-Art-Flux : strangerzonehf/Oil-Wall-Art-Flux
+ Pixelo-Flux : strangerzonehf/Pixelo-Flux
+ Abstract-Shattered : strangerzonehf/Abstract-Shattered
+ Neon-Impressionism-Flux : strangerzonehf/Neon-Impressionism-Flux
+ NewG-Art : strangerzonehf/NewG-Art

๐ŸชงDemo : prithivMLmods/FLUX-LoRA-DLC
๐Ÿค—Page : https://huggingface.co./strangerzonehf
reacted to ZennyKenny's post with ๐Ÿ‘ 1 day ago
view post
Post
1799
Okay this is pretty crazy. Snowflake has CortexAI and Uber is already teasing QueryGPT, both of which prominently feature plain text to SQL features to query your database.

I decided to see how hard it would be to put together something similar using ๐Ÿค— smolagents. Turns out, it was pretty straightforward. I managed to get it done in London Luton airport this afternoon.

ZennyKenny/sqlAgent
  • 2 replies
ยท
reacted to Jaward's post with ๐Ÿ‘€ 1 day ago
view post
Post
2732
Finally here it is: a faster, custom, scalable GRPO trainer for smaller models with < 500M params, can train on 8gb ram cpu, also supports gpu for sanity sake (includes support for vllm + flash attention). Using smolLM2-135M/360M-instructs as ref & base models. Experience your own โ€œahaโ€ moment ๐Ÿณ on 8gb ram.
Code: https://github.com/Jaykef/ai-algorithms/blob/main/smollm2_360M_135M_grpo_gsm8k.ipynb
  • 1 reply
ยท
reacted to Akjava's post with ๐Ÿ”ฅ 1 day ago
reacted to schuler's post with ๐Ÿ˜Ž 1 day ago
view post
Post
2883
๐Ÿ”ฎ GPT-3 implemented in pure Free Pascal!
https://github.com/joaopauloschuler/gpt-3-for-pascal

This implementation follows the GPT-3 Small architecture from the landmark paper "Language Models are Few-Shot Learners":
โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”
โ”‚     Input Layer       โ”‚
โ”œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค
โ”‚ Token & Positional    โ”‚
โ”‚     Embedding         โ”‚
โ”œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค
โ”‚   12x Transformer     โ”‚
โ”‚      Blocks           โ”‚
โ”‚  - 12 heads           โ”‚
โ”‚  - 768 hidden dims    โ”‚
โ”‚  - 3072 intermediate  โ”‚
โ”œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค
โ”‚   Output Layer        โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜

Clean Pascal Implementation
for CntLayer := 1 to {Layers=}12 do
begin
  Result.AddTransformerBlockCAI(
    {Heads=}12, 
    {intermediate dimensions=}4*768, 
    {NoForward=}true, 
    {HasNorm=}true, 
    false
  );
end;

reacted to onekq's post with ๐Ÿ‘ 4 days ago
view post
Post
1728
R1 is still trending. Here is a collection of works trying to replicate R1.
onekq-ai/r1-reproduction-works-67a93f2fb8b21202c9eedf0b

Players include Huggingface (Open R1), Stanford (simple scaling), Berkeley (Bespoke, Open thoughts, etc.), ServiceNow, etc. I know there is another work from HKUST but couldn't find it on ๐Ÿค—. Let me know if I miss any teams.
  • 5 replies
ยท
reacted to AdinaY's post with ๐Ÿš€ 4 days ago
view post
Post
2478
Ovis2 ๐Ÿ”ฅ a multimodal LLM released by Alibaba AIDC team.
AIDC-AI/ovis2-67ab36c7e497429034874464
โœจ1B/2B/4B/8B/16B/34B
โœจStrong CoT for deeper problem solving
โœจMultilingual OCR โ€“ Expanded beyond English & Chinese, with better data extraction
reacted to mrzjy's post with ๐Ÿ‘€ 4 days ago
view post
Post
1246
A very small project:

Introducing CreativeTinyZero:
mrzjy/Qwen2.5-1.5B-GRPO-Creative-Ad-Generation

Unlike the impressive DeepSeek-R1(-Zero), this project focuses on a pure reinforcement learning (RL) experiment applied to an open-domain task: creative advertisement generation.

Objective:

- To investigate the feasibility of applying R1-like methods to an open-domain task without a verifiable ground-truth reward, while at least demonstrating its potential.
- To explore whether <think> and <answer> rewards can be explicitly designed to provide strong guidance through RL based on human prior knowledge.

Note:
- Our goal is not to induce self-reflective thinking, but to align with human thought processes purely through RL, without any supervised fine-tuning (SFT) on any constructed dataset.

Despite its small size, the resulting 1.5B-GRPO model demonstrates intriguing generative capabilitiesโ€”though it's still far from perfect.
  • 1 reply
ยท
reacted to Duskfallcrew's post with ๐Ÿ‘ 4 days ago
view post
Post
1395
I don't have the stamina to port my articles tonight, i've been dealign with my CPTSD seizures again - but here's a fun update over on Bluesky!
https://bsky.app/profile/duskfallcrew.bsky.social/post/3li4zwdhy5c2q
HF's been my open source home since before i got on Civitai, and while i've largely left Civitai, i can't leave AI yet.
SO if y'all don't mind me trying to rebuild my "empire" one nerd block at a time, i'll keep my content easily accessible, :)

OH PSST New AI/ML Discord i made recently:
(It's also a shill for my main twitch/media/music hub)

Join us on this journey. Welcome to Ktiseos Nyx.

Our Discord:
https://discord.gg/HhBSvM9gBY

Earth & Dusk Media
https://discord.gg/5t2kYxt7An

:3 Cant' wait to hang out, and i've always linked back to HF for my E&D. content in terms of my lora backups and checkpoints!

Y'all who make diffusers versions of my content:
YOU ROCK. Do me a smidge favor: :3 aside from linking back can you maaaaaybe add the new K/N discord on there?

it's my geeky new AI safe space. XD
Also yea, if you've watched the new Beetlejuice movie, you know that i will never quit the ectoplasmic nerd train XD
  • 1 reply
ยท
reacted to as-cle-bert's post with ๐Ÿš€ 4 days ago
view post
Post
1336
๐’๐œ๐ข๐๐ž๐ฐ๐ฌ๐๐จ๐ญ - ๐‘๐ž๐ฉ๐จ๐ซ๐ญ ๐๐š๐ข๐ฅ๐ฒ ๐’๐œ๐ข๐ž๐ง๐œ๐ž ๐ง๐ž๐ฐ๐ฌ ๐จ๐ง ๐๐ฅ๐ฎ๐ž๐’๐ค๐ฒ

GitHub ๐Ÿ‘‰ https://github.com/AstraBert/SciNewsBot
BlueSky ๐Ÿ‘‰ https://bsky.app/profile/sci-news-bot.bsky.social

Hi there HF Community!๐Ÿค—
I just created a very simple AI-powered bot that shares fact-checked news about Science, Environment, Energy and Technology on BlueSky :)

The bot takes news from Google News, filters out the sources that are not represented in the Media Bias Fact Check database, and then evaluates the reliability of the source based on the MBFC metrics. After that, it creates a catchy headline for the article and publishes the post on BlueSky๐Ÿ“ฐ

The cool thing? SciNewsBot is open-source and is cheap to maintain, as it is based on mistralai/Mistral-Small-24B-Instruct-2501 (via Mistral API). You can reproduce it locally, spinning it up on your machine, and even launch it on cloud through a comfy Docker setup๐Ÿ‹

Have fun and spread Science!โœจ
reacted to m-ric's post with ๐Ÿš€ 4 days ago
view post
Post
2236
For those who haven't come across it yet, here's a handy trick to discuss an entire GitHub repo with an LLM:

=> Just replace "github" with "gitingest" in the url, and you get the whole repo as a single string that you can then paste in your LLMs
reacted to regisss's post with ๐Ÿ”ฅ 4 days ago
view post
Post
1558
Nice paper comparing the fp8 inference efficiency of Nvidia H100 and Intel Gaudi2: An Investigation of FP8 Across Accelerators for LLM Inference (2502.01070)

The conclusion is interesting: "Our findings highlight that the Gaudi 2, by leveraging FP8, achieves higher throughput-to-power efficiency during LLM inference"

One aspect of AI hardware accelerators that is often overlooked is how they consume less energy than GPUs. It's nice to see researchers starting carrying out experiments to measure this!

Gaudi3 results soon...
reacted to nicolay-r's post with ๐Ÿ‘ 4 days ago
view post
Post
1455
๐Ÿ“ข If you're around Replicate AI models and wish to use them in streaming mode via JS, then this snippet might be a quick way to experiment with streaming API usage:
https://gist.github.com/nicolay-r/86fc212086c0955d541244253ec0564b

Why it matters? The original docs has:
๐ŸŸข No the relate support for JS rather only Python/HTTP and NodeJS by using the replicate package.
๐ŸŸข Mixture of NodeJS and bash curl snippets:
https://replicate.com/docs/topics/predictions/streaming

Special thanks to the reated template for accessing APIs of other vendors like Claude / OpenAI by Simon Willson in the following post:
https://til.simonwillison.net/llms/streaming-llm-apis

Default model: meta-llama/Meta-Llama-3-70B

PS: I am happy and open for your comments related to this solution
reacted to m-ric's post with ๐Ÿ‘€ 4 days ago
view post
Post
2460
๐—š๐—ฟ๐—ฒ๐—ฎ๐˜ ๐—ณ๐—ฒ๐—ฎ๐˜๐˜‚๐—ฟ๐—ฒ ๐—ฎ๐—น๐—ฒ๐—ฟ๐˜: you can now share agents to the Hub! ๐Ÿฅณ๐Ÿฅณ

And any agent pushed to Hub get a cool Space interface to directly chat with it.

This was a real technical challenge: for instance, serializing tools to export them meant that you needed to get all the source code for a tool, verify that it was standalone (not relying on external variables), and gathering all the packages required to make it run.

Go try it out! ๐Ÿ‘‰ https://github.com/huggingface/smolagents
  • 2 replies
ยท
reacted to davanstrien's post with ๐Ÿ‘ 4 days ago
reacted to AdinaY's post with ๐Ÿ”ฅ 5 days ago
view post
Post
3485
InspireMusic ๐ŸŽต๐Ÿ”ฅ an open music generation framework by Alibaba FunAudio Lab
Model: FunAudioLLM/InspireMusic-1.5B-Long
Demo: FunAudioLLM/InspireMusic
โœจ Music, songs, audio - ALL IN ONE
โœจ High quality audio: 24kHz & 48kHz sampling rates
โœจ Long-Form Generation: enables extended audio creation
โœจ Efficient Fine-Tuning: precision (BF16, FP16, FP32) with user-friendly scripts
  • 1 reply
ยท