Quazim0t0 (Quazimoto)

reacted to merve's post with ❤️❤️ 3 months ago

Post

2468

past week had huuuge releases 💗
here's our picks 🔥 find more models, datasets, demos here merve/releases-july-11-68750452c358c98b0fa663f7

> moonshotai/Kimi-K2-Instruct is the new sota LLM with 1T total 32B active parameters 🤯

> HuggingFaceTB/SmolLM3-3B is the new best LM for it's size, offers thinking mode 💭 as well as the dataset HuggingFaceTB/smoltalk2

> Alibaba-NLP/WebSailor-3B is the new agentic LLM for complex browsing

> Google DeepMind released medical vision LMs with an agentic doctor-patient app google/medgemma-release-680aade845f90bec6a3f60c4

> fal released a LoRA to improve details on face images fal/Realism-Detailer-Kontext-Dev-LoRA

posted an update 3 months ago

Post

340

Used YoloV8n/YoloV11n Onnx + FastVLM to provide real time object detection and annotation. Works well with videos that do not have a lot of changes. In the space I used a stock security camera footage. It annotates using FastVLM while doing Object Detection using YoloV8n. The first version of this used only the FastVLM to annotate to draw the boxes then I realized that I could just be using Yolov8n. Really excited to see the future applications of ONNX models!

See the difference!
Quazim0t0/FastVLM-YoloV8n-v2
Quazim0t0/FastVLM-YoloV11n

posted an update 3 months ago

Post

1949

I loved the idea of the Boxing by sergiopaniego/vlm_object_understanding
And https://huggingface.co/spaces/webml-community/fastvlm-webgpu

So I tried to combine the two idea, unfortunately I can’t seem to get it consistent and I only worked on the File Upload side. You may have to change the prompt a bit to suite the video you upload but it seems to semi work. If anyone knows a better way to fix this, I really wanted to use this for a project but I can’t seem to figure it out.

Quazim0t0/FastVLMBoxes
I used videos from here and uploaded them to try it out.

https://pixabay.com/videos/search/branch+birds/

reacted to CultriX's post with 🔥 3 months ago

Post

1865

New Space: Generate Knowledge Graphs from input data using LLM's (OpenRouter). It's a trial project but seems to be working alright so far!

CultriX/Generate-Knowledge-Graphs

Below is an example after feeding it the wikipedia page about Elon Musk:

reacted to lbourdois's post with ❤️ 7 months ago

Post

3371

We introduce FAT5 (Flash Attention T5) ⚡

An implementation of T5 in PyTorch with UL2 objective optimized for GPGPU for both training and inference thanks to 13 different optimizations.
The main one is that we have designed a CUDA kernel to expand the Flash Attention by @tridao with RPE biases and supports other PE such as RoPE, ALiBi or FIRE.
The result kernel is 2 times faster than a SPDA implementation.
We also use Triton kernels to optimize certain parts of the architecture, such as the cross-entropy and RMSNorm layer.

The various kernels have been carefully built to be compatible with BF16 and torch.compile to go even faster and achieve efficient pretraining.

All other optimizations are described in a 📝 subsequent blog post available on @huggingface 🤗: CATIE-AQ/FAT5-report.

This methodology enabled us to efficiently pretrain as a proof of concept a FAT5 with 147M parameters in French in a reasonable time (1,461H for 419B tokens), with limited resources (1 A100 i.e. a computational budget of ~ €1,900) and a low carbon footprint (13.5kg eq CO2).

The model's weights are also available on Hugging Face: CATIE-AQ/FAT5-small.
Not very useful in practice, it's a PoC and not an instructed model (it's planned for later).

All the code is available on GitHub if you want to pretrain your own model in your own language or for a specific domain: https://github.com/catie-aq/flashT5 ⭐

Ending by indicating that was a joint project with @BorisAlbar at hf.co/CATIE-AQ.

reacted to AtAndDev's post with 👍 7 months ago

Post

4362

There seems to multiple paid apps shared here that are based on models on hf, but some ppl sell their wrappers as "products" and promote them here. For a long time, hf was the best and only platform to do oss model stuff but with the recent AI website builders anyone can create a product (really crappy ones btw) and try to sell it with no contribution to oss stuff. Please dont do this, or try finetuning the models you use...
Sorry for filling yall feed with this bs but yk...

6 replies

·

reacted to burtenshaw's post with 🤗 7 months ago

Post

2093

everybody and their dog is fine-tuning Gemma 3 today, so I thought I'd do a longer post on the tips and sharp edges I find. let's go!

1. has to be install everything form main and nightly. this is what I'm working with to get unsloth and TRL running

git+https://github.com/huggingface/transformers@main
git+https://github.com/huggingface/trl.git@main
bitsandbytes
peft

plus this with --no-deps

git+https://github.com/unslothai/unsloth-zoo.git@nightly
git+https://github.com/unslothai/unsloth.git@nightly

2. will brown's code to turn GSM8k into a reasoning dataset is a nice toy experiment https://gist.github.com/willccbb/4676755236bb08cab5f4e54a0475d6fb

3. with a learning rate of 5e-6 rewards and loss stayed flat for the first 100 or so steps.

4. so far none of my runs have undermined the outputs after 1 epoch. therefore, I'm mainly experimenting with bigger LoRA adapters.

from trl import GRPOConfig

training_args = GRPOConfig(
    learning_rate = 5e-6,
    adam_beta1 = 0.9,
    adam_beta2 = 0.99,
    weight_decay = 0.1,
    warmup_ratio = 0.1,
    lr_scheduler_type = "cosine",
    optim = "adamw_8bit",
    logging_steps = 1,
    per_device_train_batch_size = 2,
    gradient_accumulation_steps = 1,
    num_generations = 2,
    max_prompt_length = 256,
    max_completion_length = 1024 - 256,
    num_train_epochs = 1,
    max_steps = 250,
    save_steps = 250,
    max_grad_norm = 0.1,
    report_to = "none",
)

5. vision fine-tuning isn't available in TRL's GRPOTrainer, so stick to text datasets. but no need to load the model differently in transformers or Unsloth

from transformers import AutoModelForImageTextToText

model = AutoModelForImageTextToText.from_pretrained("google/gemma-3-4b-it)

if you want an introduction to GRPO, check out the reasoning course, it walks you through the algorithm, theory, and implementation in a smooth way.

reasoning-course

2 replies

·

posted an update 7 months ago

Post

1206

Thank you to the Open LLM Leaderboard's team for offering it to the community for as long as they did. I only recently joined HF, and it provided a lot of incentive and information to make better models.

Always will remember getting to #112 :D

Anyone have a solid way to test my models privately? Please let me know!

1 reply

·

reacted to onekq's post with 👍 7 months ago

Post

757

A bigger and harder pain point for reasoning model is to switch modes.

We now have powerful models capable of either system I thinking or system II thinking, but not both, much less switching between the two. But humans can do this quite easily.

ChatGPT and others push the burden to users to switch between models. I guess this is the best we have now.

2 replies

·

reacted to AdinaY's post with 🔥 7 months ago

Post

1469

R1-Omni🔥RLVR-Powered Multimodal LLM released by Alibaba

Model: StarJiaxing/R1-Omni-0.5B
Paper: R1-Omni: Explainable Omni-Multimodal Emotion Recognition with Reinforcing Learning (2503.05379)

✨0.5B with Apache2.0
✨ Improve emotion recognition with visual and audio cues

1 reply

·

reacted to thomwolf's post with 🚀 7 months ago

Post

3101

We've kept pushing our Open-R1 project, an open initiative to replicate and extend the techniques behind DeepSeek-R1.

And even we were mind-blown by the results we got with this latest model we're releasing: ⚡️OlympicCoder ( open-r1/OlympicCoder-7B and open-r1/OlympicCoder-32B)

It's beating Claude 3.7 on (competitive) programming –a domain Anthropic has been historically really strong at– and it's getting close to o1-mini/R1 on olympiad level coding with just 7B parameters!

And the best part is that we're open-sourcing all about its training dataset, the new IOI benchmark, and more in our Open-R1 progress report #3: https://huggingface.co/blog/open-r1/update-3

Datasets are are releasing:
- open-r1/codeforces
- open-r1/codeforces-cots
- open-r1/ioi
- open-r1/ioi-test-cases
- open-r1/ioi-sample-solutions
- open-r1/ioi-cots
- open-r1/ioi-2024-model-solutions

reacted to Lunzima's post with 🚀 7 months ago

Post

1373

I'm currently experimenting with the SFT dataset Lunzima/alpaca_like_dataset to further boost the performance of NQLSG-Qwen2.5-14B-MegaFusion-v9.x. This includes data sourced from DeepSeek-R1 or other cleaned results (excluding CoTs). Additionally, datasets that could potentially enhance the model's performance in math and programming/code, as well as those dedicated to specific uses like Swahili, are part of the mix.
@sometimesanotion @sthenno @wanlige

1 reply

·

reacted to awacke1's post with 🚀 7 months ago

Post

2350

I introduce MIT license

ML Model Specialize Fine Tuner app "SFT Tiny Titans" 🚀

Demo video with source.

Download, train, SFT, and test your models, easy as 1-2-3!
URL: https://huggingface.co/spaces/awacke1/TorchTransformers-NLP-CV-SFT

2 replies

·

reacted to BrigitteTousi's post with 🚀 7 months ago

Post

3450

LeRobot goes to driving school! 🚗🚗🚗

Hugging Face just announced a new collab with Yaak to bring the largest open-source self-driving dataset to LeRobot!

Major kudos to HF's @cadene , as well as @sandhawalia , @Shnissen and the Yaak team!

Check out the blog post here: https://huggingface.co/blog/lerobot-goes-to-driving-school

1 reply

·

reacted to sandhawalia's post with 🔥 7 months ago

Post

1980

LeRobot goes to driving school. World's largest open-source self driving dataset. Ready for end-to-end learning with LeRobot.

3 years, 30 German cities, 60 driving instructors and students. https://huggingface.co/blog/lerobot-goes-to-driving-school

Coming this summer — LeRobot driver.

reacted to fdaudens's post with 🤗 7 months ago

Post

5753

Honored to be named among their 12 pioneers and power players in the news industry in the 2025 Tech Trends Report from Future Today Strategy Group.

Incredible group to be part of - each person is doing groundbreaking work at the intersection of AI and journalism. Worth following them all: they're consistently sharing practical insights on building the future of news.

Take the time to read this report, it's packed with insights as always. The news & information section's #1 insight hits hard: "The most substantive economic impact of AI to date has been licensing payouts for a handful of big publishers. The competition will start shifting in the year ahead to separate AI 'haves' that have positioned themselves to grow from the 'have-nots.'"

This AI-driven divide is something I've been really concerned about. Now is the time to build more than ever!

👉 Full report here: https://ftsg.com/wp-content/uploads/2025/03/FTSG_2025_TR_FINAL_LINKED.pdf

2 replies

·

replied to their post 7 months ago

@Delta-Vector
I uploaded the 16Bit on this update. haven't tested it.

posted an update 7 months ago

Post

652

Update to the Imagine side-project.
Just uploaded the 16Bit & Q4

Samples: (Used a base Microsoft Phi4 model)
*You may experience bugs with either the model or the Open WebUI function*
Open WebUI function: https://openwebui.com/f/quaz93/imagine_phi
https://huggingface.co/Quazim0t0/Imagine-v0.5-16bit - Haven't tested
https://huggingface.co/Quazim0t0/ImagineTest-v0.5-GGUF - Tested (Pictures)

Dataset: https://huggingface.co/datasets/Quazim0t0/Amanita-Imagine
Small Dataset of 500+ entries, still working on it here and there when I can.
Pictures use the Open Web UI function I provided.

1 reply

·

replied to their post 7 months ago

I'll try to get that out for you when I get a chance.

Quazimoto PRO

AI & ML interests

Organizations

Quazimoto PRO

AI & ML interests

Organizations

Quazim0t0's activity