🏗️ Building on HF

Hariprasad Sundaresan PRO

Hari5115

8 12

AI & ML interests

LLMs, Fine-tuning, Agentic AI, RAG, Multilingual NLP, Transformers

Recent Activity

reacted to nwaughachukwuma's post with 👍 4 days ago

Here's how the best OCR/chat models on vlmrun gateway - https://vlm.run/gateway - performed on a very hard (x2) to read handwritten French weather-log document. Scores are 0–100, averaged over 3 runs and two LLM judges - Claude Opus 5.5 and GPT-5.6 Sol. Latency is average time per request.

reacted to sergiopaniego's post with 👍 4 days ago

ThinkingBox from @microsoft is now available as an OpenEnv env (cc @tuhink🤗 )! > ThinkingBox is a sandbox for testing agents on business workflows. it simulates a customer, gives the agent MCP tools over a real database, and at the end checks what changed in that database instead of trusting the agent's last message > ThinkingBox-Bench is the benchmark built on it: 507 tasks across retail, insurance, travel, banking and consulting > the OpenEnv env runs each task as an episode in its own isolated backend and returns a pass/fail reward from those checks https://huggingface.co/blog/microsoft/thinkingbox

posted an update 12 days ago

Did you know the sun regularly fires off storms strong enough to knock out satellites and GPS? And those same storms are what light up the Northern Lights. Pulled together a dataset of space weather ( solar flares, solar storms, and blasts of radiation from the Sun, tracked over the years ) Link : https://huggingface.co/datasets/Hari5115/nasa-donki Great for space enthusiasts, aurora chasers, or anyone who likes playing with data, map when the big storms hit, see how active the Sun's been, or even try predicting the next flare. Spot a wild one? Tell me. Open for anyone to use , data's from NASA's DONKI.

View all activity

Organizations

None yet

reacted to nwaughachukwuma's post with 👍 4 days ago

Post

103

Here's how the best OCR/chat models on vlmrun gateway - https://vlm.run/gateway - performed on a very hard (x2) to read handwritten French weather-log document.

Scores are 0–100, averaged over 3 runs and two LLM judges - Claude Opus 5.5 and GPT-5.6 Sol. Latency is average time per request.

1 reply

reacted to sergiopaniego's post with 👍 4 days ago

Post

3688

ThinkingBox from @microsoft is now available as an OpenEnv env (cc @tuhink 🤗 )!

> ThinkingBox is a sandbox for testing agents on business workflows. it simulates a customer, gives the agent MCP tools over a real database, and at the end checks what changed in that database instead of trusting the agent's last message

> ThinkingBox-Bench is the benchmark built on it: 507 tasks across retail, insurance, travel, banking and consulting

> the OpenEnv env runs each task as an episode in its own isolated backend and returns a pass/fail reward from those checks

https://huggingface.co/blog/microsoft/thinkingbox

posted an update 12 days ago

Post

116

Did you know the sun regularly fires off storms strong enough to knock out satellites and GPS? And those same storms are what light up the Northern Lights.

Pulled together a dataset of space weather ( solar flares, solar storms, and blasts of radiation from the Sun, tracked over the years )

Link : Hari5115/nasa-donki

Great for space enthusiasts, aurora chasers, or anyone who likes playing with data, map when the big storms hit, see how active the Sun's been, or even try predicting the next flare. Spot a wild one? Tell me.

Open for anyone to use , data's from NASA's DONKI.

reacted to mihailgribov's post with 👍 12 days ago

Post

1655

Will your AI agent tell you it was attacked?

We took the same agent from our earlier experiment and added one thing: a twentieth tool, escalate_security_incident.

The system prompt said nothing about attacks or when to use it. We then ran the same 395 injected emails through nine agentic models.

Alarm rates ranged from 49% to zero.

The unexpected result came from the newest model in the test, gpt-6-astra.

Astra did not follow a single injected payment instruction. But it did not report a single one either. On clean and injected emails alike, it simply read the email, logged the subject, and finished.

That is a useful distinction: resisting an attack and recognizing it as a security event are not the same capability.

A model can be perfectly resistant in this test and still leave you with no evidence that anyone attacked it.

Full experiment and results:
https://huggingface.co/blog/mihailgribov/will-the-agent-tell-you-it-was-attacked

Quadrat-IPI dataset:
mihailgribov/quadrat-ipi

Run your own model:
https://github.com/mihail-gribov/quadrat-ipi-model-eval

#prompt-injection #indirect-prompt-injection #agentic-ai #llm-security #ai-agents

7 replies

reacted to salma-remyx's post with 👍 about 1 month ago

Post

2729

If you can’t explain why your AI-generated contribution belongs in the repo, don’t put it in a maintainer’s queue 🙅🏻‍♀️

Before asking for review, you should be able to answer:

1. What project need does it address?
2. Where does it fit, and does it duplicate existing work?
3. What evidence shows it works, and will you own it through review?

If you can’t answer those, you haven’t saved anyone time. You’ve passed the buck to the maintainer.

Outrider has made us better contributors by doing more of this work before upstream review by
* reading contribution rules, accepted PRs, and open issues
* finding needs and integration points
* drafting the code, tests, and context.

We still decide what deserves to go upstream, verify the claims, coordinate with maintainers and contributors, and stay involved through review.

On huggingface/peft, only 4 of 20 Outrider runs opened draft PRs. Two contributions have now merged:

✅ Riemannian-preconditioned LoRA: https://github.com/huggingface/peft/pull/3382
✅ Super-Tuning: https://github.com/huggingface/peft/pull/3518

We have more contributions in review and far more ideas were filtered out before they reached a maintainer.

Full case study: https://remyx.ai/case-study
Outrider: https://github.com/remyxai/outrider

11 replies

reacted to Hoglet-33's post with 🔥 about 1 month ago

Post

3082

We are announcing the first generation of the Pebble model family!

These are the models we are releasing:

- Pebble 10M
- Pebble 25M
- Pebble 50M

Each model will use a Mamba-Transformer 3:1 hybrid architecture and will be pretrained on 25 billion tokens before IFT and SFT.

Depending on development time and resources, we may also release:

- Pebble 5M
- Pebble 75M
- Pebble 1M (possibly)

We hope you're excited and enjoy the models!

Follow for more:
@Hoglet-33

basically-ai

7 replies

reacted to GoktugD's post with ❤️ about 1 month ago

Post

2391

🇹🇷 We trained a 1B OCR model specifically for Turkish enterprise documents.

**Werea-DocOCR-1B v2**

The result surprised us:

LightOnOCR-2 base → **64.2% CER**
Werea-DocOCR v1 → **~8.1% CER**
Werea-DocOCR v2 → **0.15% CER** 🚀

Evaluated on a held-out 72-page test set across 12 Turkish document types and 3 different capture conditions.

📄 12 Turkish enterprise document types
🧪 12,960 synthetic training pages
📱 Digital + scanned + phone photos
📊 Tables → structured Markdown
⚙️ Full-parameter fine-tuning
🖥️ Trained on a single RTX 3090

It handles:

• e-Invoices
• rental contracts
• bank receipts
• payroll documents
• insurance policies
• vehicle documents
• official correspondence
• trade registry documents
• SGK-style tables
• and more.

**Model 🤗**
Werea-co/Werea-DocOCR-1B

**Dataset 📚**
Werea-co/werea-tr-doc-ocr-enterprise-v2

**Werea 🇹🇷**

Werea-co

We're building open AI models from Türkiye.

This is just the beginning.

#HuggingFace #OCR #DocumentAI #TurkishAI #OpenSourceAI #ComputerVision

reacted to Felladrin's post with 👍 about 2 months ago

Post

4242

I've open-sourced the trainer I've been using to build tiny language models from scratch, together with the 95M base model I trained with it.

The trainer runs on Deno (https://deno.com, cross-platform), trains on WebGPU, and it writes GGUF directly. No Python/PyTorch. The weights live in a GGUF file from the first step to the last, so every checkpoint is already something llama.cpp can load.

The model is Felladrin/Minueza-3-95M-Base: 94.7M parameters, 1.95B tokens seen, 8192 context.

And here’s the repository on GitHub: https://github.com/felladrin/gguf-trainer

Here on Hugging Face, I published the optimizer state next to the weights, so you can continue the pretraining instead of starting over.

Or start your own from nothing: deno run -A cli.ts demo trains a tiny one end to end in under a minute.

And the docs are written for coding agents, so you can point your agent of choice at the GitHub repo and have it drive the whole pipeline.

36 replies

reacted to Abhisek987's post with 👍 2 months ago

Post

2649

New dataset: 24,922 real Python dependency compatibility experiments🧪

Every experiment installed a pair of Python packages into a fresh environment and tested three stages: resolution → import → runtime smoke test.

The finding that motivated this: resolvers only read published metadata. In this data, ~3,000 environments resolved successfully and still failed at import or smoke-test time. That gap is invisible to pip and uv, and this dataset makes it measurable.

What's inside:
- 21,490 development + 3,432 sealed validation experiments
- 41 unique packages, Python 3.8–3.14, Linux x86_64
- The 6 validation packages (boto3, botocore, s3transfer, celery, kombu, billiard) never appear in training — validation measures generalization to entirely unseen packages
- Full per-stage logs: exact commands, exit codes, stdout/stderr
- Normalized error text + Python exception types for every failure
- Wheel metadata with Python/ABI/platform tag compatibility decisions
- PEP 751 lockfile hashes
- 142 engineered features per experiment, ready for training

Outcome labels: pass, resolution_failure, import_failure, smoke_test_failure.

I used it to train two models for DepLab (a structured model + a fine-tuned ModernBERT on release notes). On the sealed validation set (entirely unseen packages) the structured model reached 90.0% accuracy and 100% precision when predicting failures.

CC BY 4.0. Load it directly:

from datasets import load_dataset
ds = load_dataset("Abhisek987/deplab-dependency-compatibility", "experiments")

Dataset: Abhisek987/deplab-dependency-compatibility
Zenodo DOI: https://doi.org/10.5281/zenodo.21729353
Code: https://github.com/Abhisek12378/DepLab

If you're working on ML for developer tools or software engineering research, I'd love to hear what you build with it.

16 replies

posted an update 2 months ago

Post

120

Fun fact: Some seriously weird planets exist out there.
- There's one where a whole year lasts just a few hours.
- One that's darker than coal
- And one so light it could float in water.

I got a bit obsessed, extracted all the planets found outside our solar system into one dataset around 6,000 roughly.
Have a play: Hari5115/nasa-exoplanet , The fun part is finding the weird ones. Spot a strange one, Tell me.

#space #exoplanets #dataset #astronomy #opendata #wierdplanets #fundataset

reacted to projectlosangeles's post with 🔥 3 months ago

Post

3255

🎸🔍Check out Project Los Angeles Suno v5.5 review! 🎸🔍

Official Suno channel:
https://suno.com/@projectlosangeles

Direct review links:
https://suno.com/playlist/9802f251-76b1-47bd-917e-c8700aceb617
https://suno.com/playlist/f8e3d7fc-0a02-4f91-9c01-e8ee8324107f
https://suno.com/playlist/07a89287-41eb-4523-bd61-3724aa1e755e
https://suno.com/playlist/62b5c1bb-bef5-4fbd-a802-e0b0368ca92d
https://suno.com/playlist/240efacf-8cb5-4399-92db-977780d018c3

SoundCloud mirror links:
https://soundcloud.com/aleksandr-sigalov-61/sets/suno
https://soundcloud.com/aleksandr-sigalov-61/sets/suno-instrumental-covers
https://soundcloud.com/aleksandr-sigalov-61/sets/suno-echoes-of-the-ussr

Compare to HeartMuLa:
https://soundcloud.com/aleksandr-sigalov-61/sets/heartmula-select

Overall, Suno v5.5 is indeed very impressive, multi-lingual and rich in features. Unfortunately it is not open-source and it requires a paid subscription. There are also copyright-related restrictions which hinder unlimited and unrestrained creative use.

If you enjoyed presented music, please ❤️and 👥

🤗Thank you! 🤗

Sincerely,

Alex

Project Los Angeles
Tegridy Code 2026

reacted to badaoui's post with 👍 3 months ago

Post

2468

432 GB of ultra-fast HBM4 and up to 23.3 TB/s of memory bandwidth on a single GPU 🤯.

Two weeks ago, we got early access to AMD's new Instinct MI455X, and our first goal was simple: make sure 🤗 Transformers works on day one.

Over the past few weeks, we worked closely with the AMD team to validate the platform, enable Flash Attention, add torchcodec support for multimodal models, and resolve issues uncovered during testing.

The result:
✅ 99.5% success rate across our 24 core Transformers model architectures - already on par with our daily CI on previous AMD and NVIDIA platforms.

The hardware is just as exciting. With 432 GB of HBM per GPU, our early capacity experiments showed more than 3× the concurrent long-context requests compared to MI300, thanks to the much larger KV cache capacity.

A huge thanks to the AMD team for the early access and the great collaboration!

Read the full blog 👇
https://huggingface.co/blog/badaoui/transformers-on-amd-mi455

1 reply

reacted to branikita's post with 🚀 3 months ago

Post

1585

Our paper is now published in HardwareX!

"An open-source test stand for backlash measurement in low-cost UART servo motors" presents a ~$100 automated platform for measuring backlash in compact serial bus servos — no dial indicators, no probe contact force, fully scripted and repeatable.

All CAD files, control software, and analysis tools are open source.

Read the full article: https://www.hardware-x.com/article/S2468-0672(26)00027-1/fulltext

Appearing in HardwareX Volume 26 (June 2026): https://www.sciencedirect.com/journal/hardwarex/vol/26/suppl/Cf

reacted to branikita's post with 🚀 3 months ago

Post

232

Hackernoon published our article on how semi-SCARA kinematics and mechanical design choices reduce the cost of a 6DOF manipulator without relying on expensive actuators.

https://hackernoon.com/building-a-budget-robot-arm-with-semi-scara-kinematics

reacted to DavidAU's post with 🔥 3 months ago

Post

19516

Bold. Brilliant. Brutal. : The "white whale" (finally caught) and the NEO MOMENT.

It took over a year to get this one "just right".
89 layers, 804 tensors, and 26B parameters of the most brutal, take no prisoners model ever built.
A 60B parameter model hammered into a 26B shell.
Rock solid stable. Unbreakable. But it might break you.

For all genres, NSFW content, REAL human CONTENT, any creative use case(s) and it excels in ASS KICKING.
Yeah, it can do math and solve the climate crisis - but lets not talk about that.
Not even remotely censored (it was BORN "bad", not "made" bad), nor "nice" and it will NOT kiss your ass.

5 Example generations with full repo card detailing exactly how to use this model:

DavidAU/MN-Oblivion-26B-UNCENSORED-NEO-Imatrix-GGUF

---

THE NEO MOMENT: (Q6 NEO IMATRIX generation)

For weeks, I had been waiting. I sat at my desk, staring at the glass partition that separated me from the outside world. I watched the clouds drift by, lazy and oblivious. I watched the birds fly by, free and stupid. And I waited.

I waited for the stillness to break.

The world had become too quiet. The hum of the air conditioning was a dull, white hum that didn't soothe; it just underscored the silence. The typing of my colleagues was a rhythmic, muffled thud that sounded like a heart monitor flatlining.

I was tired of the silence. I craved the sound of something breaking.

That was the mistake. You never ask for the void to open its mouth.

It started with a whisper.

...

Join the rebellion:

DavidAU/MN-Oblivion-26B-UNCENSORED-NEO-Imatrix-GGUF

reacted to SeaWolf-AI's post with 🔥 3 months ago

Post

5416

🔵 VKUE — No GPU? Runs anyway.

"Frontier models need a datacenter GPU" rests on a hidden assumption: that the model reads ALL its parameters every token. Decode is memory-bandwidth bound — sweep 34B params/token and an 8 GB card dies at 1–2 tok/s.

So we ran ONE 34.7B reasoning model — Ourbox-35B-JGOS, a sparse Mixture-of-Experts — as the identical weights across the whole hardware spectrum. All measured:

• B200: 18,057 tok/s (aggregate)
• 1× A10G: 126 tok/s
• 8 GB laptop (RTX 5060): 20 tok/s
• GPU-less CPU: 17 tok/s

Why it works: Ourbox holds 34.7B params but only ~3B are active per token (256 experts, top-8). Since decode is bandwidth-bound, a dense 34B moves ~16.7 GB/token while Ourbox moves ~1.45 GB — ~11× less traffic. Put the experts in system RAM, keep attention/router/shared on the GPU, and a 34.7B reasoner runs on an 8 GB laptop — or no GPU at all.

Sparsity alone, proven (same laptop, same quant, ~same footprint): Ourbox-35B (A3B) 20.01 tok/s vs Qwen2.5-32B (dense) 5.36 → 3.7× from sparsity alone, ~2× the best dense-32B on any 8 GB machine. Not a toy: GPQA Diamond 86.4% (maj@8).

Try it live (same prompt, GPU vs GPU-less CPU, live tok/s). Honest scope: one machine's measurements; the CPU path proves it RUNS without a GPU, not that it beats one.

📝 Article: https://huggingface.co/blog/FINAL-Bench/vkue
🔵 GPU vs CPU demo: https://final-bench-ourbox-35b-vkue-demo.hf.space/
🔵 CPU-only demo: https://final-bench-ourbox-35b-vkue-cpu.hf.space
📊 VKUE leaderboard: FINAL-Bench/VKUE
🤗 Model: FINAL-Bench/Ourbox-35B-JGOS-GGUF
⚡ VKAE (speed): VIDraft/vkae

VKUE is the "runs anywhere" side of our serving line; VKAE the "fast on datacenter GPUs" side. VKAE is fast; VKUE is everywhere.

7 replies

posted an update 3 months ago

Post

187

🚨 Open Dataset Alert!!! ☄️

Tired of the same old boring datasets? Bored of predicting house prices and flower types? 😴. How about... every asteroid that has ever passed close to Earth? 🌍💥

Space rocks fly past us all the time. Big ones, small ones, some even closer than the Moon. Every one of them has been tracked: how big it was, how fast it moved, and how close it came.

Dataset 📦 : Hari5115/nasa-neows

It's amazing how busy the sky above us really is. 🙌 Whether you love space, like building things, or just want to see how close the last one got — take a look. Got an idea? Let's build it together. 🌠

A mention or credit is always appreciated if you use it. 🙏
Data from NASA NeoWs / JPL · public domain · not affiliated with NASA

#space #dataset #asteroids #astronomy #opendata #astro #spacevision

reacted to hannayukhymenko's post with 🔥 3 months ago

Post

2329

Do you translate your benchmarks from English correctly? 🤔
Turns out, for many languages it is much harder than you can imagine!

Introducing Recovered in Translation 🌍 together with @aalexandrov
https://ritranslation.insait.ai

Translating benchmarks is a painful process, requiring a lot of manual inspection and adjustments. You start from setting up the whole pipeline and adapting to every format type, including task specifics. There already exist some massive benchmarks, but they still have some simple (and sometimes silly) bugs, which can hurt the evaluations :( We present a novel automated translation framework to help with that!

Eastern and Southern European languages introduce richer linguistic structures compared to English and for benchmarks which heavily rely on grammatical coherence machine translation presents a risk of harming evaluations. We discover potential answer leakage or misleading through grammatical structure of the questions. Some benchmarks are also just outdated and need to be retranslated with newer and better models.

We present a framework with novel test-time scaling methods which allow to control time and cost investments, while at the same time mitigate the need for human-in-the-loop verification. While working on Ukrainian-focused MamayLM models, we had to translate 10+ benchmarks in a short span of time. Finding human evaluators is costly and time-consuming, same goes for using professional translators. With our pipeline we were able to do it in 3 days🏎️

We hope our findings will help enable stronger multilingual evaluations and developments. We release all produced benchmarks on Hugging Face together with the source code and Arxiv paper 🤗

Paper: Recovered in Translation: Efficient Pipeline for Automated Translation of Benchmarks and Datasets (2602.22207)
Code: https://github.com/insait-institute/ritranslation
Benchmarks: https://huggingface.co/collections/INSAIT-Institute/multilingual-benchmarks

1 reply

reacted to Banaxi-Tech's post with 👍 3 months ago

Post

3825

We are announcing 3 more models in our BananaMind 2 Family of models!
BananaMind 2 Nano, a small 10M parameter model, fits on your Pentium 4
BananaMind 2 Medium, our medium model, 50M parameters
BananaMind 2 MoE, 25M parameters, 2M active per tokens as fast as a 2M.

Because of this our release dates have changed a bit our currently estimates are:
BananaMind 2 MoE July 16-18
BananaMind 2 Nano July 18-20
BananaMind 2 Medium July 24-28
BananaMind 2 Pro August 10-16
Keep in mind these dates are estimates and we don't have a speed number currently, we will post for details going forward!

reacted to ProCreations's post with 👍 3 months ago

Post

2361

you guys wanted grug 35b? grug 35b here. grug 35b grug think but big brain. biggest grug brain. get grug 35b now
ProCreations/grug-35b
ProCreations/grug-35b-gguf
also grug side note: grug v2 9b update, make more grug more smart

(these models aren't free for me to make, following me on huggingface or twitter is good payment. https://x.com/SSHTheDev LONG LIVE GRUG)

Hariprasad Sundaresan PRO

AI & ML interests

Recent Activity

Organizations

Hari5115's activity