Hugging Face's logo

Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up

All HF Hub posts
danielhanchenΒ 
posted an update 2 days ago
Banaxi-TechΒ 
posted an update about 13 hours ago
view post
Post
1367
AGI has arrived.


Just gotta wait for the GLM distill.
  • 5 replies
Β·
GGUFGuyΒ 
posted an update about 16 hours ago
view post
Post
1109
wait why can i post
  • 10 replies
Β·
RN0311Β 
posted an update 2 days ago
view post
Post
3597
SONAR is now open-source! πŸ”Š

SONAR is an evaluation toolkit for multilingual ASR that goes beyond WER/CER. It combines semantic similarity, the Poseidon Score, and analysis across dialect, demographic, and metadata-based failure modes. 🌍

Our goal is to make it easier for everyone to understand why an ASR model fails, not just how often. πŸ” You can plug in your own models + audio, extend it to new languages and datasets, or contribute directly. πŸ› οΈ

MIT licensed. Would love feedback from the HF community! πŸ€—

πŸ”— GitHub: https://github.com/PSDN-AI/SONAR-OSS/
πŸ”— Blog: https://www.psdn.ai/blog/open-source-multilingual-asr-evaluation
anakin87Β 
posted an update 2 days ago
view post
Post
3309
I made a 1.1M ModernBERT encoder play Doom in real time on a CPU

Some time ago, VAGO Solutions released SauerkrautLM-Doom-MultiVec-1.3M, a tiny model trained to play Doom Defend the Center scenario from 31k human gameplay examples.

My first thought: cool! I love both Doom and Small Language Models.

Then another idea: I bet I can do better :-)

What I did?
- evaluated the original model and found it's better than reported
- changed a bit the architecture
- generated SFT data with a scripted oracle
- SFT + PPO refinement on consumer hardware

Got a smaller, faster and killer model
Can even fit a floppy with int8 quantization πŸ’Ύ

Watch it play/read the article: anakin87/tiny-doom-defender
HannesVonEssenΒ 
posted an update about 9 hours ago
view post
Post
587
πŸ“£ HF Viewer now has a HF space! πŸ€—

embedl/hfviewer

Visualize any model directly on Hugging Face - now 4,727 graphs!

If you like it, feel free to give the space a heart to help it grow! ❀️

And you can reply with any feedback or feature requests here!
OppaAIΒ 
posted an update 2 days ago
view post
Post
3092
Another small 4B model comes out yesterday.
NeoHorse 1 4B
TokenRhythm/NeoHorse-1-4B

There are quite a few good smaller parameter models that are capable for Agentic tasks:

The ones from the chart, I have tried a few already in my Jetson Orin Nano,
❌Gemma4 E2B IT - cannot fit my RAM usage if use with TTS and embedder
❓Qwen3.5 4B - just barely fit my RAM usage, need to add think/no_think
❌Spark X2.5 4B - need to build the forked llama.cpp; no vision
➑️Nanbeige 4.2 3B - need to build the forked llama.cpp; slower than Ministral3-3B by 25%; no vision but good for coding; maybe run this is separate server for doing coding tasks
➑️Agents A1 4B - This one is quite interesting. Another Qwen3.5 4B base. I just learnt this right now. This model may surpassed the Ministral3-3B that I'm currently running.
➑️NeoHorse 1 4B - wait for GGUF version comes out; Qwen 3.5 4B base with vision striped
➑️Needle2 45M - need to use separately from llama.cpp server; currently testing to see if it can be used as spawning sub-agents to do parallel tasks
  • 7 replies
Β·
nwaughachukwumaΒ 
posted an update 2 days ago
view post
Post
3170
It’s easy to get distracted by benchmarks, throughput (tok/s), and all the hype around frontier model releases.

This is Shiny Model Syndrome, which makes engineers and teams forget the basic physics of production software, i.e., using the right tool for the job and optimizing for ease of integration.

- Teams spend huge amounts of money on frontier models for document parsing, OCR, detection, segmentation, and other task-specific visual AI workflows.

- Inference marketplaces don’t find it profitable to list task-specific models like glm-ocr, paddleocr, or dots.mocr, even though they’re all superior to frontier VLMs for document parsing and OCR.

- Engineers stitch together multiple endpoints for different use cases across the long tail of visual AI. Those who choose to self-host instead deal with painful infrastructure and GPU ops.

At VLM Run, we wanted one place to run OCR models, VLMs, and ViTs that we could confidently use for our own internal agents and evals. The gateway was born out of that need, and we’ve since opened it to the public.

The gateway exposes a single OpenAI-compatible endpoint for the long tail of visual AI across OCR, document parsing, VQA, detection, segmentation, embeddings, and transcription. Simply point the base_url of your OpenAI SDK at gateway.vlm.run/v1/openai, or ask your agent to connect via MCP (gateway.vlm.run/mcp).

You can swap the model name to compare glm-ocr, dots.mocr, paddleocr-vl-1.6, qwen3.8-27b, gemma4-26b-a4b, and more. We handle serving, runtime, and pipelining behind the scenes to give you high-quality visual intelligence.

- https://vlm.run/gateway
- https://huggingface.co/blog/vlm-run/introducing-gateway
- https://www.vlm.run/blog/introducing-gateway
GoktugDΒ 
posted an update 2 days ago
view post
Post
2349
πŸ‡ΉπŸ‡· One of our small Turkish models quietly reached **500+ monthly downloads** on Hugging Face.

**Werea-TR-TextRestore β€” only 300M parameters.**

Its job is simple:

istanbulda hava cok guzel
β†’ Δ°stanbul'da hava Γ§ok gΓΌzel.

A lightweight model for restoring Turkish text:
β€’ diacritics
β€’ punctuation
β€’ casing
β€’ corrupted text

**96.5% word accuracy** on real Turkish news sentences.

And it runs without sending your text to a cloud API.

πŸ€— Try the model:
Werea-co/Werea-TR-TextRestore

πŸ‡ΉπŸ‡· Built in TΓΌrkiye. Open source.

If you're working on Turkish NLP, I'd love to hear what we should build next.

#TurkishNLP #HuggingFace #OpenSourceAI #NLP
  • 2 replies
Β·
bghiraΒ 
posted an update 3 days ago
view post
Post
2647
I've been gradually recaptioning aged text-to-image datasets with better vision models! These datasets are also repackaged into the more modern webshart format (https://github.com/bghira/webshart) which includes built-in aspect bucketing and caption delivery.

The first two datasets are ready for use!

- webshart/terminusresearch-photo-anatomy
- webshart/terminusresearch-photo-aesthetics

"anatomy" is a bunch of human-centric images containing people holding or otherwise interacting with objects or positioned in complex ways.

"aesthetics" is a collection of visually striking images - high contrast, diverse colouration, and cinematic framing (among other factors).

These two datasets from 2023 were recaptioned with GLM 5.3 Flash via Ollama Cloud and Zhipu AI APIs, with a smaller portion run over 4x H100 with GLM 5.3 Flash in W4A16 precision - these outputs were checked by hand for quality, and the 4bit run was continued at a batch size of 64.

What datasets would you like to see recaptioned next? A better CC12M is on its way!
  • 3 replies
Β·
Recently active users