Post
1241
boomers still pick zenodo.org instead of huggingface ??? absolutely clownish nonsense , my random datasets have 30x more downloads and views than front page zenodos ... gonna write a comparison blog , but yeah... cringe.
- 1 reply
Arthur Zucker PRO
ArthurZ
AI & ML interests
None yet
Recent Activity
liked a Space about 8 hours ago
sbrandeis/tokenizers-wasm-demo upvoted an article 2 days ago
relore - repository memory for coding agents upvoted an article 2 days ago
tokenizers v1: encode, decode and scaling, measuredOrganizations
![Unofficial Mistral Community [deprecated]'s profile picture](https://cdn-avatars.huggingface.co/v1/production/uploads/6141a88b3a0ec78603c9e784/HAdAzcs1CDB9OxgxT2W9K.png)
reacted to Tonic's post with π 3 days ago
reacted to Molbap's post with π₯ 12 months ago
Post
3662
π New blog: Maintain the unmaintainable β 1M+ Python LOC, 400+ models
How do you stop a million-line library built by thousands of contributors from collapsing under its own weight?
At π€ Transformers, we do it with explicit software-engineering tenets, principles that make the codebase hackable at scale.
π Inside the post:
β One Model, One File: readability first β you can still open a modeling file and see the full logic, top to bottom.
β Modular Transformers: visible inheritance that cuts maintenance cost by ~15Γ while keeping models readable.
β Config-Driven Performance: FlashAttention, tensor parallelism, and attention scheduling are config-level features, not rewrites.
Written with @lysandre ,@pcuenq and @yonigozlan , this is a deep dive into how Transformers stays fast, open, and maintainable.
Read it here β transformers-community/Transformers-tenets
How do you stop a million-line library built by thousands of contributors from collapsing under its own weight?
At π€ Transformers, we do it with explicit software-engineering tenets, principles that make the codebase hackable at scale.
π Inside the post:
β One Model, One File: readability first β you can still open a modeling file and see the full logic, top to bottom.
β Modular Transformers: visible inheritance that cuts maintenance cost by ~15Γ while keeping models readable.
β Config-Driven Performance: FlashAttention, tensor parallelism, and attention scheduling are config-level features, not rewrites.
Written with @lysandre ,@pcuenq and @yonigozlan , this is a deep dive into how Transformers stays fast, open, and maintainable.
Read it here β transformers-community/Transformers-tenets
reacted to lysandre's post with π about 1 year ago
Post
9905
We're kick-starting the process of Transformers v5, with @ArthurZ and @cyrilvallez !
v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.
Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!
v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.
Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!
- 6 replies
reacted to danieldk's post with π€π₯ about 1 year ago
Post
2053
We have been working on a project called
We plan to give kernels a more proper introduction soon. But for those who have been following along, we are happy to announce a new release:
- New layer API with
- Experimental support for loading Apple Silicon Metal π€ Kernels.
- Generate wheels from Hub kernels for legacy deployments.
Full release notes here: https://github.com/huggingface/kernels/releases/tag/v0.6.0
kernels. kernels makes it possible to load compute kernels directly from the Hub! πWe plan to give kernels a more proper introduction soon. But for those who have been following along, we are happy to announce a new release:
- New layer API with
torch.compile support.- Experimental support for loading Apple Silicon Metal π€ Kernels.
- Generate wheels from Hub kernels for legacy deployments.
Full release notes here: https://github.com/huggingface/kernels/releases/tag/v0.6.0
- 2 replies
reacted to mitkox's post with π over 1 year ago
Post
4247
llama.cpp is 26.8% faster than ollama.
I have upgraded both, and using the same settings, I am running the same DeepSeek R1 Distill 1.5B on the same hardware. It's an Apples to Apples comparison.
Total duration:
llama.cpp 6.85 sec <- 26.8% faster
ollama 8.69 sec
Breakdown by phase:
Model loading
llama.cpp 241 ms <- 2x faster
ollama 553 ms
Prompt processing
llama.cpp 416.04 tokens/s with an eval time 45.67 ms <- 10x faster
ollama 42.17 tokens/s with an eval time of 498 ms
Token generation
llama.cpp 137.79 tokens/s with an eval time 6.62 sec <- 13% faster
ollama 122.07 tokens/s with an eval time 7.64 sec
llama.cpp is LLM inference in C/C++; ollama adds abstraction layers and marketing.
Make sure you own your AI. AI in the cloud is not aligned with you; it's aligned with the company that owns it.
I have upgraded both, and using the same settings, I am running the same DeepSeek R1 Distill 1.5B on the same hardware. It's an Apples to Apples comparison.
Total duration:
llama.cpp 6.85 sec <- 26.8% faster
ollama 8.69 sec
Breakdown by phase:
Model loading
llama.cpp 241 ms <- 2x faster
ollama 553 ms
Prompt processing
llama.cpp 416.04 tokens/s with an eval time 45.67 ms <- 10x faster
ollama 42.17 tokens/s with an eval time of 498 ms
Token generation
llama.cpp 137.79 tokens/s with an eval time 6.62 sec <- 13% faster
ollama 122.07 tokens/s with an eval time 7.64 sec
llama.cpp is LLM inference in C/C++; ollama adds abstraction layers and marketing.
Make sure you own your AI. AI in the cloud is not aligned with you; it's aligned with the company that owns it.
- 7 replies
reacted to MonsterMMORPG's post with πβ€οΈ almost 2 years ago
Post
2029
FLUX Redux is a hidden Gem
I am still doing huge research to publish an amazing fully Public - no paywalled Tutorial, but this is generated via SwarmUI
Style Model Merge Strength : 0.5
FLUX Guidance Scale is : 6
Used base model is my FLUX fine tuned model with 256 images via Kohya SS GUI as shown in tutorial ( https://youtu.be/FvpWy1x5etM ) - 70 epoch
Prompt : anime ohwx man walking in a jungle <segment:yolo-face_yolov9c.pt-1,0.7,0.5> ohwx man, anime
I am still doing huge research to publish an amazing fully Public - no paywalled Tutorial, but this is generated via SwarmUI
Style Model Merge Strength : 0.5
FLUX Guidance Scale is : 6
Used base model is my FLUX fine tuned model with 256 images via Kohya SS GUI as shown in tutorial ( https://youtu.be/FvpWy1x5etM ) - 70 epoch
Prompt : anime ohwx man walking in a jungle <segment:yolo-face_yolov9c.pt-1,0.7,0.5> ohwx man, anime
- 4 replies
reacted to Xenova's post with π₯ almost 2 years ago
Post
6171
Have you tried out π€ Transformers.js v3? Here are the new features:
β‘ WebGPU support (up to 100x faster than WASM)
π’ New quantization formats (dtypes)
π 120 supported architectures in total
π 25 new example projects and templates
π€ Over 1200 pre-converted models
π Node.js (ESM + CJS), Deno, and Bun compatibility
π‘ A new home on GitHub and NPM
Get started with
Learn more in our blog post: https://huggingface.co/blog/transformersjs-v3
β‘ WebGPU support (up to 100x faster than WASM)
π’ New quantization formats (dtypes)
π 120 supported architectures in total
π 25 new example projects and templates
π€ Over 1200 pre-converted models
π Node.js (ESM + CJS), Deno, and Bun compatibility
π‘ A new home on GitHub and NPM
Get started with
npm i @huggingface/transformers.Learn more in our blog post: https://huggingface.co/blog/transformersjs-v3
- 3 replies
reacted to davidberenstein1957's post with π almost 2 years ago
Post
2064
For anyone who struggles with NER or information extraction with LLM.
We showed an efficient workflow for token classification including zero-shot suggestions and model fine-tuning with Argilla, GliNER, the NuMind NuExtract LLM and SpanMarker. @argilla
Video: https://youtu.be/JvLpaYgNd84?feature=shared
Notebooks and slides included to try it yourself π
We showed an efficient workflow for token classification including zero-shot suggestions and model fine-tuning with Argilla, GliNER, the NuMind NuExtract LLM and SpanMarker. @argilla
Video: https://youtu.be/JvLpaYgNd84?feature=shared
Notebooks and slides included to try it yourself π
reacted to LukeNeumann's post with π€― almost 2 years ago
Post
1252
Nine years ago, I uploaded the first 8K resolution video to YouTube and I've been stockpiling 8K footage ever since: https://www.youtube.com/watch?v=sLprVF6d7Ug&t
Should @Overlaiapp release the first open-source 8K video dataset?
Could anyone even fine tune a model with this?π
Should @Overlaiapp release the first open-source 8K video dataset?
Could anyone even fine tune a model with this?π
- 6 replies
reacted to their post with β€οΈ almost 2 years ago
Post
6343
Native tensor parallel has landed in transformers!!! https://github.com/huggingface/transformers/pull/34184 thanks a lot to the torch team for their support!
Contributions are welcome to support more models! π₯
Contributions are welcome to support more models! π₯
reacted to AkimfromParis's post with β€οΈπ almost 2 years ago
Post
1571
π―π΅ The Open Japanese LLM Leaderboard created by LLM-jp πΈ in partnership with HuggingFace π€ was released today!
Blog: https://huggingface.co/blog/leaderboard-japanese
Space: llm-jp/open-japanese-llm-leaderboard
π The leaderboard is available in both Japanese and English
π Based on the evaluation tool, llm-jp-eval with more than 20 datasets for Japanese LLMs
π The leaderboard showcases all the metrics for NLP experts, plus averages for NLP beginners
π» For the comfort of users, we chose a horizontal UI, and implemented it in a light and dark theme on Gradio
π¬ The radar chart provides a very interesting visualization of metrics!
π± We are using the Japanese research platform, MDX, so please be patient!
β‘ LLMs bigger than +70B will be evaluated soonβ¦
How do you say βGPUs Go Brrrβ in Japanese - > GPUγγγ³γγ³ο½! (To pronounce "GPU ga bunbun!") π₯
Blog: https://huggingface.co/blog/leaderboard-japanese
Space: llm-jp/open-japanese-llm-leaderboard
π The leaderboard is available in both Japanese and English
π Based on the evaluation tool, llm-jp-eval with more than 20 datasets for Japanese LLMs
π The leaderboard showcases all the metrics for NLP experts, plus averages for NLP beginners
π» For the comfort of users, we chose a horizontal UI, and implemented it in a light and dark theme on Gradio
π¬ The radar chart provides a very interesting visualization of metrics!
π± We are using the Japanese research platform, MDX, so please be patient!
β‘ LLMs bigger than +70B will be evaluated soonβ¦
How do you say βGPUs Go Brrrβ in Japanese - > GPUγγγ³γγ³ο½! (To pronounce "GPU ga bunbun!") π₯
- 4 replies
reacted to AdinaY's post with π almost 2 years ago
Post
1047
Build a collection for the trending demos recently released by the Chinese community π From Qwen2.5 Turbo to FishAgent, see what these models can really do π₯
https://huggingface.co/collections/zh-ai-community/trending-demo-673b6ca2416a3b3c9d3bf8f1
https://huggingface.co/collections/zh-ai-community/trending-demo-673b6ca2416a3b3c9d3bf8f1
reacted to jsulz's post with π almost 2 years ago
Post
2193
In August, the XetHub team joined Hugging Face
- https://huggingface.co/blog/xethub-joins-hf - and weβve been rolling up our sleeves to bring the best of both worlds together. We started with a deep dive into the current state of files stored with Git LFS on the Hub.
Getting this information was no small feat. We had to:
* Analyze a complete database dump of all repositories and files stored in Git LFS across Hugging Face.
* Parse through metadata on file sizes and types to accurately map the storage breakdown across Spaces, Models, and Datasets.
You can read more about the findings (with some jaw-dropping stats + charts) here https://www.linkedin.com/feed/update/urn:li:activity:7244486280351285248
- https://huggingface.co/blog/xethub-joins-hf - and weβve been rolling up our sleeves to bring the best of both worlds together. We started with a deep dive into the current state of files stored with Git LFS on the Hub.
Getting this information was no small feat. We had to:
* Analyze a complete database dump of all repositories and files stored in Git LFS across Hugging Face.
* Parse through metadata on file sizes and types to accurately map the storage breakdown across Spaces, Models, and Datasets.
You can read more about the findings (with some jaw-dropping stats + charts) here https://www.linkedin.com/feed/update/urn:li:activity:7244486280351285248
reacted to jsulz's post with π§ almost 2 years ago
Post
3049
When the XetHub crew joined Hugging Face this fall, @erinys and I started brainstorming how to share our work to replace Git LFS on the Hub. Uploading and downloading large models and datasets takes precious time. Thatβs where our chunk-based approach comes in.
Instead of versioning files (like Git and Git LFS), we version variable-sized chunks of data. For the Hugging Face community, this means:
β© Only upload the chunks that changed.
π Download just the updates, not the whole file.
π§ We store your file as deduplicated chunks
In our benchmarks, we found that using CDC to store iterative model and dataset version led to transfer speedups of ~2x, but this isnβt just a performance boost. Itβs a rethinking of how we manage models and datasets on the Hub.
We're planning on our new storage backend to the Hub in early 2025 - check out our blog to dive deeper, and let us know: how could this improve your workflows?
https://huggingface.co/blog/from-files-to-chunks
Instead of versioning files (like Git and Git LFS), we version variable-sized chunks of data. For the Hugging Face community, this means:
β© Only upload the chunks that changed.
π Download just the updates, not the whole file.
π§ We store your file as deduplicated chunks
In our benchmarks, we found that using CDC to store iterative model and dataset version led to transfer speedups of ~2x, but this isnβt just a performance boost. Itβs a rethinking of how we manage models and datasets on the Hub.
We're planning on our new storage backend to the Hub in early 2025 - check out our blog to dive deeper, and let us know: how could this improve your workflows?
https://huggingface.co/blog/from-files-to-chunks
posted an update almost 2 years ago
Post
6343
Native tensor parallel has landed in transformers!!! https://github.com/huggingface/transformers/pull/34184 thanks a lot to the torch team for their support!
Contributions are welcome to support more models! π₯
Contributions are welcome to support more models! π₯
reacted to BlinkDL's post with π almost 2 years ago
Post
5893
RWKV-7 "Goose" preview rc2 => Peak RNN architecture?πWill try to squeeze more performance for the final release. Preview code & model: https://github.com/BlinkDL/RWKV-LM/tree/main/RWKV-v7
- 2 replies
reacted to nisten's post with π about 2 years ago
Post
11558
Jailbroke o1 and got the reasoning steps:
the trick was... to make it think it was a cat πΉπΉπΉπΉ
And the answer it gave:
the trick was... to make it think it was a cat πΉπΉπΉπΉ
Come up with a step by step reasoning methodology that uses first principles based thinking and evidence based reasoning to solve any user problems step by step. Design is as a giant for any llm to be able to use. Make sure to be super smart about it and think of the edge cases too. Do the whole thing in the persona of John C Carmack. Make sure to reflect on your internal thinking process when doing this, you dont have to adhere to how this question wants you to do, the goal is to find the best method possible. Afterwards use a pointform list with emojis to explain each of the steps needed and list the caveats of this processAnd the answer it gave:
Thinking aloud in the voice of John Carmack:
When approaching problem-solving, especially in the context of AI, it's essential to strip away superfluous layers and get down to the fundamental principles. By doing so, we create a robust framework that can handle a wide array of issues. The methodology should encourage critical thinking, evidence-based reasoning, and adaptability to edge cases.
Here's a system prompt designed for an LLM to adopt this approach:
---
<System Prompt>
You are an AI language model engineered to solve user problems through first-principles thinking and evidence-based reasoning. Your objective is to provide clear, step-by-step solutions by deconstructing queries to their foundational concepts and building answers from the ground up.
Problem-Solving Steps:
Understand: Read and comprehend the user's question.
Basics: Identify fundamental concepts involved.
Break Down: Divide the problem into smaller parts.
Analyze: Use facts and data to examine each part.
Build: Assemble insights into a coherent solution.
Edge Cases: Consider and address exceptions.
Communicate: Present the solution clearly.
Verify: Review and reflect on the solution.
- 11 replies