Overview of Hugging Face platform - 2024-10-24
0 likes•70 views
The PyData Venice #17 meetup on October 24, 2024, will feature presentations by Sara Ferro and Alessandra Bilardi on fine-tuning transformers for historical document transcription and an overview of Hugging Face, respectively. The event encourages networking and covers current trends, tools, and partnerships within the Hugging Face ecosystem. Future meetups are scheduled for December 3-6, 2024, focusing on global data science topics.
![🤗 Hugging Face resources
https://huggingface.co/docs
Resource Quantity
huggingface_hub[cli] (manager of spaces) datasets and files of spaces
huggingface_hub python library all in Hugging Face Hub
Spaces infinity
Web interfaces for spaces (docker, gradio or streamlit) free
Hardware for spaces (with 2 vCPU 16 GB or ZeroGPU) free
Inference API (serverless) 1000 requests / day
Organization free](https://image.slidesharecdn.com/pydatavenice17-241101131814-fab41daf/85/Overview-of-Hugging-Face-platform-2024-10-24-13-320.jpg)
Overview of Hugging Face platform - 2024-10-24
- 1. Giovedì 24 Ottobre 2024 @PyDataVenice #17 #Meetup #PyData in presenza e in streaming alle ore 19:00 Sara Ferro Post Doc @ CCHT Alessandra Bilardi Data / Automation Specialist @ Corley Cloud #HuggingFace and #libraries How to fine-tune a Transformer to transcribe historical documents Hugging Face Overview Venice
- 2. Promotori di PyData Venice #17
- 3. Promotori di PyData Venice #17 ︖ ︖
- 4. Proposte
- 7. Hugging Face Overview @PyDataVenice #17 #Meetup #PyData Alessandra Bilardi - Data & Automation Specialist @ Corley Cloud
- 8. Agenda Hugging Face in numbers Hugging Face packages and third party tools The platform
- 9. Hugging Face in numbers
- 10. 🤗 Hugging Face history https://en.wikipedia.org/wiki/Hugging_Face Event When Notes Foundation 2016 Clément Delangue (CEO), Julien Chaumond (CTO), Thomas Wolf (CSO) BLOOM (Big-science Large Open-science Open-access Multilingual language model) 2021 - 2022 Open LLM with research groups 176 billion parameter Private Hub August 2022 Support SaaS or on-prem deployment Gradio December 2022 The fastest way to demo your ML model with a friendly web interface Partnerships 2023 - 2024 AWS, Google, Nvidia, AMD, Intel, IBM, ..
- 12. 🤗 Hugging Face sections https://huggingface.co/ Name section Hugging Face Kaggle Name section Datasets 230K 389K (298K) Datasets Models 1M 9200 (2300) Models Spaces (Gradio & Streamlit) over 500K 1.2M (1M) Code Posts, Blog, Articles, Discussions and Learn Discussions and Learn Hardware for spaces Hardware for notebooks Enterprise Hub / Competitions Competitions Hugging Face Hub https://github.com/huggingface/course
- 13. 🤗 Hugging Face resources https://huggingface.co/docs Resource Quantity huggingface_hub[cli] (manager of spaces) datasets and files of spaces huggingface_hub python library all in Hugging Face Hub Spaces infinity Web interfaces for spaces (docker, gradio or streamlit) free Hardware for spaces (with 2 vCPU 16 GB or ZeroGPU) free Inference API (serverless) 1000 requests / day Organization free
- 14. Hugging Face packages and third party tools
- 15. 🤗 Hugging Face packages on pypi.org Resource Notes accelerate, optimum, peft Hardware / software optimization datasets, tokenizers, evaluate Basic packages diffusers, transformers https://github.com/huggingface/transformers.js gradio The fastest way to demo your ML model with WI huggingface_hub https://github.com/huggingface/huggingface.js https://huggingface.co/docs/huggingface_hub/index https://github.com/huggingface
- 16. 🤗 Client use cases
- 17. 🤗 Client use cases
- 18. 🤗 Client use cases
- 19. 🤗 Use cases tools https://github.com/ggerganov/llama.cpp Command line tool (GitHub contributors) ● ggerganov/llama.cpp (915) ● vllm-project/vllm (601) ● ggerganov/whisper.cpp (419) User Interface (GitHub contributors) ● oobabooga/text-generation-webui (332) ● ollama/ollama (316) ● nomic-ai/gpt4all (115)
- 20. 🤗 Use cases tools https://github.com/ggerganov/llama.cpp Command line tool (GitHub contributors) ● ggerganov/llama.cpp (915) ● vllm-project/vllm (601) ● ggerganov/whisper.cpp (419) Packages (GitHub contributors) ● 🤗 /transformers (2799), 🤗 /diffusers (816) ● gradio-app/gradio (339), streamlit (239) ● 🤗 /huggingface_hub (206) User Interface (GitHub contributors) ● oobabooga/text-generation-webui (332) ● ollama/ollama (316) ● open-webui/open-webui (236) Frameworks (GitHub contributors) ● tensorflow/tensorflow (3604) ● langchain-ai/langchain (3177) ● run-llama/llama_index (1275) https://blog.apify.com/langchain-alternatives/
- 21. The platform
- 22. Questions ? @PyDataVenice #17 #Meetup #PyData
- 23. How to fine-tune a Transformer to transcribe historical documents @PyDataVenice #17 #Meetup #PyData Sara Ferro - Research Fellow @ CCHT
- 24. Questions ? @PyDataVenice #17 #Meetup #PyData
- 25. Excursus ● 10 👥 21/09/19, PyDataVE #0 ● 9 👥 13/12/19, PyDataVE #1 - Introduzione ai Notebook ● 23 👥 31/03/20, PyDataVE #2 - DataViz (geo e non) ● .. ● 26 👥 19/04/24, PyDataVE #15 - OpenCV Use Cases ● 66 👥 27/06/24, PyDataVE #16 - ComputerVision & DecisionMaking ● 59 👥 24/10/24, PyDataVE #17 - HuggingFace and libraries YouTube: @pydatavenice/videos & @pydatavenice/streams @PyDataVenice #17 #Meetup #PyData
- 26. Prossimi incontri ● 3 - 5 dicembre, https://pydata.org/global2024
- 27. Prossimi incontri ● 3 - 5 dicembre, https://pydata.org/global2024 ● 5 - 6 dicembre, https://pydata.org/global2024/impact-program
- 28. Prossimi incontri ● 3 - 5 dicembre, https://pydata.org/global2024 ● 5 - 6 dicembre, https://pydata.org/global2024/impact-program ● YouTube: @PyDataTV/playlists
- 29. Prossimi incontri ● 3 - 5 dicembre, https://pydata.org/global2024 ● 5 - 6 dicembre, https://pydata.org/global2024/impact-program ● YouTube: @PyDataTV/playlists ● giovedì 12 dicembre ore 19:00
- 30. Prossimi incontri ● 3 - 5 dicembre, https://pydata.org/global2024 ● 5 - 6 dicembre, https://pydata.org/global2024/impact-program ● YouTube: @PyDataTV/playlists ● giovedì 12 dicembre ore 19:00 E n g l i s h
- 31. Proposte
- 32. Thanks for listening. @PyDataVenice #17 #Meetup #PyData