A hint about Clibrain Lab 🧪 next release...

It seems that TPUs v5 will be the winning chip in terms of throughput per dollar ⚙️/💲. Highly recommended for inference⚡

There is a huge difference between hosting a fine-tuned model in your own environment where you have full control, compared to using a managed provider, where you share resources and infrastructure with others. 🤔

Price per million tokens might not be the best metric for that🤷🏻‍♂️

Research 🔬 in #LLMs is expensive, both in terms of money and time 💸⌛ Thank you to all who contribute to it 🤗

231 ⭐️ Thank you! 🙏
star-history.com/#mrm8488/share…

github.com/mrm8488/shared…

Check out amazing work of Hugging Face Fellows in collections 👉 huggingface.co/hugging-fellows 🖼️ 🎨
Browse each Fellow's collection on the sidebar 🤗
Learn more about our Fellows program here 👉 huggingface.co/blog/fellowship

Oh shit! I am here..

Good to know

👀

Companies are currently deciding whether to invest in fine-tuning their own open LLMs (e.g. Llama 2) or to use closed LLMs via an API (e.g. Claude or GPT4).

Of course, you can build quick & cheap prototypes with closed LLMs, but fine-tuning on high quality domain data is cheaper…

🚀 Explore the creative frontier! Crave a #MidJourney inspired image but unsure of the prompt? 🤔 Discover Hugging Face space by the innovative Clibrain labs team! 🌟 Simply upload a picture 📸, grab the caption crafted by #IDEFICS 🖋, and unleash it on your text2img model! 🎨…

MURCIA 🚄 MADRID ✈️ NYC 🗽

I highly recommend this course of Sebastian Raschka

🔵 ojooo, recordemos que si bien OpenAI anunció que #CharGPT usará su versión multimodal (es decir que ahora puede recibir e interpretar imágenes) #Bard de Google AI lo lleva haciendo ya hace unos meses y con buenos resultados, la competencia sigue y se viene “Gemini” por parte…

Or play with 🖼️➡️🔤 'image to Midjourney like captions' (describe function) here: huggingface.co/spaces/mrm8488…