[2/8] Healthsea is a super interesting @spacy_io pipeline to analyze reviews of supplementary products. It has a nice blog post and GitHub repo as well. 💫💫
Interactive tool for object tracking, segmentation, and inpainting. Take a video, click the object of interest, and do things like inpainting and tracking! 😼
🔥15B LLM with 8k context
🥳Trained on permissively-licensed code
💻Acts as tech assistant
🤯80+ programming languages
🚀Open source and data
💫Online demos
🧑💻VSCode plugin
🪅1 trillion tokens
Follow this amazing experience! 🧵
The model can generate and complete code, but it also has nice chatty properties. Some things it can do
- Write code snippets
- Translate Python to C++
- Explain concepts (what's recursion)
- Act as a terminal
- Tell me how to exit vim
🔥
Two models were trained:
- StarCoderBase, trained on 1 trillion tokens from The Stack (hf.co/datasets/bigco…), which is permissively licensed with inspection tools, deduplication and opt-out
- StarCoder, a fine-tuned version of 35B Python tokens
Graphs are **everywhere**, from social media and knowledge systems to molecules and meshes! 🧑🏫
Want to learn about Machine Learning for Graphs? Check out this thread! 🧵
Graphs are denoted as G(V,E), with V being its vertices (nodes in the graph) and E being its edges (connections or links).
In a social network, the vertices would be its users, and the edges would be the connections. 🐦
Many tasks with industry applications can be solved with graphs!
💊 Graph generation: used in drug discovery
🔥 Graph evolution: physics
⚛️Graph level prediction (e.g. classification or regression): e.g. for toxicity of a molecule
🧵I asked ChatGPT to explain what Transfer Learning is, but
- Rapping
- As if I were 5 years old
- A corn lover
- To Homer Simpson teaching Bart
- As Schmidhuber talking with Geoffrey Hinton
Follow this thread! Transfer learning but with some rap! 🎶
A Stable Diffusion model for...anything? 😅I actually don't know much about this model (that's why model documentation is important). The results are great though!