Erlang Developers 🇺🇦

The largest #Erlang Developers Group on Linkedin

[Mark Ericksen @brainlid]

Wanted an @ElixirLang version of #LangChain so I created it! Happy to announce the initial release! Do you want to experiment with creating a #ChatGPT integration into your #ElixirLang app? Now it’s easy! Check out the announcement post! More soon! https://t.co/urttbRpU1Y

[Erlang Developers 🇺🇦 @ErlangDevelop]

Wow! Llama-2 #LLM (13B learned parameters) running on #GPU backend in #Elixir 💥

[Chris McCord @chris_mccord]

👀 Llama2-13b running on GPU with Elixir/Bumblee/Phoenix

[Erlang Developers 🇺🇦 @ErlangDevelop]

Happy Birthday to #NodeJS 🎂🎉🥳

[Node.js @nodejs]

14 years of Node.js 💚

To celebrate this milestone, we invite our community to contribute to our Node.js time capsule. Share your favorite memories, stories, features or swag, or share your predictions for the future of Node.js.

Let us know below!

[Loreto Parisi @loretoparisi]

Never seems a comparison between running 🦙Llama on Apple Silicon with MPS in 🐍 #Python and @ggerganov C++ in terms of peak memory, cpu 🌡️Temperature and 🪫power consumption and - a new 📏 measure unit to me - Tokens/1 Wh (Tokens per 1 Watt per Hour).
https://t.co/LgU9PlF0Tk

[Loreto Parisi @loretoparisi]

#WebGPU benchmarks for #LLM inference in the browser: encoding: 27.7424 tokens/sec, decoding: 13.3699 tokens/sec 🏎️ model: vicuna-7b int4 quantized sharded; tokenizer: SentencePiece/BPE WASM module; backend: @ApacheTVM Unity / IRModule (TensorIR) + @emscripten

[Loreto Parisi @loretoparisi]

How to serialize/deserialize to #Protobuf a #JSON object in #javascript #NodeJS
https://t.co/yEVOXejrRd

[Loreto Parisi @loretoparisi]

65k context window (65536 tokens!) for the first time in a #LLM available with Apache 2.0 License. But what does it mean 65 window in practice? [1/3]

[Loreto Parisi @loretoparisi]

It's interesting that a pure RNN large language model is in the top 10 #LLM. It's @BlinkDL_AI Raven 14 billion parameters. It has no context length limits, the embedding is in the last hidden state, there is no attention. Ah, code is 150 #Python lines.
https://t.co/niIrPkVF9I

[Erlang Developers 🇺🇦 @ErlangDevelop]

#erlang developers 🎂🎂🎉

[Loreto Parisi @loretoparisi]

Woah! @elixirlang powered by @huggingface spaces is huge!

[German Velasco @germsvel]

There are two ways to write @elixirphoenix LiveView form tests:

- Using view |> render_submit/3
- Using view |> form/3 |> render_submit/2

🔥Always use the `form/3` helper🔥

It validates the form data against the HTML to ensure the elements exists!

#MyElixirStatus

[Loreto Parisi @loretoparisi]

Diffusion Transformers (DiTs) are a new class of latent diffusion models (LDM) based on the transformers, obtained replacing the U-Net backbone with a transformer that operates on latent patches. DiTs are very similar to Vision Transformer (ViT)
https://t.co/w3nB9BlVTn

[Erlang Developers 🇺🇦 @ErlangDevelop]

How to run #GPT #stablediffusion in #elixir Bumblebee server less. 💥

[Loreto Parisi @loretoparisi]

From GPT2 to #StableDiffusion: @huggingface arrives to the #Elixir #Erlang community 💥💥💥
https://t.co/Kch1fwoBL2

[Erlang Developers 🇺🇦 @ErlangDevelop]

#deeplearning for #computervision in #erlang 💥

[Grey Neurons - Experience the wisdom @GreyNeurons]

A nice article on why Sonic Pi ( https://t.co/ZlL3x3msy0 ) decided to rewrite some part of its server using Erlang. Pleasantly surprised to know that initial version was written by Joe Armstrong

https://t.co/AidKrG8fp2

#elixir #erlang

[Erlang Developers 🇺🇦 @ErlangDevelop]

❤️❤️❤️

[Grey Neurons - Experience the wisdom @GreyNeurons]

A nice article on why Sonic Pi ( https://t.co/ZlL3x3msy0 ) decided to rewrite some part of its server using Erlang. Pleasantly surprised to know that initial version was written by Joe Armstrong

https://t.co/AidKrG8fp2

#elixir #erlang

[Erlang Developers 🇺🇦 @ErlangDevelop]

#ComputerVision in @elixirphoenix live view right in the #browser running torch @onnxruntime converted model! 💥 #myelixirstatus