Interesting idea of "code-boarding" using LLMs which explain the code base for new developers

Demo: https://github.com/resemble-ai/chatterbox/blob/4164aa66c46d4a88f06f6cc65a732dcf1ab82db4/.codeboarding/on_boarding.md

#ai #nlp #llm

https://github.com/D4Vinci/Scrapling

#python #data

GitHub added hashes to binary files

Really useful thing

https://huggingface.co/Mozilla/smart-tab-topic

#nlp #ai

Demo with gemma-3-1b-it-Q6_K model.

Generation is faster, pre-filling is enabled

#llm #ai

Fine-tuned Whisper model for Multilingual Speech-to-Phonemes recognition:

https://colab.research.google.com/drive/1cJD_yQMM71IaLxkC0mEgYLOb2DX7p_LA?usp=sharing

#audio #ai

Made a simple Containerfile to run llamafile using podman/docker:

https://github.com/crs-org/multilingual-pii-detection/blob/main/containers/llamafile/Containerfile

Size of the image: 6.68 GB

#ai #llm

I think I've seen it before but let's write here as well about this model - https://huggingface.co/osmosis-ai/Osmosis-Structure-0.6B

This model gives an ability to extract information in text using JSON schema

Quantized model (8-bit) requires about 1602MiB of GPU

#ai #llm

https://arxiv.org/abs/2409.09506

#ai #audio

nota bene

Maybe this update is number three in the last two weeks

Google cooking very well on its AI

https://qwen3.org/vl/

#vision #ai

https://github.com/sparckles/Robyn/releases/tag/v0.69.0

#rust #python

Mercator projection vs real size