doing something
@doing_something
712 subscribers
912 photos
36 videos
36 files
1.94K links
@smlkw doing something, just my notes to keep updates
Download Telegram
About
Blog
Apps
Platform
Join
doing something
712 subscribers
doing something
https://github.com/kyutai-labs/sphn/issues/12

#daily_pain
GitHub
python 3.11 and sphn 0.1.8 (from PyPI) gives an error · Issue #12 · kyutai-labs/sphn
I use https://github.com/Rikorose/DeepFilterNet to clean signals and when I've updated sphn to 0.1.8 it does not work as expected. Issue: TypeError: argument 'input': 'ndarray' ...
123 views
doing something
"Mom, I wanna be him"
124 views
doing something
Using olmOCR to get text data from JFK assasination files

#cv #ai
👍1
120 views
doing something
doing something
Using olmOCR to get text data from JFK assasination files #cv #ai
https://huggingface.co/datasets/Yehor/jfk-pdfs-ocr
huggingface.co
Yehor/jfk-pdfs-ocr · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
106 views
doing something
doing something
Using olmOCR to get text data from JFK assasination files #cv #ai
A colab to reproduce:

https://colab.research.google.com/drive/1xsXHYIaEU50tS0eTwuI5QdkGrbvjHQpW?usp=sharing

#cv
Google
olmOCR for JFK Assasination Records.ipynb
Colab notebook
443 views
doing something
https://github.com/ml-explore/mlx/releases/tag/v0.24.0

#mlx
GitHub
Release v0.24.0 · ml-explore/mlx
Highlights

Much faster fused attention with support for causal masking

Benchmarks
Improvements in prompt processing speed and memory use, benchmarks
Much faster small batch fused attention for e....
121 views
doing something
Yehor Smoliakov
👋🏻 Вітаю, шановне товариство! Багато людей знає, що я займаюся в ШІ направленням розпізнавання та синтезу мови. Зараз я організував ініціативу Speech-UK (підписуйтесь на Hugging Face) в яку вже опублікував дві колекції з датасетами по українській та кр…
I’m sampling AI HOUSE podcast to test stability of a tool that prepares the data

#rust #audio
125 viewsedited  
doing something
doing something
I’m sampling AI HOUSE podcast to test stability of a tool that prepares the data #rust #audio
https://huggingface.co/datasets/speech-uk/test-ai-house-podcast
huggingface.co
speech-uk/test-ai-house-podcast · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
130 views
doing something
Forwarded from Feed-Master
This Week in Rust 591
this-week-in-rust.org
This Week in Rust 591 · This Week in Rust
127 views
doing something
https://github.com/langgenius/dif

#llm
127 views
doing something
https://github.com/niklak/dom_smoothie

#rust
GitHub
GitHub - niklak/dom_smoothie: A Rust crate for extracting readable content from web pages.
A Rust crate for extracting readable content from web pages. - niklak/dom_smoothie
127 views
doing something
https://github.com/huggingface/transformers/releases/tag/v4.50.0

#ml
GitHub
Release Release v4.50.0 · huggingface/transformers
Release v4.50.0
New Model Additions
Model-based releases
Starting with version v4.49.0, we have been doing model-based releases, additionally to our traditional, software-based monthly releases. Th...
117 views
doing something
doing something
https://github.com/huggingface/transformers/releases/tag/v4.50.0 #ml
😐
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥2
114 views
doing something
doing something
https://github.com/huggingface/transformers/releases/tag/v4.50.0 #ml
https://quark.docs.amd.com/latest/index.html

#optimization
114 viewsedited  
doing something
https://arxiv.org/abs/2503.05931

#ai #audio
arXiv.org
Training and Inference Efficiency of Encoder-Decoder Speech Models
Attention encoder-decoder model architecture is the backbone of several recent top performing foundation speech models: Whisper, Seamless, OWSM, and Canary-1B. However, the reported data and...
117 views
doing something
doing something
https://arxiv.org/abs/2503.05931 #ai #audio
Gonna read it
128 views
doing something
Apple recently added ability to record phone calls

Tested my Gemini bot with these recordings

So, it works well

Next, it would be awesome to do RAG over these transcripts
150 views
doing something
Here's a popular alternative to Pandas - Polars

It's written in Rust

I have adapted their code in my tiny utility to extract audio files from parquet files in parallel using rayon

All we need is blazingly fast programs

#rust
138 views
doing something
doing something
https://github.com/glotlabs/gdrive #rust
So useful, uploading 110 GB to gdrive

https://github.com/glotlabs/gdrive

#rust
127 viewsedited  
doing something
https://iree.dev/guides/ml-frameworks/onnx/#samples

Learned that we can upgrade ONNX version using https://github.com/onnx/onnx

#ml #optimization
118 views