doing something
@doing_something
710 subscribers
909 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
710 subscribers
doing something
doing something
At least, it works somehow Quantized it using https://pypi.org/project/llmcompressor/ Data used for calibration: https://huggingface.co/datasets/Yehor/cv10-uk-testset-clean-punctuated Quantized model: https://huggingface.co/Yehor/whisper-large-v2-quantized…
Ok, correct configuration gives the result I needed

#ai #speech
🤯1
332 viewsedited  
doing something
doing something
At least, it works somehow Quantized it using https://pypi.org/project/llmcompressor/ Data used for calibration: https://huggingface.co/datasets/Yehor/cv10-uk-testset-clean-punctuated Quantized model: https://huggingface.co/Yehor/whisper-large-v2-quantized…
How to quantize:
https://colab.research.google.com/drive/1TsCMxwq9kqsWV8jabihFN7J78RKgyvnD?usp=sharing

How to inference:
https://colab.research.google.com/drive/1fd6orSZqisnvjMm3Via6CdU247I5JbKo?usp=sharing

#ai #speech #whisper
Google
LLM Compressor for Whisper Large V2: Calibrate the model on Yehor/cv10-uk-testset-clean-punctuated dataset.ipynb
Colab notebook
381 viewsedited  
doing something
Forwarded from Hacker News
Why are QR Codes with capital letters smaller than QR codes with lower-case let
Article, Comments
Terence Eden’s Blog
Why are QR Codes with capital letters smaller than QR codes with lower-case letters?
Take a look at these two QR codes. Scan them if you like, I promise there's nothing dodgy in them.     Left is upper-case HTTPS://EDENT.TEL/ and right is lower-case https://edent.tel/ You can clearly see that the one on the left is a "smaller" QR as it has…
111 views
doing something
#linux
112 views
doing something
Interesting projects:

- https://pypi.org/project/satya/
- https://pypi.org/project/bhumi/

#rust
PyPI
satya
Blazingly fast data validation for Python, powered by Rust
108 views
doing something
https://huggingface.co/papers/2410.02056

#ai #audio
huggingface.co
Paper page - Synthio: Augmenting Small-Scale Audio Classification Datasets with
Synthetic Data
Join the discussion on this paper page
111 views
doing something
https://app.laravel.cloud

#php
Laravel Cloud
Deploy and scale Laravel applications without managing servers. Built-in autoscaling, databases, caching, and security. Get started in under 60 seconds with zero configuration needed.
112 views
doing something
https://github.com/deepseek-ai/FlashMLA

#ai #optimization
GitHub
GitHub - deepseek-ai/FlashMLA: FlashMLA: Efficient Multi-head Latent Attention Kernels
FlashMLA: Efficient Multi-head Latent Attention Kernels - deepseek-ai/FlashMLA
115 views
doing something
https://docs.nvidia.com/nemo-framework/user-guide/24.09/nemotoolkit/tools/intro.html

#ai #speech
109 views
doing something
doing something
How to quantize: https://colab.research.google.com/drive/1TsCMxwq9kqsWV8jabihFN7J78RKgyvnD?usp=sharing How to inference: https://colab.research.google.com/drive/1fd6orSZqisnvjMm3Via6CdU247I5JbKo?usp=sharing #ai #speech #whisper
Quantized Whisper V3 Turbo quite fast

#ai #speech #optimization
107 viewsedited  
doing something
doing something
Published Open Source Crimean Tatar Text-to-Speech dataset to Hugging Face: https://huggingface.co/datasets/Yehor/qirimtatar-tts #ai #speech
Now, it's in different datsets to make it more useful (with train/test splits):

- https://huggingface.co/datasets/speech-uk/qirimtatar-abibullah-tts

- https://huggingface.co/datasets/speech-uk/qirimtatar-sevil-tts

- https://huggingface.co/datasets/speech-uk/qirimtatar-arslan-tts

#ai #speech
huggingface.co
speech-uk/tts-crh-abibullah · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
👍1
121 viewsedited  
doing something
Ukrainian voices for Text-to-Speech now in separated datasets:

https://huggingface.co/collections/speech-uk/ukrainian-text-to-speech-67bd059d61b2598f3a2a7969

#ai #speech
huggingface.co
Ukrainian Text-to-Speech datasets - a speech-uk Collection
Five voices: Mykyta, Oleksa, Lada, Kateryna or Tetiana
🔥2
583 views
doing something
Added a section with phonemizers to https://github.com/egorsmkv/speech-recognition-uk

https://github.com/egorsmkv/speech-recognition-uk/issues/57

#ai #speech
GitHub
GitHub - egorsmkv/speech-recognition-uk: 🇺🇦 Speech Recognition & Synthesis for Ukrainian
🇺🇦 Speech Recognition & Synthesis for Ukrainian. Contribute to egorsmkv/speech-recognition-uk development by creating an account on GitHub.
317 views
doing something
https://github.com/huggingface/setfit

#ai #nlp
GitHub
GitHub - huggingface/setfit: Efficient few-shot learning with Sentence Transformers
Efficient few-shot learning with Sentence Transformers - huggingface/setfit
124 views
doing something
Let's cook, guys

#ai #speech
🔥1
124 views
doing something
https://burn.dev/blog/why-quantization-matters/

#ai #optimization
burn.dev
Why Quantization Matters
Modern deep learning models, such as large language models (LLMs), are heavily constrained by memory bandwidth. GPUs can execute floating-point operations (FLOPs) much faster than they can fetch weights from memory. For instance, an NVIDIA A10 has a peak…
134 views
doing something
https://huggingface.co/blog/Steveeeeeeen/llasagna

#ai #speech #tts #vc
huggingface.co
From Llasa to Llasagna 🍕: Finetuning LLaSA to generates Italian speech and other languages
A Blog post by Steven Zheng on Hugging Face
136 viewsedited  
doing something
doing something
Let's cook, guys #ai #speech
As always, it's #daily_pain
122 views
doing something
doing something
Let's cook, guys #ai #speech
It's done.

https://huggingface.co/datasets/speech-uk/voice-of-america

#ai #asr #speech
🔥2
411 views
doing something
https://t.me/neuroComputational
Telegram
Neural Encoder 🧠🌊📟
Spiky notes on computational neuroscience and biology 🔬🧬 ML / AI / AGI.
Let’s compute and fire together — so we can wire up cross-disciplinary topics.
124 views
doing something
Polars even helps to code

#ai #polars
❤2
119 views