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
doing something
bs=24, whisper large-v3 https://github.com/egorsmkv/optimized-whisper #ai #asr
Tested a new Turbo model with HQQ, below an example how a ~2 hour audio is recognized
300 views
doing something
doing something
arestovich_ukrainian.txt
turbo_arestovich_ukrainian.txt
159.7 KB
- All Duration: 6465.6441
- All RTF: 0.003
- All elapsed: 19.2815
304 views
doing something
doing something
turbo_arestovich_ukrainian.txt
v3-arestovich.txt
153.1 KB
This test was on a 3090 card, so this file is also tested on the same hardware and software

RTF:

- V3: 0.0103
- V3 turbo: 0.003
316 viewsedited  
doing something
doing something
# test file: batch_1.zip - 715M RAYON_NUM_THREADS=25 ./hyperfine --runs 10 './ripunzip unzip-file --output-directory ripunzip-files batch_1.zip' Benchmark 1: ./ripunzip unzip-file --output-directory ripunzip-files batch_1.zip Time (mean ± σ): 745.1…
Powerful thing
56 views
doing something
https://github.com/skirdey/voicerestore

#ai #audio
GitHub
GitHub - skirdey/voicerestore: VoiceRestore: Flow-Matching Transformers for Universal Speech Restoration
VoiceRestore: Flow-Matching Transformers for Universal Speech Restoration - skirdey/voicerestore
55 views
doing something
https://www.youtube.com/watch?v=FH5wiwOyPX4

#ai #optimization
YouTube
GPU MODE IRL 2024 Keynotes
The server met IRL for the first time in the beautiful Accel office https://t.co/hq24uIE0fG

Since then we've rebranded to GPU MODE to be more inclusive of different hardware vendors and programming languages discord.gg/gpumode

We were lucky to invite some…
55 views
doing something
https://huggingface.co/spaces/jasperai/Flux.1-dev-Controlnet-Upscaler

#cv #genai
huggingface.co
Flux.1-dev Upscaler - a Hugging Face Space by jasperai
This app lets you upload a small picture and increase its size using an advanced AI upscaler. You can choose how much to enlarge the image and adjust a few settings, then receive a clearer, higher‑...
59 views
doing something
doing something
https://huggingface.co/spaces/jasperai/Flux.1-dev-Controlnet-Upscaler #cv #genai
Demo
57 views
doing something
https://github.com/slashml/amd_inference

#ai
GitHub
GitHub - slashml/amd_inference: Docker-based inference engine for AMD GPUs
Docker-based inference engine for AMD GPUs. Contribute to slashml/amd_inference development by creating an account on GitHub.
52 views
doing something
https://huggingface.co/grounded-ai/phi3.5-hallucination-judge

#nlp #ai
huggingface.co
grounded-ai/phi3.5-hallucination-judge · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
57 views
doing something
https://colab.research.google.com/drive/1o9b2JQ8l9a39uOZZi9DWXQ15BXlCIfEu?usp=sharing

600M model with 4-bits uses about 1GB of VRAM

#nlp #ai #hqq
Google
Quantized NLLB using HQQ.ipynb
Colab notebook
376 viewsedited  
doing something
doing something
decord can lag (it does not get the batch of frames) on GPU device if they have high values
This story has ended with a script that uses StreamReader from torchaudio.

I'm using the seek method to skip unrelated frames.

One important thing is that you need compiled ffmpeg with CUDA to do appropriate decoding.
64 viewsedited  
doing something
https://engineering.fb.com/2024/10/03/ml-applications/open-source-ai-population-maps-meta/
Engineering at Meta
How open source AI can improve population estimates, sustainable energy, and the delivery of climate change interventions
Data for Good at Meta is open-sourcing the data used to train our AI-powered population maps. We’re hoping that researchers and other organizations around the world will be able to leverage these t…
68 views
doing something
An idea for a weekend project:

Adapt https://github.com/Gadersd/whisper-burn to the recently publish Whisper Turbo model.

Need to make only two things:

- update the repo to latest burn 🔥 version
- get down the number of decoder layers

#ai #asr
GitHub
GitHub - Gadersd/whisper-burn: A Rust implementation of OpenAI's Whisper model using the burn framework
A Rust implementation of OpenAI's Whisper model using the burn framework - Gadersd/whisper-burn
57 viewsedited  
doing something
https://youtube.com/playlist?list=PLoROMvodv4rPOWA-omMM6STXaWW4FvJT8&feature=shared

https://deepgenerativemodels.github.io/

#ai #genai
YouTube
Stanford CS236: Deep Generative Models I 2023 I Stefano Ermon
For more information about Stanford's Artificial Intelligence programs visit: https://stanford.io/ai View the course website: https://deepgenerativemodels.gi...
73 viewsedited  
doing something
Finally we can cut texts from memes to make other ones

https://huggingface.co/spaces/OzzyGT/diffusers-image-fill

#ai #genai
58 views
doing something
A nice tool to make models smaller by compression

https://github.com/zipnn/zipnn
302 views