lewtun

# This is a modified version of TRL's `SFTTrainer` example (https://github.com/huggingface/trl/blob/main/examples/scripts/sft_trainer.py), 
# adapted to run with DeepSpeed ZeRO-3 and Mistral-7B-V1.0. The settings below were run on 1 node of 8 x A100 (80GB) GPUs.
#
# Usage:
#   - Install the latest transformers & accelerate versions: `pip install -U transformers accelerate`
#   - Install deepspeed: `pip install deepspeed==0.9.5`
#   - Install TRL from main: pip install git+https://github.com/huggingface/trl.git
#   - Clone the repo: git clone github.com/huggingface/trl.git
#   - Copy this Gist into trl/examples/scripts
#   - Run from root of trl repo with: accelerate launch --config_file=examples/accelerate_configs/deepspeed_zero3.yaml --gradient_accumulation_steps 8 examples/scripts/sft_trainer.py  

scottire

import numpy as np
import panel as pn
import holoviews as hv 
hv.extension("bokeh")
# hv.extension('matplotlib')
from holoviews.streams import Stream, Params
from scipy.io import wavfile
from scipy.signal import spectrogram

rate, data = wavfile.read('/filepath.wav')

glasslion

"""
Convert YouTube subtitles(vtt) to human readable text.

Download only subtitles from YouTube with youtube-dl:
youtube-dl  --skip-download --convert-subs vtt <video_url>

Note that default subtitle format provided by YouTube is ass, which is hard
to process with simple regex. Luckily youtube-dl can convert ass to vtt, which
is easier to process.

torjusb

var clipboardData = window.clipboardData || event.clipboardData || event.originalEvent && event.originalEvent.clipboardData;

var pastedText = clipboardData.getData("Text") || clipboardData.getData("text/plain");

if (!pastedText && pastedText.length) {
    return;
}

// Parse the pasted text from Excel into rows.
// Pasted text is usually separated by a new line for each row,

snipe

"\e[A": history-search-backward
"\e[B": history-search-forward
set show-all-if-ambiguous on
set completion-ignore-case on