Update README.md

Add GitHub sponsor link to repo

Update README.md

Merge pull request #3 from IThinkIAmCool/patch-1

added native support for v query parameter.

Fix incorrect segment output format

Update default classifier model

Separate missing and incorrect detection logic

Add safe_print method

Output raw text during evaluation

Fix duplicate argument in function definition

Use logger instead of printing when loading datasets

Fix training arguments dataclasses

Code formatting

Fix classifier preprocessing

Add max_segment_duration argument

Delete moderate.py

Use new classifier for evaluation

Remove PreprocessingDatasetArguments class

Removed unused code

Improve prediction pipeline

Merge duplicated training dataclasses

Ignore prediction if categorized as nothing by classifier and extractor

Set device to cpu (-1) if device index is None

Update streamlit app to use new classifier

Upgrade classifier to transformer-based model

Update requirements.txt

Add output folder to .gitignore

Remove classifiers folder

Add language preference list

Fix logging messages in predict script

Only consider spoken words when calculating metrics

Ensure event duration is non-negative

Remove zero-width spaces from text

Fix classifier train command

Add support for mute action type and remove videos with full action type

Initialize logging in each script

Do not allow predictions to miss start of video

Fix `--no_cuda` argument for preprocessing

Revert model input size back to 512 tokens

Fix conflicting `--no_cuda` argument

Use correct logger per script

Update preprocessing script to use logging module

Add `no_cuda` argument to not use GPU

Update README to include installation instructions

Fix button colour on dark theme

Remove redundant calls to change device

Add `output_as_json` argument for inference

Adjust tokenizer input size based on model input size

Fix typo in prediction command