Fix incorrect segment output format
d7725ec
Joshua Lochner
commited on
Spaces:
Running
Running
Commit History
Update default classifier model
6a8bf30
Joshua Lochner
commited on
Separate missing and incorrect detection logic
dccb47b
Joshua Lochner
commited on
Add safe_print method
67d0193
Joshua Lochner
commited on
Output raw text during evaluation
c0313f5
Joshua Lochner
commited on
Fix duplicate argument in function definition
bb853de
Joshua Lochner
commited on
Use logger instead of printing when loading datasets
eaa79a8
Joshua Lochner
commited on
Fix training arguments dataclasses
d34e3fe
Joshua Lochner
commited on
Code formatting
7dbc778
Joshua Lochner
commited on
Fix classifier preprocessing
508e8b2
Joshua Lochner
commited on
Add max_segment_duration argument
2115d78
Joshua Lochner
commited on
Delete moderate.py
18c7914
Joshua Lochner
commited on
Use new classifier for evaluation
787a8df
Joshua Lochner
commited on
Remove PreprocessingDatasetArguments class
643d00a
Joshua Lochner
commited on
Removed unused code
a0ca50e
Joshua Lochner
commited on
Improve prediction pipeline
813b772
Joshua Lochner
commited on
Merge duplicated training dataclasses
490a61c
Joshua Lochner
commited on
Ignore prediction if categorized as nothing by classifier and extractor
e77b67b
Joshua Lochner
commited on
Set device to cpu (-1) if device index is None
79b40d9
Joshua Lochner
commited on
Upgrade classifier to transformer-based model
36f7534
Joshua Lochner
commited on
Add language preference list
62ea1e5
Joshua Lochner
commited on
Fix logging messages in predict script
4d4de75
Joshua Lochner
commited on
Only consider spoken words when calculating metrics
9f15397
Joshua Lochner
commited on
Ensure event duration is non-negative
2439d9a
Joshua Lochner
commited on
Remove zero-width spaces from text
884d564
Joshua Lochner
commited on
Add support for mute action type and remove videos with full action type
1286fe5
Joshua Lochner
commited on
Initialize logging in each script
c4f250e
Joshua Lochner
commited on
Do not allow predictions to miss start of video
aa018be
Joshua Lochner
commited on
Fix `--no_cuda` argument for preprocessing
87b2dec
Joshua Lochner
commited on
Revert model input size back to 512 tokens
721bf64
Joshua Lochner
commited on
Fix conflicting `--no_cuda` argument
09cabec
Joshua Lochner
commited on
Use correct logger per script
e3d3d3f
Joshua Lochner
commited on
Update preprocessing script to use logging module
cfbd4d5
Joshua Lochner
commited on
Add `no_cuda` argument to not use GPU
de9c8c4
Joshua Lochner
commited on
Remove redundant calls to change device
8981122
Joshua Lochner
commited on
Add `output_as_json` argument for inference
52340fc
Joshua Lochner
commited on
Adjust tokenizer input size based on model input size
9604abd
Joshua Lochner
commited on
Remove unused utilities
0e18e8c
Joshua Lochner
commited on
Move `load_datasets` to train script
086ca93
Joshua Lochner
commited on
Improve how transcripts are stored and how manual transcripts are segmented
583f4cf
Joshua Lochner
commited on
Add boilerplate code to detect whether segment was split due to length
df35612
Joshua Lochner
commited on
Revert evaluation script to use `processed_file` by default
8fc746d
Joshua Lochner
commited on
Fix segmentation using binary search
de9c264
Joshua Lochner
commited on
Add fallback for old transcript version
c445f1a
Joshua Lochner
commited on
Fix `num_tokens` key in words
83dc695
Joshua Lochner
commited on
Optimize segment generation and extraction
4b4c9f0
Joshua Lochner
commited on
Abstract inference code
8b71088
Joshua Lochner
commited on
Improve caching and downloading of classifier for predictions
fb87012
Joshua Lochner
commited on
Create `ClassifierLoadError`
02e576a
Joshua Lochner
commited on
Download classifier and vectorizer if not present
d7a594b
Joshua Lochner
commited on