3 contributors

Keep only best performing model in main (all models still available in develop branch). Best performing model is clip_spanish_141230_samples with a loss of 2.231235980987549

864f8c7 about 3 years ago

logs
Keep only best performing model in main (all models still available in develop branch). Best performing model is clip_spanish_141230_samples with a loss of 2.231235980987549 about 3 years ago
.gitattributes

737 Bytes

Add training scripts and initial model trained on 1% of the data. over 3 years ago
README.md

1.64 kB

Update README and add a training doc about 3 years ago
config.json

4.16 kB

Copy latest model to repo root about 3 years ago
configuration_hybrid_clip.py

4.39 kB

Add actual files instead of symlinks about 3 years ago
discard_incorrect_files.py

1.11 kB

Add all necessary files to replicate training run about 3 years ago
flax_model.msgpack

792 MB
LFS

Copy latest model to repo root about 3 years ago
join_datasets_custom_split.py

1.35 kB

Add all necessary files to replicate training run about 3 years ago
modeling_hybrid_clip.py

18 kB

Add actual files instead of symlinks about 3 years ago
prepare_wit.py

4.27 kB

Add all necessary files to replicate training run about 3 years ago
run-clip.sh

805 Bytes

Add all necessary files to replicate training run about 3 years ago
run_hybrid_clip.py

21.7 kB

Add all necessary files to replicate training run about 3 years ago
scale_convert.py

1.39 kB

Add all necessary files to replicate training run about 3 years ago
test_on_image.py

1.47 kB

Refactor test_on_image.py about 3 years ago
training.md

530 Bytes

Update README and add a training doc about 3 years ago