Releases: Labbeti/conette-audio-captioning
Releases · Labbeti/conette-audio-captioning
Version 0.3.2
Version 0.3.1
[0.3.1] 2024-06-10
Fixed
- Remove duplicated files in
results
folder. - CNN10 and CNN14D pre-process transform instantiation.
Version 0.3.0
[0.3.0] 2024-04-18
Changed
- Update dependencies with
torchoutil
, and clean a lot of dead code. - Convnext model (BL version) is now automatically selected during HDF preparation. (#5)
Fixed
Version 0.2.2
[0.2.2] 2024-01-15
Added
- Multiple candidates, predictions and probabilities in model outputs.
train_and_enable_grad
method inCoNeTTEModel
class.
Changed
- Rename
eval_and_detach
toeval_and_disable_grad
inCoNeTTEModel
class.
Version 0.2.1
[0.2.1] 2024-01-12
Added
conette-predict
now support CNext-trans (baseline) model.
Version 0.2.0
[0.2.0] 2024-01-12
Added
- CoNeTTE training source code, with entire data processing.
- ConvNeXt-trans baseline training source code, with entire data processing.
- ConvNeXt tag logits to CoNeTTE model outputs during inference.
Version 0.1.4
[0.1.4] 2023-11-20
Fixed
- Fix forbid repetition mode argument.
Version 0.1.3
[0.1.3] 2023-11-20
Added
- Forbid repetition mode argument to LightningModule and HuggingFace wrapper.
Version 0.1.2
[0.1.2] 2023-11-17
Fixed
- Task embeddings inputs
wavcaps_audioset_sl
andwavcaps_bbc_sound_effects
.
Version 0.1.1
[0.1.1] 2023-11-09
Added
- Unittests for hf model.
Fixed
- Fix sample path for PyPI package.