Releases: ControlNet/tensorneko
Releases · ControlNet/tensorneko
[0.3.14] TensorNeko Release
Fix:
- Fix
ar_1d
AR calculation when the num proposals is less than the max_proposals.
[0.3.13] TensorNeko Release
Features:
- Add
file_key
arguments forap_1d
andar_1d
- Add stub for
tensorneko_lib
Fix:
- Wrong computation of
ap_1d
[0.3.12] TensorNeko Release
Features:
- new library
tensorneko_lib
for high performance rust implementations. - npy reader and writer in
tensorneko_util.io
tensorneko_util.Timer.lines
to automatically time each lines in the wrapped function.
Fix:
- opencv image writer cannot save tensor as image.
[0.3.11] TensorNeko Release
Features:
Singleton
support with class with init args- More utils for download files
- Typing for
Registry
[0.3.10] TensorNeko Release
Features:
Registry
for easier decorator-style class/func registrationread
andwrite
are also in the top level oftensorneko_util
[0.3.9] TensorNeko Release
Features:
JsonReader
now can load jsonl filesWeightReader
andWeightWriter
that can read and write thept
,ckpt
,safetensors
filesread
andwrite
now is in the top level of thetensorneko
packageimport_tqdm
andimport_tqdm_auto
now are not private
[0.3.8] TensorNeko Release
Features:
- Group the metrics, losses, and system stats in logger
- Add
input_shape
arg forBinaryClassifier
Fix:
NekoModel
cannot train when the logger isNone
[0.3.7] TensorNeko Release
Fix:
- Training error when using
lightning
2.1 - Docs are not clear for
VideoReader
,RoundRobinDataset
,NestedDataset
,ListDataset
.
[0.3.6] TensorNeko Release
Features:
- Better type hint for eventbus and dispatch
- Dynamic import for
mita_tqdm
on_step
andon_epoch
arguments forGpuStatsLogger
andSystemStatsLogger
Timer
now doesn't require parentheses as decorator
Fix:
crop_with_padding
using numpy arraysample_indexes
not importedFID
not working withprogress_bar=True
- import error with pytorch lightning 2.1.3
[0.3.5] TensorNeko Release
Features:
- Add support for pytorch 2.1 and lightning 2.1
- Auto infer for
notebook.display
, now supports audio, video, image, and code. ListDataset
, likeTensorDataset
intorch
but directly uses listSequentialIterSampler
, can split dataset to multiple epochs. Normally can be used for training in large-scale dataset but want to utilize the learning rate scheduler based on epochs.BinaryClassifier
. A predefined model for binary classification.sample_indexes
. Sample a continious indexes giving a source sequence length.
Fix:
- Import problem raise by
RoundRobinDataset
FID
calculation problem when setnum_workers > 0
.iou_2d
calculation error