mirrors
/
fast-reid
mirror of https://github.com/JDAI-CV/fast-reid.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
fast-reid/CHANGELOG.md

39 lines
832 B
Markdown
Raw Permalink Normal View History

feat: support multi-teacher kd Summary: support multi-teacher kd with logits and overhaul distillation
2021-01-29 17:25:31 +08:00
# Changelog
faster dataloader with pre-fetch and cuda stream (#456) Summary: add a background thread to create a generator with pre-fetch, and create a new cuda stream to copy tensor from cpu to gpu in parallel. Reviewed by: l1aoxingyu
2021-04-12 15:03:35 +08:00
### v1.3
Bugfix for cls_layer In `any_softmax`, all operations are in-place, so pass into the `logits.clone()` to prevent outside logits changed.
2021-05-31 17:32:24 +08:00
#### New Features
- Vision Transformer backbone, see config in `configs/Market1501/bagtricks_vit.yml`
- Self-Distillation with EMA update
- Gradient Clip
faster dataloader with pre-fetch and cuda stream (#456) Summary: add a background thread to create a generator with pre-fetch, and create a new cuda stream to copy tensor from cpu to gpu in parallel. Reviewed by: l1aoxingyu
2021-04-12 15:03:35 +08:00
Bugfix for cls_layer In `any_softmax`, all operations are in-place, so pass into the `logits.clone()` to prevent outside logits changed.
2021-05-31 17:32:24 +08:00
#### Improvements
faster dataloader with pre-fetch and cuda stream (#456) Summary: add a background thread to create a generator with pre-fetch, and create a new cuda stream to copy tensor from cpu to gpu in parallel. Reviewed by: l1aoxingyu
2021-04-12 15:03:35 +08:00
- Faster dataloader with pre-fetch thread and cuda stream
Bugfix for cls_layer In `any_softmax`, all operations are in-place, so pass into the `logits.clone()` to prevent outside logits changed.
2021-05-31 17:32:24 +08:00
- Optimize DDP training speed by removing `find_unused_parameters` in DDP
faster dataloader with pre-fetch and cuda stream (#456) Summary: add a background thread to create a generator with pre-fetch, and create a new cuda stream to copy tensor from cpu to gpu in parallel. Reviewed by: l1aoxingyu
2021-04-12 15:03:35 +08:00
update fastreid v1.2 readme and changelog
2021-04-06 20:09:13 +08:00
### v1.2 (06/04/2021)
#### New Features
- Multiple machine training support
- [RepVGG](https://github.com/DingXiaoH/RepVGG) backbone
- [Partial FC](projects/FastFace)
#### Improvements
- Torch2trt pipeline
- Decouple linear transforms and softmax
- config decorator
feat: support multi-teacher kd Summary: support multi-teacher kd with logits and overhaul distillation
2021-01-29 17:25:31 +08:00
### v1.1 (29/01/2021)
#### New Features
update readme about FastRT
2021-01-30 16:11:08 +08:00
- NAIC20(reid track) [1-st solution](projects/NAIC20)
feat: support multi-teacher kd Summary: support multi-teacher kd with logits and overhaul distillation
2021-01-29 17:25:31 +08:00
- Multi-teacher Knowledge Distillation
update readme about FastRT
2021-01-30 16:11:08 +08:00
- TRT network definition APIs in [FastRT](projects/FastRT)
feat: support multi-teacher kd Summary: support multi-teacher kd with logits and overhaul distillation
2021-01-29 17:25:31 +08:00
#### Bug Fixes
#### Improvements