mmrazor icon indicating copy to clipboard operation
mmrazor copied to clipboard

OpenMMLab Model Compression Toolbox and Benchmark.

Results 184 mmrazor issues
Sort by recently updated
recently updated
newest added

## Motivation Add `DerivedMutable` and refactor dynamic ops. ## Modification 1. Add `DerivedMutable` 2. Refactor dynamic ops 3. Add `BigNASMobileNet` as an example for hand-wirtten model without tracer ## BC-breaking...

## Motivation Add Decoupled KD Loss (DKDLoss). Add [DKDLoss](https://arxiv.org/pdf/2203.08679.pdf) config, readme and UT. Revise `test_distillation_losses` to mock data_samples for [DKDLoss](https://arxiv.org/pdf/2203.08679.pdf) and [WSLDLoss](https://arxiv.org/abs/2102.00650).

### Describe the question you meet I successfully ran through Point_rend with CWD, but at the beginning of the training, all the losses were Nan. What is the reason Your...

Dose mmrazor support mmaction2(posec3d)?

## Motivation Add BigNAS algorithm. ## Modification 1. Includes modifications mentioned in pr [open-mmlab/mmrazor#215](https://github.com/open-mmlab/mmrazor/pull/215) 2. Pull out the grouping logic of mutator as mixin 3. Add AttentiveNAS and BigNAS search...

## Motivation 1. Support unroll method in the search phase of darts. 2. Support DDP training. 3. Add docstring for darts algorithm. Results: ![image](https://user-images.githubusercontent.com/29230784/182513710-44d42bd3-43e6-4c36-abed-4d53295e96ba.png) Docstring coverage: ![image](https://user-images.githubusercontent.com/29230784/182500244-62d9f530-ca89-4d0b-874f-062f9fd290a0.png) TestCase coverage: ![image](https://user-images.githubusercontent.com/29230784/182639649-b05ee20e-18cb-483d-b7e1-68ecb504709b.png)

## Motivation - [x] Move `build_arch_param` function from `DiffMutableModule` (including `DiffMutableOP`, `DiffChoiceRoute` and `GumbelChoiceRoute`) to `DiffModuleMutator`. - [x] Rename cls.data to cls.structures. For more information see: https://github.com/open-mmlab/mmclassification/pull/941 ## Modification -...

1. Fix bug in test_convmodule_connector function.

## Motivation Add graph-related modules which are essential for pruning. ## Modification add BaseNode/BaseGraph, ModuleNode/ModuleGraph, and related test.

When is the quantitative algorithm coming out? I noticed SenseTime's MQBench, is there any difference and connection between the coming quantization algorithm and MQBench?