PocketFlow

PocketFlow copied to clipboard

**Describe the bug** A clear and concise description of what the bug is. **To Reproduce** The command to reproduce the issue ``` ./scripts/run_local.sh nets/mobilenet_at_ilsvrc12_run.py --mobilenet_version 2 --ws_prune_ratio_prtl uniform --ws_prune_ratio 0.75...

yuanyuanli85

If still exec the compress process when inference?

21

I read the code in `export_pb_tflite_models.py`. The following code compress the model, but I think these operation will save in the `pb` file, and the new output `x` rely on...

as754770178

Accuracy drooped after using export utility

16

I am trying to calculate manually the accuracy of a model with uniform-tf learner. After calling export_quant_tflite_model, a Pb file was generated, python ./tools/conversion/export_quant_tflite_model.py --model_dir ./models_uqtf_eval I am trying to...

besherh

Hyper-parameters about learners on Resnet-50

26

I can't get the model converge on every learner , either imagenet or cifar10。Any suggestion about the hyper-parameters?Thanks

hzhyhx1117

Can you give the channel pruning performance on mobilenet version2? There is only mobilenet version1 relased

3

Please make sure that this is a documentation issue. **System information** - PocketFlow version: - Doc Link: **Describe the documentation issue** **We welcome contributions by users. Will you be able...

Anonymity2022

restore model at the end of `__train_pruned_model`

1

I want to know why restore model at the end of `__train_pruned_model`, because no operation after this. ``` def __calc_grads_pruned(self, grads_origin): ...... if self.__is_primary_worker(): with self.pruner.model.g.as_default(): self.pruner.saver = tf.train.Saver() #...

as754770178

can't runing multi-GPU

7

when I use one GPU and it finished without any problem , but when using multi-GPU, it hung when runing bcast operation, I don't know how to solve it. code:...

liaocz

ChannelPrunedLearner don't compress the model

7

I prune resnet_20 at cifar_10 by ChannelPrunedLearner, but I reader `original_model.ckpt`, `pruned_model.ckpt` and `best_model.ckpt`, the size of weight is same. `origin val` is the size of weight in `original_model.ckpt`, `pruned...

as754770178

Hi @psyyz10, It's really hard to train the ddpg-agent and I have tried many combinations of hyper-parameters but only get 66.38% top1-acc on MobileNetV1 with 50% FLOPs. I will appreciate...

GoldenSpark

ChannelPrunedLearner don't exec

13

I compress the model by ChannelPrunedLearner, when executed to `self.sess_train.run(self.train_op)` in `__train_pruned_model(self, finetune=False)` fun, the program don't continue execution。 I don't know This is because my machine is too card...

as754770178