facebookresearch kill-the-bits issues

facebookresearch / kill-the-bits

Code for: "And the bit goes down: Revisiting the quantization of neural networks"

Other

636 stars 124 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Inference bug.

#41 zhangkui669 closed 3 years ago
1
Address semisup R50 part of #38 and #39

#40 una-dinosauria closed 3 years ago
0
Reproducing memory of MaskRCNN and semisupervised ResNet-50

#39 una-dinosauria closed 3 years ago
4
About the results on semi-supervised ResNet-50

#38 una-dinosauria closed 3 years ago
2
How does dynamic X determine fixed Q(W)?

#37 bigdan12 closed 3 years ago
1
CentroidSGD in centroid_sgd.py

#36 cscftao closed 4 years ago
1
Why the activation is the input?

#35 yw155 closed 4 years ago
1
已压缩模型进行指定图片测试

#34 findlazygirl closed 4 years ago
2
Question about open code and different accuracy when infer by using quantized pth file

#33 Eunhui-Kim closed 4 years ago
1
RuntimeError for resnet50 small&large

#32 ghost closed 4 years ago
1
How to illustrate 'in-domain' and 'out-of-domain'?

#31 yw155 closed 4 years ago
2
cannot reproduce the results

#30 minhson closed 4 years ago
0
What should I do if I want to quantize Mask-RCNN by myself

#29 gitarya closed 4 years ago
1
I got a bad performance with "mask_r_cnn.pth" .How can I improve it, pelase ?

#28 xscjun closed 4 years ago
7
=

#27 EdwardLiuYu closed 4 years ago
2
About multi-gpu training(i can't reopen the former issue)

#26 Fight-hawk closed 4 years ago
0
About multi-gpu training

#25 Fight-hawk closed 4 years ago
6
Compression scheduler

#24 rafikg closed 4 years ago
1
processing time

#23 rafikg closed 4 years ago
1
small typo in section 4.2

#22 rafikg closed 4 years ago
1
footprint memory cost!

#21 rafikg closed 4 years ago
1
How to implement multi-gnu training?

#20 yw155 closed 4 years ago
3
How to use '_register_hooks'?

#19 yw155 closed 4 years ago
2
Why the second GPU has memory consumption?

#18 yw155 closed 4 years ago
4
The definition of 'quantizer' in quantize.py

#17 yw155 closed 4 years ago
1
The process of testing MaskRCNN was killed by system due to over RAM.

#16 mainguyenanhvu closed 5 years ago
3
I can't run python inference.py --model maskrcnn_resnet50_fpn --state-dict-compressed models/compressed/mask_r_cnn.pth --device cuda --data-path YOUR_COCO_PATH

#15 mainguyenanhvu closed 5 years ago
5
The variable 'in_activations_current' becomes 0 valued at the 9th layer of ResNet50

#14 yw155 closed 5 years ago
3
RuntimeError: svd_cuda: the updating process of SBDSDC did not converge

#13 alanMachineLeraning closed 5 years ago
3
train acc is about 60%, but test acc is only 8%

#12 xiaodao2049 closed 5 years ago
2
counting ops

#11 shenyann closed 5 years ago
1
RuntimeError: svd_cuda: the updating process of SBDSDC did not converge (error: 8)

#10 alanMachineLeraning closed 5 years ago
0
suspicious that the code of the compression model saved in the published source code is wrong.

#9 alanMachineLeraning closed 5 years ago
8
TypeError: 'module' object is not callable

#8 alanMachineLeraning closed 5 years ago
1
Empty cluster still remain

#7 tuanho27 closed 5 years ago
5
About small datasets training

#6 wnma3mz closed 5 years ago
8
The val set of imagenet is not stored by its "label folder"!

#5 talenz closed 5 years ago
5
Does this Quantization model accuracy is calculated on fixed point or floating point?

#4 tiru1930 closed 5 years ago
3
Fix version specification

#3 take-cheeze closed 5 years ago
2
how to quantize Mask R-CNN?

#2 92ypli closed 5 years ago
2
inference time？

#1 shiyongde closed 5 years ago
1