facebookresearch/ConvNeXt-V2

Issues

• Weights trained using pre-training script produce NaN values when running fine-tuning script

  #64 opened 8 months ago by RienkF
  3

• Why not provide 22k-supervised finetuning model??? I am really shocked by that every available ConvNeXt-V2 pre-training weights has been finetuned on imagenet-1k. Please make 22k-supervised ConvNeXt-V2 open just like ConvNeXt-V1 !!!!!! 🙏🙏🙏

  #72 opened 6 months ago by yan-hao-tian
  2

• The reconstruction images have a lot of noise

  #48 opened a year ago by powermano
  6

• Pre-trained weights incompatible with backbone

  #65 opened 7 months ago by LucFrachon
  2

• ConvNeXt or ConvNeXt-V2 for drone target view geo-localization problem ?

  #77 opened 2 months ago by blackpearl1022
  0

• The model used contains deepcopy, and a TypeError: cannot pickle 'MinkowskiConvolutionFunction' object error was reported when using this project

  #27 opened 2 years ago by Joker9194
  1

• ImageNet-1K pre-trained weights ConvNeXt V2 supervised

  #61 opened a year ago by OmarEl-Khashab
  1

• Onnx Export

  #68 opened 7 months ago by sulaimanvesal
  1

• GRN in paper uses x_i / sum x_j whereas code uses x_i / mean x_j

  #76 opened 3 months ago by sebLopezCot
  0

• Colab Implementation

  #75 opened 3 months ago by emrekeles-arch
  0

• Cannot install MinkowskiEngine with provided instructions

  #63 opened 8 months ago by gholste
  2

• downsample_layer model order

  #74 opened 3 months ago by deeperlearner
  0

• can grn add to resnet

  #73 opened 3 months ago by dianjingshilezhi
  0

• cannot import name 'MinkowskiDepthwiseConvolution' from 'MinkowskiEngine'.

  #35 opened 2 years ago by jayepep
  5

• ImageNet 22k(21k) Traning loss at the end of training

  #71 opened 6 months ago by TheDarkKnight-21th
  0

• Finetuning with limited Labels ?

  #70 opened 6 months ago by rohit-gupta
  0

• [Question] Cosine Similarity

  #69 opened 7 months ago by dhkim0225
  0

• Reason for double mask in Jax implementation?

  #45 opened 7 months ago by sebastian-sz
  2

• deployment issue in trt fp16

  #66 opened 7 months ago by sipie800
  0

• could you share the dense masked conv based sparse encoding ?

  #60 opened a year ago by hengshan123
  2

• Unable to update submodule (publickey error)

  #50 opened a year ago by JonathanBhimaniBurrows
  4

• Could you release a pre-trained model with a decoder?

  #39 opened 2 years ago by Totorol
  2

• GRN can be used on any convent with FCMAE ？anyone tried this ?

  #62 opened 9 months ago by Vincent630
  0

• Will the methods work on other CNNs?

  #23 opened 2 years ago by dashidhy
  2

• visualization and feature consine distance

  #51 opened a year ago by Mogul5306
  2

• Adapting ConvNetV2 for Time Series: Inconsistencies between Pre-training Visuals and Fine-Tuning Performance

  #59 opened a year ago by heng3366
  1

• Make image and patch sizes dynamic

  #57 opened a year ago by lrettenberger
  0

• About the size of the patches

  #30 opened 2 years ago by colindecourt
  8

• Could you please provide a demo code for how to visualize the effect of mask decoder, such as the following figure?

  #42 opened a year ago by littletomatodonkey
  2

• On masking input images

  #56 opened a year ago by xiaohao-lin1
  1

• unable to install apex

  #52 opened a year ago by wgx8130
  2

• How can I reproduce the supervised training results?

  #43 opened a year ago by rayleizhu
  1

• segmentation

  #49 opened a year ago by mingming0611yes
  0

• Question on decoder resolution

  #25 opened 2 years ago by cmsflash
  1

• The official pre-training weights does not match the parameters of the model

  #47 opened a year ago by morningbearscau
  1

• What is the loss of the pretrain model in fcmae stage?

  #44 opened a year ago by xcyuyuyu
  0

• Problem with ConvNeXt V2-B 384x384 weights

  #24 opened a year ago by kaitolucifer
  4

• How about the log files of pretraining?

  #37 opened 2 years ago by ayumiymk
  1

• custom LayerNorm

  #41 opened a year ago by starsky68
  0

• No key named 'lr_scale' as input parameter to optimizer.AdamW

  #40 opened 2 years ago by DanielGeorgeMathew
  0

• Training Setting for ConvNeXt V1-A Supervised (75.7%) in Table 14

  #38 opened 2 years ago by techmonsterwang
  0

• RuntimeError: Error(s) in loading state_dict for ConvNeXtV2:

  #33 opened 2 years ago by weihongwei-zg
  5

• MinkowskiLinear is missing from MinkowskiEngine, MinkowskiGELU code files

  #36 opened 2 years ago by 464hee
  0

• How to get the predicted results of the test set?

  #34 opened 2 years ago by weihongwei-zg
  0

• Do you consider comparing the scores of Cityscapes in the future to facilitate the data disclosure

  #32 opened 2 years ago by 464hee
  0

• Question regarding sparse convolution dimension

  #31 opened 2 years ago by orsic
  0

• Reproduction of segmentation networks

  #29 opened 2 years ago by 464hee
  0

• LayerNorm vs torch.nn.LayerNorm

  #28 opened 2 years ago by Michael-H777
  1

• self-supervised pre-trained weights miss Parameters

  #26 opened 2 years ago by youzhaxiaomayi
  0

• Is the gamma and beta initial value wrong?Does zero work？

  #22 opened 2 years ago by Elaineok
  0