Dear participants,
Please register any use of public pre-trained models by mentioning it in this thread to keep equity among participants.
The deadline to register pre-trained model is Monday 22nd March (few days before private test phase).
Thanks for your understanding,
The EmoRecCom organizers.
I use a pre-trained model on ImageNet and Wiki.en
Posted by: DETA @ March 22, 2021, 10:40 a.m.I have utilized the pretrained model of bert from the HuggingFace Website. And the pretrained resnet50 model is downloaded from the Pytorch Website.
Posted by: NELSLIP @ March 22, 2021, 11:31 a.m.I have used the pretrilned ResNet50 from TensorFlow
Posted by: datqngo @ March 22, 2021, 1:34 p.m.Currently, I have used some pre-trained models on ImageNet like ResNet, VGG, Xception, ... for the visual approach. Besides, I have used glove embedding for the text approach. Can I try other approaches after registering the pre-trained models?
Posted by: phamnhattruong.skyo @ March 22, 2021, 1:57 p.m.I used pretrained Resnet model on Imagenet and GloVe word embeddings, now try to use bert pretrained model.
Posted by: HU_XIAORAN @ March 22, 2021, 2:30 p.m.My pretrained models:
- ResNet50/101/152/50V2/101V2/152V2, DenseNet121/169/201, EffieciennetB0-B7, ResNeXt50/101
- glove.840B.300d, wiki.en.vec, crawl-300d-2M.vec, wiki-news-300d-1M.vec
- bert base/large cased/uncased, distilbert base/large cased/uncased, roberta base/large, Albert, electra, xlm-roberta, bertweet
I used pretrained Efficientnet model on Imagenet and Noisy student and BERT/RoBERTa embeddings.
Posted by: Samson @ March 22, 2021, 5:01 p.m.bert-large-uncased, roberta-large, ernie-2.0-large-en resnet50
Posted by: DeepBlueAI @ March 23, 2021, 2:07 a.m.bert-large-uncased, roberta-large, ernie-2.0-large-en resnet50
Posted by: DeepBlueAI @ March 23, 2021, 2:07 a.m.unet