CoCa - Pytorch
Implementation of CoCa, Contrastive Captioners are Image-Text Foundation Models, in Pytorch. They were able to elegantly fit in contrastive learning to a conventional encoder / decoder (image to text) transformer, achieving SOTA 91.0% top-1 accuracy on ImageNet with a finetuned encoder.
This repository also chooses to adopt the specific transformer architecture from PaLM, for both the unimodal and multimodal transformers as well as the cross attention blocks (parallel SwiGLU feedforwards)
Install
$ pip install coca-pytorch
Usage
First install the vit-pytorch for the image encoder, which needs to be pretrained
$ pip install vit-pytorch
Then
import torch
# import vision transformer
from vit_pytorch import ViT
from vit_pytorch.extractor import Extractor
vit = ViT(
image_size = 256,
patch_size = 32,
num_classes = 1000,
dim = 1024,
depth = 6,
heads = 16,
mlp_dim = 2048
)
# do your vision transformer training
vit = Extractor(vit, return_embeddings_only = True)
# extractor will enable it so the vision transformer returns its embeddings
# import CoCa and instantiate it
from coca_pytorch.coca_pytorch import CoCa
coca = CoCa(
dim = 512, # model dimension
img_encoder = vit, # vision transformer - image encoder, returning image embeddings as (batch, seq, dim)
image_dim = 1024, # image embedding dimension, if not the same as model dimensions
num_tokens = 20000, # number of text tokens
unimodal_depth = 6, # depth of the unimodal transformer
multimodal_depth = 6, # depth of the multimodal transformer
dim_head = 64, # dimension per attention head
heads = 8, # number of attention heads
caption_loss_weight = 1., # weight on the autoregressive caption loss
contrastive_loss_weight = 1., # weight on the contrastive loss between image and text CLS embeddings
).cuda()
# mock text and images
text = torch.randint(0, 20000, (4, 512)).cuda()
images = torch.randn(4, 3, 256, 256).cuda()
# train by giving CoCa your text and images with `return_loss = True`
loss = coca(
text = text,
images = images,
return_loss = True # set this to True to get the full caption + contrastive loss
)
loss.backward()
# do the above for as much text and images...
# then you can get the caption logits as so
logits = coca(
text = text,
images = images
) # (4, 512, 20000)
# and the CLIP-like text and image embeddings as
text_embeds, image_embeds = coca(
text = text,
images = images,
return_embeddings = True
) # (4, 512), (4, 512)
Citations
@inproceedings{Yu2022CoCaCC,
title = {CoCa: Contrastive Captioners are Image-Text Foundation Models},
author = {Jiahui Yu and Zirui Wang and Vijay Vasudevan and Legg Yeung and Mojtaba Seyedhosseini and Yonghui Wu},
year = {2022}
}
@inproceedings{Chowdhery2022PaLMSL,
title = {PaLM: Scaling Language Modeling with Pathways},
author = {Aakanksha Chowdhery and Sharan Narang and Jacob Devlin and Maarten Bosma and Gaurav Mishra and Adam Roberts and Paul Barham and Hyung Won Chung and Charles Sutton and Sebastian Gehrmann and Parker Schuh and Kensen Shi and Sasha Tsvyashchenko and Joshua Maynez and Abhishek Rao and Parker Barnes and Yi Tay and Noam M. Shazeer and Vinodkumar Prabhakaran and Emily Reif and Nan Du and Benton C. Hutchinson and Reiner Pope and James Bradbury and Jacob Austin and Michael Isard and Guy Gur-Ari and Pengcheng Yin and Toju Duke and Anselm Levskaya and Sanjay Ghemawat and Sunipa Dev and Henryk Michalewski and Xavier Garc{\'i}a and Vedant Misra and Kevin Robinson and Liam Fedus and Denny Zhou and Daphne Ippolito and David Luan and Hyeontaek Lim and Barret Zoph and Alexander Spiridonov and Ryan Sepassi and David Dohan and Shivani Agrawal and Mark Omernick and Andrew M. Dai and Thanumalayan Sankaranarayana Pillai and Marie Pellat and Aitor Lewkowycz and Erica Oliveira Moreira and Rewon Child and Oleksandr Polozov and Katherine Lee and Zongwei Zhou and Xuezhi Wang and Brennan Saeta and Mark Diaz and Orhan Firat and Michele Catasta and Jason Wei and Kathleen S. Meier-Hellstern and Douglas Eck and Jeff Dean and Slav Petrov and Noah Fiedel},
year = {2022}
}
137 Nov 09, 2022
196 Jan 04, 2023
880 Jan 07, 2023
17 Nov 05, 2022
244 Dec 27, 2022
38 Jan 02, 2023
5 Nov 03, 2022
1 Jan 07, 2022
138 Dec 17, 2022
35 Dec 21, 2022
49 Nov 28, 2022
1k Nov 29, 2022
695 Jan 05, 2023
5 Nov 25, 2022
88 May 15, 2022
3.7k Jan 05, 2023
451 Dec 26, 2022
9 Oct 15, 2022
30 Dec 28, 2022
41 Oct 29, 2022