{"id":16271684,"url":"https://github.com/jianzhnie/multimodaltransformers","last_synced_at":"2025-09-15T09:05:33.897Z","repository":{"id":208367001,"uuid":"714955930","full_name":"jianzhnie/MultimodalTransformers","owner":"jianzhnie","description":"lmmtoolkit is a toolkit for Multi-Modal Learning","archived":false,"fork":false,"pushed_at":"2023-11-21T11:40:52.000Z","size":23,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-02-14T11:53:04.654Z","etag":null,"topics":["image-text","multi-modal-learning","text-image","text-to-video"],"latest_commit_sha":null,"homepage":"https://jianzhnie.github.io/llmtech/","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jianzhnie.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2023-11-06T07:37:18.000Z","updated_at":"2024-10-15T07:38:20.000Z","dependencies_parsed_at":null,"dependency_job_id":"6faa9771-1ce6-4f85-b900-6680af3ec5e8","html_url":"https://github.com/jianzhnie/MultimodalTransformers","commit_stats":{"total_commits":13,"total_committers":1,"mean_commits":13.0,"dds":0.0,"last_synced_commit":"47033c72f24415ce4498b3656feaca5e3f504eb1"},"previous_names":["jianzhnie/multimodaltransformers"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jianzhnie%2FMultimodalTransformers","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jianzhnie%2FMultimodalTransformers/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jianzhnie%2FMultimodalTransformers/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jianzhnie%2FMultimodalTransformers/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jianzhnie","download_url":"https://codeload.github.com/jianzhnie/MultimodalTransformers/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247871308,"owners_count":21010020,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["image-text","multi-modal-learning","text-image","text-to-video"],"created_at":"2024-10-10T18:14:24.871Z","updated_at":"2025-04-08T15:34:42.588Z","avatar_url":"https://github.com/jianzhnie.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# MultimodalTransformers\n\n## CLIP\n\nThis is a simple implementation of **Natural Language-based Image Search** inspired by the [CLIP](https://openai.com/blog/clip/) approach as proposed by the paper [**Learning Transferable Visual Models From Natural Language Supervision**](https://arxiv.org/abs/2103.00020) by OpenAI in [**PyTorch Lightning**](https://www.pytorchlightning.ai/). We also use [**Weights \u0026 Biases**](wandb.ai) for experiment tracking, visualizing results, comparing performance of different backbone models, hyperparameter optimization and to ensure reproducibility.\n\n```shell\npython examples/train_clip.py\n```\n\nThis command will initialize a CLIP model with a **ResNet50** image backbone and a **distilbert-base-uncased** text backbone.\n\n## 📚 CLIP: Connecting Text and Images\n\nCLIP (Contrastive Language–Image Pre-training) builds on a large body of work on zero-shot transfer, natural language supervision, and multimodal learning. CLIP pre-trains an image encoder and a text encoder to predict which images were paired with which texts in our dataset. This behavior turns CLIP into a zero-shot classifier. All of a dataset’s classes are converted into captions such as “a photo of a dog” followed by predicting the class of the caption in which CLIP estimates best pairs with a given image.\n\nYou can read more about CLIP [here](https://openai.com/blog/clip/) and [here](https://arxiv.org/abs/2103.00020)\n\n## 💿 Dataset\n\nThis implementation of CLIP supports training on two datasets [Flickr8k](https://forms.illinois.edu/sec/1713398) which contains ~8K images with 5 captions for each image and [Flickr30k](https://aclanthology.org/Q14-1006/) which contains ~30K images with corresponding captions.\n\n## 🤖 Model\n\nA CLIP model uses a text encoder and an image encoder. This repostiry supports pulling image models from [PyTorch Image Models](https://github.com/rwightman/pytorch-image-models) and transformer models from [huggingface transformers](https://github.com/huggingface/transformers).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjianzhnie%2Fmultimodaltransformers","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjianzhnie%2Fmultimodaltransformers","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjianzhnie%2Fmultimodaltransformers/lists"}