# OpenCLIP

## Docs

- [Custom Models](https://mintlify.wiki/mlfoundations/open_clip/advanced/custom-models.md): Create custom CLIP model architectures with custom vision and text encoders
- [Gradient Accumulation](https://mintlify.wiki/mlfoundations/open_clip/advanced/gradient-accumulation.md): Simulate larger batch sizes with gradient accumulation to overcome memory limitations
- [Int8 Support](https://mintlify.wiki/mlfoundations/open_clip/advanced/int8-support.md): Train and run inference with 8-bit quantization for improved speed and reduced memory usage
- [Model Distillation](https://mintlify.wiki/mlfoundations/open_clip/advanced/model-distillation.md): Distill knowledge from larger CLIP models into smaller, more efficient models
- [Push to Hub](https://mintlify.wiki/mlfoundations/open_clip/advanced/push-to-hub.md): Upload trained CLIP models to Hugging Face Hub for sharing and deployment
- [Remote Training](https://mintlify.wiki/mlfoundations/open_clip/advanced/remote-training.md): Train models with automatic cloud storage backup and resume from remote checkpoints
- [AugmentationCfg](https://mintlify.wiki/mlfoundations/open_clip/api/augmentation-cfg.md): Configuration dataclass for image augmentation during training
- [build_zero_shot_classifier](https://mintlify.wiki/mlfoundations/open_clip/api/build-zero-shot-classifier.md): Build zero-shot classifier weights from class names and text templates
- [ClipLoss](https://mintlify.wiki/mlfoundations/open_clip/api/clip-loss.md): Contrastive loss function for CLIP model training with distributed support
- [CLIP](https://mintlify.wiki/mlfoundations/open_clip/api/clip-model.md): Core CLIP model implementation for contrastive learning between images and text
- [CoCaLoss](https://mintlify.wiki/mlfoundations/open_clip/api/coca-loss.md): Combined contrastive and captioning loss for training CoCa (Contrastive Captioner) models
- [CoCa](https://mintlify.wiki/mlfoundations/open_clip/api/coca-model.md): Contrastive Captioner model combining CLIP-style contrastive learning with autoregressive text generation
- [create_model](https://mintlify.wiki/mlfoundations/open_clip/api/create-model.md): Creates and configures a contrastive vision-language model
- [create_model_and_transforms](https://mintlify.wiki/mlfoundations/open_clip/api/create-model-and-transforms.md): Creates a model with preprocessing transforms for training and validation
- [create_model_from_pretrained](https://mintlify.wiki/mlfoundations/open_clip/api/create-model-from-pretrained.md): Creates a model from pretrained weights with optional preprocessing transform
- [CustomTextCLIP](https://mintlify.wiki/mlfoundations/open_clip/api/custom-text-clip.md): CLIP variant with a separately constructed text encoder tower for maximum flexibility
- [decode](https://mintlify.wiki/mlfoundations/open_clip/api/decode.md): Convert token IDs back to text strings
- [get_pretrained_cfg](https://mintlify.wiki/mlfoundations/open_clip/api/get-pretrained-cfg.md): Get configuration for a specific pretrained model and tag
- [get_tokenizer](https://mintlify.wiki/mlfoundations/open_clip/api/get-tokenizer.md): Get the appropriate tokenizer for a CLIP model
- [image_transform_v2](https://mintlify.wiki/mlfoundations/open_clip/api/image-transform.md): Create image preprocessing transforms with configuration objects
- [list_models](https://mintlify.wiki/mlfoundations/open_clip/api/list-models.md): Enumerate available model architectures
- [list_pretrained](https://mintlify.wiki/mlfoundations/open_clip/api/list-pretrained.md): List all available pretrained model/weight combinations
- [load_checkpoint](https://mintlify.wiki/mlfoundations/open_clip/api/load-checkpoint.md): Load a checkpoint into an existing model
- [PreprocessCfg](https://mintlify.wiki/mlfoundations/open_clip/api/preprocess-cfg.md): Configuration dataclass for image preprocessing
- [SigLipLoss](https://mintlify.wiki/mlfoundations/open_clip/api/siglip-loss.md): Sigmoid-based contrastive loss for language-image pre-training with improved efficiency and performance
- [tokenize](https://mintlify.wiki/mlfoundations/open_clip/api/tokenize.md): Tokenize text strings for CLIP models
- [Zero-Shot Metadata](https://mintlify.wiki/mlfoundations/open_clip/api/zero-shot-metadata.md): Pre-defined templates and class names for zero-shot classification
- [CLIP Overview](https://mintlify.wiki/mlfoundations/open_clip/concepts/clip-overview.md): Understanding the CLIP architecture and how it learns visual-semantic representations
- [Contrastive Learning](https://mintlify.wiki/mlfoundations/open_clip/concepts/contrastive-learning.md): Understanding how CLIP learns visual-semantic embeddings through contrastive loss
- [Zero-Shot Classification](https://mintlify.wiki/mlfoundations/open_clip/concepts/zero-shot-classification.md): How CLIP performs classification on unseen categories without fine-tuning
- [Benchmark Results](https://mintlify.wiki/mlfoundations/open_clip/evaluation/benchmarks.md): Comprehensive benchmark results for OpenCLIP models across 38 evaluation datasets
- [Evaluation Metrics](https://mintlify.wiki/mlfoundations/open_clip/evaluation/metrics.md): Understanding the metrics used to evaluate OpenCLIP models
- [Zero-Shot Evaluation](https://mintlify.wiki/mlfoundations/open_clip/evaluation/zero-shot.md): Learn how to evaluate OpenCLIP models on zero-shot tasks during and after training
- [Installation](https://mintlify.wiki/mlfoundations/open_clip/installation.md): Install OpenCLIP for inference, training, or development
- [Introduction](https://mintlify.wiki/mlfoundations/open_clip/introduction.md): OpenCLIP is an open-source implementation of OpenAI's CLIP (Contrastive Language-Image Pre-training) for training and using vision-language models.
- [Quickstart](https://mintlify.wiki/mlfoundations/open_clip/quickstart.md): Get started with OpenCLIP in minutes - load a model and perform zero-shot image classification
- [CoCa Training](https://mintlify.wiki/mlfoundations/open_clip/training/coca.md): Train CoCa (Contrastive Captioner) models for image captioning and contrastive learning
- [Training Configuration](https://mintlify.wiki/mlfoundations/open_clip/training/configuration.md): Complete reference for all OpenCLIP training parameters and hyperparameters
- [Data Preparation](https://mintlify.wiki/mlfoundations/open_clip/training/data-preparation.md): Prepare training data in CSV and WebDataset formats for CLIP training
- [Distributed Training](https://mintlify.wiki/mlfoundations/open_clip/training/distributed-training.md): Advanced techniques for efficient large-scale CLIP training
- [Fine-tuning](https://mintlify.wiki/mlfoundations/open_clip/training/fine-tuning.md): Fine-tune pretrained CLIP models for improved performance on specific tasks
- [Multi-Node Training](https://mintlify.wiki/mlfoundations/open_clip/training/multi-node.md): Scale CLIP training across multiple machines with torchrun and SLURM
- [Training Overview](https://mintlify.wiki/mlfoundations/open_clip/training/overview.md): Overview of training CLIP models with OpenCLIP
- [Single-Node Training](https://mintlify.wiki/mlfoundations/open_clip/training/single-node.md): Train CLIP models on a single machine with multiple GPUs
- [Image Preprocessing](https://mintlify.wiki/mlfoundations/open_clip/usage/image-preprocessing.md): Configure image transformations, normalization, augmentation, and preprocessing for OpenCLIP models
- [Inference](https://mintlify.wiki/mlfoundations/open_clip/usage/inference.md): Complete guide to running inference with OpenCLIP models including image/text encoding, similarity computation, and batch processing
- [Loading Models](https://mintlify.wiki/mlfoundations/open_clip/usage/loading-models.md): Learn how to load OpenCLIP models from various sources including pretrained weights, HuggingFace Hub, and local directories
- [Pretrained Models](https://mintlify.wiki/mlfoundations/open_clip/usage/pretrained-models.md): Complete catalog of pretrained OpenCLIP models with performance metrics, training details, and usage examples
- [Tokenization](https://mintlify.wiki/mlfoundations/open_clip/usage/tokenization.md): Text tokenization for OpenCLIP models including SimpleTokenizer, HFTokenizer, and SigLipTokenizer with context length handling

## OpenAPI Specs

- [openapi](https://mintlify.wiki/mlfoundations/open_clip/api-reference/openapi.json)