cstr/clip-vit-large-patch14-336-GGUF

CLIP ViT-L/14@336 Vision Encoder (GGUF)

GGUF conversion of openai/clip-vit-large-patch14-336 for use with CrispEmbed.

• —Architecture: CLIP ViT-L/14 vision encoder (336px variant)
• —Parameters: 304M
• —Output: 768-dimensional L2-normalized embeddings (1024d internal, projected to 768d)
• —Input: 336x336 RGB image with CLIP normalization
• —Size: ~1.2 GB
• —Source: openai/clip-vit-large-patch14-336

Usage

# Embed a single image
crispembed -m clip-vit-large-patch14-336 --image photo.jpg

# Batch processing
crispembed -m clip-vit-large-patch14-336 --image-dir ./photos/ --output embeddings.bin

Higher input resolution (336x336 vs 224x224) captures finer spatial detail compared to the standard ViT-L/14.

Cross-modal pairing

Shares an embedding space with cstr/clip-text-large-GGUF for zero-shot image-text matching.

Notes

• —All output embeddings are L2-normalized.
• —This is a GGUF conversion; weights are numerically equivalent to the original HuggingFace model.

Provenance and EU AI Act Art. 53 note

• —Upstream model: openai/clip-vit-large-patch14-336 — published by openai.
• —Upstream licence: mit. This repository redistributes under the same terms; it grants no rights the upstream licence does not.
• —What was done here: format conversion and/or quantisation only (GGUF/GGML). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs.
• —Training data: documented — where it is documented at all — by the upstream provider; see the upstream model card. No training data was used, added or selected by this repository.
• —Provider status: under Regulation (EU) 2024/1689 the upstream authors remain the provider of this model. Converting the serialisation format does not make this repository the provider of a new general-purpose AI model, and no such claim is made. Questions about training content, copyright policy or model capability belong upstream.