Skip to content
ModelListedExternal
Download

Fashion Clip Onnx

ONNX export of patrickjohncyh/fashion-clip split into separate vision and text encoders. Hosted for use by @looped-app/garment-classifier as a third backbone option alongside the standard OpenAI CLIP variants from...

Sourcehuggingface.co/Frapic/fashion-clip-onnx/tree/12eb79267363fd03b8983a25903cd9097b1ec76c
KindOtherBase modelpatrickjohncyh/fashion-clipVersionRevision 12eb792

Model source

Kind
Other
File size
243–336 MB
Base model
patrickjohncyh/fashion-clip
Version
vRevision 12eb792
Updated
Sep 17, 2026
Downloads
14
Source
Hugging Face

Source description

FashionCLIP ONNX (split vision + text encoders)

ONNX export of patrickjohncyh/fashion-clip split into separate vision and text encoders. Hosted for use by @looped-app/garment-classifier as a third backbone option alongside the standard OpenAI CLIP variants from Xenova.

Files

FileSizePurpose
vision_model.onnx~353 MBImage encoder. Input pixel_values: float32 [B, 3, 224, 224] → output image_embeds: float32 [B, 512] (not L2-normalized).
text_model.onnx~255 MBText encoder. Input input_ids: int64 [B, sequence_length] → output text_embeds: float32 [B, 512] (not L2-normalized).
tokenizer.json3.6 MBOpenAI CLIP BPE tokenizer (49408 vocab + SOT 49406 / EOT 49407). Compatible with @huggingface/tokenizers / @huggingface/transformers AutoTokenizer.
config.json, vocab.json, merges.txt, special_tokens_map.json, tokenizer_config.json, preprocessor_config.jsonmiscStandard transformers metadata.

Caveats

  • Text encoder batch size must be 1 at runtime due to a torch 2.12 export-time dynamic-axis baking quirk. Run prompts sequentially (one forward per prompt). Our garment-classifier package does this naturally in its prompt-embedding precompute loop.
  • Outputs are NOT L2-normalized. Normalize in the consumer code before cosine similarity (CLIP convention).
  • Image preprocessing: requires standard CLIP-style normalization: resize to 224×224 + mean=[0.48145466, 0.4578275, 0.40821073] / std=[0.26862954, 0.26130258, 0.27577711] applied per-channel.

Provenance

  • Base model: patrickjohncyh/fashion-clip (282 likes, 3M downloads on HF as of 2026-05-24)
  • Export: torch 2.12.0 + onnx 1.21.0 via torch.onnx.export on separate CLIPVisionModelWithProjection / CLIPTextModelWithProjection wrappers.
  • Verification: real-image cosine vs PyTorch reference. Output is functional (no NaN, sensible top-1 predictions on apparel imagery).
  • Export script: /tmp/fashionclip-split-export.py in the original session (orchestrator-direct).
  • License: MIT (inherits from patrickjohncyh/fashion-clip).

Sources

1 source
Hugging FacePrimaryVersion 12eb79267363fd03b8983a25903cd9097b1ec76c
huggingface.co/Frapic/fashion-clip-onnx/tree/12eb79267363fd03b8983a25903cd9097b1ec76c

Verified Sep 17

Model artifacts

2 artifacts

text_model.onnx

onnx · 243 MB · SHA-256 4051b35aa2a5…06bf · Hugging Face

Download

vision_model.onnx

onnx · 336 MB · SHA-256 0f87d09467f9…1a85 · Hugging Face

Download

Source excerpts

2 excerpts
ReadmeSep 17huggingface.co/Frapic/fashion-clip-onnx

--- license: mit tags: - clip - onnx - fashion - zero-shot-image-classification language: - en pipeline_tag: zero-shot-image-classification base_model: patrickjohncyh/fashion-clip --- # FashionCLIP ONNX (split vision + text encoders) ONNX export of [`patrickjohncyh/fashion-clip`](https://huggingface.co/patrickjohncyh/fashion-clip) split into separate vision and text encoders. Hosted for use by [`@looped-app/garment-classifier`](https://github.com/looped-app/packages/tree/main/garment-classifier) as a third backbone option alongside the standard OpenAI CLIP variants from Xenova. ## Files | File | Size | Purpose | |---|---|---| | `vision_model.onnx` | ~353 MB | Image encoder. Input `pixel_values: float32 [B, 3, 224, 224]` → output `image_embeds: float32 [B, 512]` (not L2-normalized). | | `text_model.onnx` | ~255 MB | Text encoder. Input `input_ids: int64 [B, sequence_length]` → output `text_embeds: float32 [B, 512]` (not L2-normalized). | | `tokenizer.json` | 3.6 MB | OpenAI CLIP BPE tokenizer (49408 vocab + SOT 49406 / EOT 49407). Compatible with `@huggingface/tokenizers` / `@huggingface/transformers` `AutoTokenizer`. | | `config.json`, `vocab.json`, `merges.txt`, `special_tokens_map.json`, `tokenizer_config.json`, `preprocessor_config.json` | misc | Standard transformers metadata. | ## Caveats - **Text encoder batch size must be 1** at runtime due to a torch 2.12 export-time dynamic-axis baking quirk. Run prompts sequentially (one forward per prompt). Our `garment-classifier` package does this naturally in its prompt-embedding precompute loop. - **Outputs are NOT L2-normalized**. Normalize in the consumer code before cosine similarity (CLIP convention). - **Image preprocessing**: requires standard CLIP-style normalization: resize to 224×224 + mean=[0.48145466, 0.4578275, 0.40821073] / std=[0.26862954, 0.26130258, 0.27577711] applied per-channel. ## Provenance - Base model: `patrickjohncyh/fashion-clip` (282 likes, 3M downloads on HF as of 2026-05-24) - Export: torch 2.12.0 + onnx 1.21.0 via `torch.onnx.export` on separate `CLIPVisionModelWithProjection` / `CLIPTextModelWithProjection` wrappers. - Verification: real-image cosine vs PyTorch reference. Output is functional (no NaN, sensible top-1 predictions on apparel imagery). - Export script: `/tmp/fashionclip-split-export.py` in the original session (orchestrator-direct). - License: MIT (inh...

NoteSep 17huggingface.co/Frapic/fashion-clip-onnx

Source context: 14 downloads · 0 likes · Pipeline zero-shot-image-classification · Repo Frapic/fashion-clip-onnx