Upload 11 files

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+text.onnx.data filter=lfs diff=lfs merge=lfs -text
+visual.onnx.data filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,80 @@
+# open_clip_rs ONNX export of timm/MobileCLIP2-S2-OpenCLIP
+
+This model is an export of [timm/MobileCLIP2-S2-OpenCLIP](https://huggingface.co/timm/MobileCLIP2-S2-OpenCLIP). It is meant to be used with the [`open_clip_inference`](https://crates.io/crates/open_clip_inference) rust crate.
+
 ---
+---
+tags:
+- clip
+- mobileclip2
+library_name: open_clip
+pipeline_tag: zero-shot-image-classification
 license: apple-amlr
 ---
+# Model card for MobileCLIP2-S2-OpenCLIP
+These weights and model card are adapted from the original Apple model at https://huggingface.co/apple/MobileCLIP2-S2. This version uses canonical OpenCLIP configs and weight naming.
+
+MobileCLIP2 was introduced in [MobileCLIP2: Improving Multi-Modal Reinforced Training](http://arxiv.org/abs/2508.20691) (TMLR August 2025 <mark>Featured</mark>), by Fartash Faghri, Pavan Kumar Anasosalu Vasu, Cem Koc, Vaishaal Shankar, Alexander T Toshev, Oncel Tuzel, Hadi Pouransari.
+
+This repository contains the **MobileCLIP2-S2** checkpoint.
+
+### Highlights
+
+* `MobileCLIP2-S4` matches the accuracy of SigLIP-SO400M/14 with 2x fewer parameters and surpasses DFN ViT-L/14 at 2.5x lower latency measured on iPhone12 Pro Max.
+* `MobileCLIP-S3/S4` are our new architectures trained on MobileCLIP’s training dataset, DataCompDR-1B (dashed lines).
+* Our smallest variant `MobileCLIP-S0` obtains similar zero-shot performance as [OpenAI](https://arxiv.org/abs/2103.00020)'s ViT-B/16 model while being 4.8x faster and 2.8x smaller.
+* `MobileCLIP-S2` obtains better avg zero-shot performance than [SigLIP](https://arxiv.org/abs/2303.15343)'s ViT-B/16 model while being 2.3x faster and 2.1x smaller, and trained with 3x less seen samples.
+* `MobileCLIP-B (LT)` attains zero-shot ImageNet performance of **77.2%** which is significantly better than recent works like [DFN](https://arxiv.org/abs/2309.17425) and [SigLIP](https://arxiv.org/abs/2303.15343) with similar architectures or even [OpenAI's ViT-L/14@336](https://arxiv.org/abs/2103.00020).
+
+
+## Checkpoints and Results (Original Apple links)
+
+| Model                                                     | # Seen <BR>Samples (B) | # Params (M) <BR> (img + txt) | Latency (ms) <BR> (img + txt) | IN-1k Zero-Shot <BR> Top-1 Acc. (%) | Avg. Perf. (%) <BR> on 38 datasets |
+|:----------------------------------------------------------|:----------------------:|:-----------------------------:|:-----------------------------:|:-----------------------------------:|:----------------------------------:|
+| [MobileCLIP2-S0](https://hf.co/apple/MobileCLIP2-S0)      |           13           |          11.4 + 42.4          |           1.5 + 1.6           |               71.5                 |                59.7                 |
+| [MobileCLIP2-S2](https://hf.co/apple/MobileCLIP2-S2)      |           13           |          35.7 + 63.4          |           3.6 + 3.3           |               77.2                 |                64.1                 |
+| [MobileCLIP2-B](https://hf.co/apple/MobileCLIP2-B)        |           13           |          86.3 + 63.4          |          10.4 + 3.3           |               79.4                 |                65.8                 |
+| [MobileCLIP2-S3](https://hf.co/apple/MobileCLIP2-S3)      |           13           |         125.1 + 123.6         |           8.0 + 6.6           |               80.7                 |                66.8                 |
+| [MobileCLIP2-L/14](https://hf.co/apple/MobileCLIP2-L-14)  |           13           |         304.3 + 123.6         |          57.9 + 6.6           |               81.9                 |                67.8                 |
+| [MobileCLIP2-S4](https://hf.co/apple/MobileCLIP2-S4)      |           13           |         321.6 + 123.6         |          19.6 + 6.6           |               81.9                 |                67.5                 |
+| [MobileCLIP-S0](https://hf.co/apple/MobileCLIP-S0)        |           13           |          11.4 + 42.4          |           1.5 + 1.6           |               67.8                 |                58.1                 |
+| [MobileCLIP-S1](https://hf.co/apple/MobileCLIP-S1)        |           13           |          21.5 + 63.4          |           2.5 + 3.3           |               72.6                 |                61.3                 |
+| [MobileCLIP-S2](https://hf.co/apple/MobileCLIP-S2)        |           13           |          35.7 + 63.4          |           3.6 + 3.3           |               74.4                 |                63.7                 |
+| [MobileCLIP-B](https://hf.co/apple/MobileCLIP-B)          |           13           |          86.3 + 63.4          |          10.4 + 3.3           |               76.8                 |                65.2                 |
+| [MobileCLIP-B (LT)](https://hf.co/apple/MobileCLIP-B-LT)  |           36           |          86.3 + 63.4          |          10.4 + 3.3           |               77.2                 |                65.8                 |
+| [MobileCLIP-S3](https://hf.co/apple/MobileCLIP-S3)        |           13           |         125.1 + 123.6         |           8.0 + 6.6           |               78.3                 |                66.3                 |
+| [MobileCLIP-L/14](https://hf.co/apple/MobileCLIP-L-14)    |           13           |         304.3 + 123.6         |          57.9 + 6.6           |               79.5                 |                66.9                 |
+| [MobileCLIP-S4](https://hf.co/apple/MobileCLIP-S4)        |           13           |         321.6 + 123.6         |          19.6 + 6.6           |               79.4                 |                68.1                 |
+
+
+## How to Use
+
+```py
+import torch
+import open_clip
+from PIL import Image
+from urllib.request import urlopen
+from timm.utils import reparameterize_model
+
+model, _, preprocess = open_clip.create_model_and_transforms('MobileCLIP2-S0', pretrained='dfndr2b')
+model.eval()
+tokenizer = open_clip.get_tokenizer('MobileCLIP2-S0')
+
+# For inference/model exporting purposes, optionally reparameterize for better performance
+model = reparameterize_model(model)
+
+image = Image.open(urlopen(
+    'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
+))
+image = preprocess(image).unsqueeze(0)
+text = tokenizer(["a diagram", "a dog", "a cat", "a doughnut"])
+
+with torch.no_grad(), torch.amp.autocast(image.device.type):
+    image_features = model.encode_image(image)
+    text_features = model.encode_text(text)
+    image_features /= image_features.norm(dim=-1, keepdim=True)
+    text_features /= text_features.norm(dim=-1, keepdim=True)
+    text_probs = (100.0 * image_features @ text_features.T).softmax(dim=-1)
+
+print("Label probs:", text_probs)
+```

model_config.json

@@ -0,0 +1,8 @@
+{
+  "logit_scale": 55.869911193847656,
+  "logit_bias": 0.0,
+  "activation_function": "softmax",
+  "tokenizer_needs_lowercase": false,
+  "pad_id": 0,
+  "vocab_size": 49408
+}

open_clip_config.json

@@ -0,0 +1,38 @@
+{
+  "model_cfg": {
+    "embed_dim": 512,
+    "vision_cfg": {
+      "timm_model_name": "fastvit_mci2",
+      "timm_model_pretrained": false,
+      "timm_pool": "avg",
+      "timm_proj": null,
+      "timm_drop": 0.0,
+      "timm_drop_path": 0.0,
+      "image_size": 256
+    },
+    "text_cfg": {
+      "context_length": 77,
+      "vocab_size": 49408,
+      "width": 512,
+      "heads": 8,
+      "layers": 12,
+      "no_causal_mask": true,
+      "hf_tokenizer_name": "timm/MobileCLIP2-S2-OpenCLIP"
+    },
+    "custom_text": true
+  },
+  "preprocess_cfg": {
+    "mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "interpolation": "bilinear",
+    "resize_mode": "shortest"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

text.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7046e2a3c4078fd0b10865f3063bfa8e89c0808de3bd2fcaf89ba0d0a2fe90f
+size 1517435

text.onnx.data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0e6225b1a9e8b15b1abedcc430ebb369c9695831c9b0922acf9e83906e03e47
+size 253755392

tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "49406": {
+      "content": "<|startoftext|>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "49407": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<|startoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": true,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 77,
+  "pad_token": "<|endoftext|>",
+  "tokenizer_class": "CLIPTokenizer",
+  "unk_token": "<|endoftext|>"
+}

visual.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bb691c4085f8767c8237e324816a430f8264b5c2d4542ec2d9f88d701faa823
+size 2310022

visual.onnx.data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:365f1b95476a0852aaf5b136432383ff9405eedc3deb5347c2b8398a3838f50c
+size 143589376