Upload folder using huggingface_hub

README.md

@@ -0,0 +1,131 @@
+---
+license: apache-2.0
+pipeline_tag: feature-extraction
+tags:
+- clip
+- vision
+datasets:
+- Ziyang/yfcc15m
+- conceptual_captions
+---
+<h1 align="center">UForm</h1>
+<h3 align="center">
+Multi-Modal Inference Library<br/>
+For Semantic Search Applications<br/>
+</h3>
+
+---
+
+UForm is a Multi-Modal Modal Inference package, designed to encode Multi-Lingual Texts, Images, and, soon, Audio, Video, and Documents, into a shared vector space!
+
+This is model card of the __English only model__ with:
+
+* 12 layers BERT (6 layers for unimodal encoding and rest layers for multimodal encoding)
+* ViT-L/14 (image resolution is 224x224)
+* Multiple embedding sizes: 64, 256, 512, 768
+
+
+If you need Multilingual model, check [this](https://huggingface.co/unum-cloud/uform-vl-multilingual).
+
+## Evaluation
+
+The following metrics were obtained with multimodal re-ranking (text-to-image retrieval):
+
+| Dataset   |Recall@1 |  Recall@5 | Recall@10 |
+| :------   | ------: | --------: | --------: |
+| Zero-Shot Flickr    | 0.693 | 0.875 | 0.923 |
+| Zero-Shot MS-COCO   | 0.382 | 0.617 | 0.728 |
+
+ImageNet-Top1: 0.518 \
+ImageNet-Top5: 0.756
+
+## Installation
+
+```bash
+pip install uform[torch]
+```
+
+## Usage
+
+To load the model:
+
+```python
+import uform
+
+model, processor = uform.get_model('unum-cloud/uform-vl-english-large')
+```
+
+To encode data:
+
+```python
+from PIL import Image
+
+text = 'a small red panda in a zoo'
+image = Image.open('red_panda.jpg')
+
+image_data = processor.preprocess_image(image)
+text_data = processor.preprocess_text(text)
+
+image_features, image_embedding = model.encode_image(image_data, return_features=True)
+text_features, text_embedding = model.encode_text(text_data, return_features=True)
+joint_embedding = model.encode_multimodal(image=image_data, text=text_data)
+```
+
+To get features:
+
+```python
+image_features, image_embedding = model.encode_image(image_data, return_features=True)
+text_features, text_embedding = model.encode_text(text_data, return_features=True)
+```
+
+These features can later be used to produce joint multimodal encodings faster, as the first layers of the transformer can be skipped:
+
+```python
+joint_embedding = model.encode_multimodal(
+    image_features=image_features,
+    text_features=text_features,
+    attention_mask=text_data['attention_mask']
+)
+```
+
+There are two options to calculate semantic compatibility between an image and a text: [Cosine Similarity](#cosine-similarity) and [Matching Score](#matching-score).
+
+### Cosine Similarity
+
+```python
+import torch.nn.functional as F
+
+similarity = F.cosine_similarity(image_embedding, text_embedding)
+```
+
+The `similarity` will belong to the `[-1, 1]` range, `1` meaning the absolute match.
+
+__Pros__:
+
+- Computationally cheap.
+- Only unimodal embeddings are required, unimodal encoding is faster than joint encoding.
+- Suitable for retrieval in large collections.
+
+__Cons__:
+
+- Takes into account only coarse-grained features.
+
+
+### Matching Score 
+
+Unlike cosine similarity, unimodal embedding are not enough.
+Joint embedding will be needed and the resulting `score` will belong to the `[0, 1]` range, `1` meaning the absolute match.
+
+```python
+score = model.get_matching_scores(joint_embedding)
+```
+
+__Pros__:
+
+- Joint embedding captures fine-grained features.
+- Suitable for re-ranking – sorting retrieval result.
+
+__Cons__:
+
+- Resource-intensive.
+- Not suitable for retrieval in large collections.

config.json

@@ -0,0 +1,45 @@
+{
+    "text_encoder": {
+        "tokenizer_class": "bert",
+        "model_type": "bert",
+        "dim": 512,
+        "context_dim": 1024,
+        "vocab_size": 30522,
+        "padding_idx": 0,
+        "num_layers": 12,
+        "num_heads": 8,
+        "embedding_dim": 768,
+        "multimodal_layers_ids": [
+            6,
+            7,
+            8,
+            9,
+            10,
+            11
+        ],
+        "head_one_neuron": false,
+        "pooling": "cls",
+        "max_position_embeddings": 64,
+        "dropout_prob": 0.1
+    },
+    "image_encoder": {
+        "normalization_means": [
+            0.48145466,
+            0.4578275,
+            0.40821073
+        ],
+        "normalization_deviations": [
+            0.26862954,
+            0.26130258,
+            0.27577711
+        ],
+        "dim": 1024,
+        "patch_size": 14,
+        "image_size": 224,
+        "num_layers": 24,
+        "num_heads": 16,
+        "embedding_dim": 768,
+        "pooling": "cls",
+        "num_reg_tokens": 4
+    }
+}

image_encoder.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9e7674bf0b3cb4888c07f856afdf4a2c726d9897abea51d2dc09bbf891f2b87
+size 219785

image_encoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c7e5554fd2b3c1a28c338d3dad05ec621660d63c2c10e2d12851276c76c2fe2
+size 1216100992

image_encoder.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "5D0A1DEC-240F-4858-9B65-E24C5ECD8FA0": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        },
+        "B69ED582-7925-4F67-8551-49D55D91B556": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        }
+    },
+    "rootModelIdentifier": "5D0A1DEC-240F-4858-9B65-E24C5ECD8FA0"
+}

image_encoder.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d811e8175034ff69bb81af939eb9de9e43a190e51418d64c3a80d1014f9bea04
+size 306693409

image_encoder.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7eb875122465970de652c96b1c5d10f3d8a92ead5e986d8fef139c6be4a526a1
+size 608182826

image_encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b351cecc9bd6a336ca0691f89e3d4356dd8568d7425bdd6a23deff1f680dd987
+size 608077816

text_encoder.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad16992287a7c84131e8c2a8c19c44030f10379726581dab51a3dc9905306341
+size 56199

text_encoder.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b666e07fb083deef657480a19e6b8c8b323a01a001e3d2913c9566cc5a19eab
+size 139883968

text_encoder.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "88941D4F-6EE3-4492-92F1-4714D520C2B2": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        },
+        "C5FE30C0-9C0C-4F26-96FC-29FA64D92CE7": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        }
+    },
+    "rootModelIdentifier": "C5FE30C0-9C0C-4F26-96FC-29FA64D92CE7"
+}

text_encoder.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57df732e74740b70ecf1788478d7936bdffae3cec4eb997bfb6c7a59aa06f6af
+size 35272692

text_encoder.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b7ab3f8e506099dc62cadb85705bf8b91bfbc000474858448173d27ff7dab33
+size 121525138

text_encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:893185006938771e00437e95a6a40504e327ffdca1b7bb4c6bc78024e6c09881
+size 121461852