add dalle visualization

clip_advice.py

@@ -36,6 +36,10 @@
 import omegaconf
 import ast
 
+# from dalle2_laion import ModelLoadConfig, DalleModelManager
+# from dalle2_laion.scripts import InferenceScript
+# from test import ExampleInference
+
 parser = argparse.ArgumentParser(description='CLIP Advice')
 parser.add_argument('--config', default='configs/Noop.yaml', help="config file")
 parser.add_argument('overrides', nargs='*', help="Any key=value arguments to override config values "
@@ -221,6 +225,14 @@ def flatten_config(dic, running_key=None, flattened_dict={}):
             except:
                 print(f"sample idx {sample_idx} is not a valid index")
         wandb.log({"train features NN": wandb.Image(f), "domain consistency acc": domain_acc, "class consistency acc": class_acc, "unique nn": prop_unique})
+
+        sample_idxs = random.sample(list(range(len(train_features))), 10)
+        with open('dalle/embeddings.npy', 'wb') as f:
+            for i in sample_idxs:
+                np.save(f, train_features[i])
+                np.save(f, train_features[i+1])
+                print(f"original domain: {train_domains[i]}, augmented domain: {train_domains[i+1]}")
+
         wandb.sklearn.plot_confusion_matrix(sample_domains, neighbor_domains, dataset_domains)
         if args.EXP.LOG_EMB_DRIFT:
             sample_idxs = random.sample(list(range(len(val_features))), min([len(val_features), 1000]))

dalle/dalle2.json

@@ -0,0 +1,38 @@
+{
+    "decoder": {
+        "unet_sources": [
+            {
+                "unet_numbers": [1],
+                "default_cond_scale": [1.7],
+                "load_model_from": {
+                    "load_type": "url",
+                    "path": "https://huggingface.co/laion/DALLE2-PyTorch/resolve/main/decoder/v1.0.2/latest.pth",
+                    "cache_dir": "./dalle/models",
+                    "filename_override": "new_decoder.pth"
+                }
+            },
+            {
+                "unet_numbers": [2],
+                "load_model_from": {
+                    "load_type": "url",
+                    "path": "https://huggingface.co/Veldrovive/upsamplers/resolve/main/working/latest.pth",
+                    "cache_dir": "./dalle/models",
+                    "filename_override": "second_decoder.pth"
+                },
+                "load_config_from": {
+                    "load_type": "url",
+                    "path": "https://huggingface.co/Veldrovive/upsamplers/raw/main/working/decoder_config.json",
+                    "checksum_file_path": "https://huggingface.co/Veldrovive/upsamplers/raw/main/working/decoder_config.json",
+                    "cache_dir": "./dalle/models",
+                    "filename_override": "second_decoder_config.json"
+                }
+            }
+        ]
+    },
+    "clip": {
+        "make": "openai",
+        "model": "ViT-L/14"
+    },
+    "devices": "cuda:0",
+    "strict_loading": false
+}

dalle/setup.py

@@ -0,0 +1,17 @@
+from setuptools import setup, find_packages
+
+setup(
+    name = "dalle2-laion",
+    version = "0.0.1",
+    packages = find_packages(exclude=[]),
+    include_package_data = True,
+    install_requires = [
+        "packaging>=21.0",
+        "pydantic>=1.9.0",
+        "torch>=1.10",
+        "Pillow>=9.0.0",
+        "numpy>=1.20.0",
+        "click>=8.0.0"
+        "dalle2-pytorch"
+    ]
+)

dalle/visualization.py

@@ -0,0 +1,38 @@
+from dalle2_laion import ModelLoadConfig, DalleModelManager
+from dalle2_laion.scripts import InferenceScript
+import torch
+import numpy as np
+from PIL import Image
+
+class ExampleInference(InferenceScript):
+    def run(self, text, image_embedding) -> Image:
+        image_map = self._sample_decoder(text=text, image_embed=image_embedding)
+        return image_map[0][0]
+
+model_config = ModelLoadConfig.from_json_path("./dalle/dalle2.json")
+model_manager = DalleModelManager(model_config)
+inference = ExampleInference(model_manager)
+
+original_embed = []
+augmented_embed = []
+with open('./dalle/embeddings.npy', 'rb') as f:
+    for _ in range(10):
+        a = np.load(f)
+        original_embed.append(torch.from_numpy(a))
+        b = np.load(f)
+        augmented_embed.append(torch.from_numpy(b))
+
+# original_domain = [1, 1, 0, 0, 0, 0, 1, 1, 0, 1]
+# original_text = ['a photo of a landbird in the forest.', 'a photo of a waterbird on water.']
+# augmented_text = ['a photo of a waterbird in the forest.', 'a photo of a landbird on water.']
+for i in range(10):
+    image = inference.run(["a photo"], [original_embed[i]])
+    image.save(f'./dalle/images/waterbird-original-photo-{i}.png')
+    image = inference.run(["a photo"], [augmented_embed[i]])
+    image.save(f'./dalle/images/waterbird-augment-photo-{i}.png')
+
+for i in range(10):
+    image = inference.run(["a car"], [original_embed[i]])
+    image.save(f'./dalle/images/waterbird-original-car-{i}.png')
+    image = inference.run(["a car"], [augmented_embed[i]])
+    image.save(f'./dalle/images/waterbird-augment-car-{i}.png')