fix florence

cog_test_train.sh

@@ -3,10 +3,10 @@ GPU_ID="device=3"
 
 cog predict --gpus $GPU_ID \
     -i name="xander_sdxl_cog" \
-    -i lora_training_urls="https://storage.googleapis.com/public-assets-xander/A_workbox/lora_training_sets/xander_big.zip" \
+    -i lora_training_urls="https://storage.googleapis.com/public-assets-xander/A_workbox/lora_training_sets/xander.zip" \
     -i concept_mode="face" \
     -i sd_model_version="sdxl" \
-    -i max_train_steps="360" \
-    -i caption_model="blip" \
+    -i max_train_steps="300" \
+    -i caption_model="florence" \
     -i debug="False" \
     -i seed="0"

predict.py

@@ -175,9 +175,13 @@ class Predictor(BasePredictor):
             
             # Add instructions README:
             tar.add("instructions_README.md", arcname="README.md")
-            tar.add("comfyUI_workflow_lora_txt2img.json", arcname="comfyUI_workflow_lora_txt2img.json")
-            if sd_model_version == "sd15":
-                tar.add("comfyUI_workflow_lora_adiff.json", arcname="comfyUI_workflow_lora_adiff.json")
+            comfy_workflows_path = "ComfyUI_workflows"
+            if os.path.exists(comfy_workflows_path) and os.path.isdir(comfy_workflows_path):
+                for root, dirs, files in os.walk(comfy_workflows_path):
+                    for file in files:
+                        file_path = os.path.join(root, file)
+                        arcname = os.path.relpath(file_path, os.path.dirname(comfy_workflows_path))
+                        tar.add(file_path, arcname=arcname)
 
         attributes = {}
         attributes['grid_prompts'] = config.training_attributes["validation_prompts"]

requirements.txt

@@ -21,4 +21,5 @@ ujson==5.10.0
 bitsandbytes==0.43.1
 setuptools==70.3.0
 torchtyping==0.1.5
-einops
+einops
+timm

train_configs/training_args_style_sdxl.json

@@ -1,5 +1,5 @@
 {
-    "name": "clipx",
+    "name": "twisting_realities",
     "sd_model_version": "sdxl",
     "lora_training_urls": "https://edenartlab-lfs.s3.amazonaws.com/datasets/twisting_realities.zip",
     "concept_mode": "style",
@@ -7,7 +7,7 @@
     "seed": 0,
     "resolution": 512,
     "train_batch_size": 4,
-    "n_sample_imgs": 8,
+    "n_sample_imgs": 6,
     "max_train_steps": 300,
     "checkpointing_steps": 200,
 

trainer/preprocess.py

@@ -562,7 +562,7 @@ def florence_caption_dataset(images, captions):
     with patch("transformers.dynamic_module_utils.get_imports", fixed_get_imports): #workaround for unnecessary flash_attn requirement
         model = AutoModelForCausalLM.from_pretrained("microsoft/Florence-2-large", attn_implementation="sdpa", device_map=device, torch_dtype=torch_dtype,trust_remote_code=True)
             
-    processor = AutoProcessor.from_pretrained("microsoft/Florence-2-large", trust_remote_code=True)
+    processor = AutoProcessor.from_pretrained("microsoft/Florence-2-large", trust_remote_code=True, cache_dir = model_paths.get_path("FLORENCE"))
 
     for i, image in enumerate(tqdm(images)):
         if captions[i] is None:
@@ -675,15 +675,15 @@ def random_crop(image, scale=(0.85, 0.95)):
 
     return image.crop((left, top, left + new_width, top + new_height))
 
-def gaussian_blur(image):
-    return image.filter(ImageFilter.GaussianBlur(radius=1))
+def gaussian_blur(image, radius = 1.0):
+    return image.filter(ImageFilter.GaussianBlur(radius=radius))
 
 def augment_image(image):
     image = hue_augmentation(image)
     image = color_jitter(image)
     image = random_crop(image)
     if random.random() < 0.5:
-        image = gaussian_blur(image, blur = random.uniform(0.0, 1.0))
+        image = gaussian_blur(image, radius = random.uniform(0.0, 1.0))
     return image
 
 def round_to_nearest_multiple(x, multiple):