Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,45 +6,58 @@ import pandas as pd
|
|
| 6 |
from datasets import load_dataset
|
| 7 |
from sklearn.metrics.pairwise import cosine_similarity
|
| 8 |
import numpy as np
|
|
|
|
|
|
|
| 9 |
|
| 10 |
# Load Florence-2 model and processor
|
| 11 |
model_name = "microsoft/Florence-2-base"
|
| 12 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 13 |
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
|
| 14 |
|
|
|
|
| 15 |
model = AutoModelForCausalLM.from_pretrained(
|
| 16 |
model_name,
|
| 17 |
torch_dtype=torch_dtype,
|
| 18 |
-
trust_remote_code=True
|
|
|
|
| 19 |
).to(device)
|
| 20 |
processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)
|
| 21 |
|
| 22 |
# Load CivitAI dataset (limited to 1000 samples)
|
|
|
|
| 23 |
dataset = load_dataset("thefcraft/civitai-stable-diffusion-337k", split="train[:1000]")
|
| 24 |
df = pd.DataFrame(dataset)
|
|
|
|
| 25 |
|
| 26 |
# Create cache for embeddings to improve performance
|
| 27 |
text_embedding_cache = {}
|
| 28 |
|
| 29 |
def get_image_embedding(image):
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
|
| 35 |
def get_text_embedding(text):
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
|
| 47 |
-
# Pre-compute text embeddings for all prompts in the dataset
|
| 48 |
def precompute_embeddings():
|
| 49 |
print("Pre-computing text embeddings...")
|
| 50 |
for idx, row in df.iterrows():
|
|
@@ -55,21 +68,21 @@ def precompute_embeddings():
|
|
| 55 |
print("Finished pre-computing embeddings")
|
| 56 |
|
| 57 |
def find_similar_images(uploaded_image, top_k=5):
|
| 58 |
-
# Get embedding for uploaded image
|
| 59 |
query_embedding = get_image_embedding(uploaded_image)
|
|
|
|
|
|
|
| 60 |
|
| 61 |
-
# Calculate similarities with dataset
|
| 62 |
similarities = []
|
| 63 |
for idx, row in df.iterrows():
|
| 64 |
prompt_embedding = get_text_embedding(row['prompt'])
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
|
|
|
| 71 |
|
| 72 |
-
# Sort by similarity and get top k results
|
| 73 |
sorted_results = sorted(similarities, key=lambda x: x['similarity'], reverse=True)
|
| 74 |
top_models = []
|
| 75 |
top_prompts = []
|
|
@@ -94,21 +107,28 @@ def process_image(input_image):
|
|
| 94 |
if input_image is None:
|
| 95 |
return "Please upload an image.", "Please upload an image."
|
| 96 |
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
|
| 110 |
# Pre-compute embeddings when starting the application
|
| 111 |
-
|
|
|
|
|
|
|
|
|
|
| 112 |
|
| 113 |
# Create Gradio interface
|
| 114 |
iface = gr.Interface(
|
|
|
|
| 6 |
from datasets import load_dataset
|
| 7 |
from sklearn.metrics.pairwise import cosine_similarity
|
| 8 |
import numpy as np
|
| 9 |
+
import warnings
|
| 10 |
+
warnings.filterwarnings('ignore')
|
| 11 |
|
| 12 |
# Load Florence-2 model and processor
|
| 13 |
model_name = "microsoft/Florence-2-base"
|
| 14 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 15 |
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
|
| 16 |
|
| 17 |
+
# Modify model loading to disable flash attention
|
| 18 |
model = AutoModelForCausalLM.from_pretrained(
|
| 19 |
model_name,
|
| 20 |
torch_dtype=torch_dtype,
|
| 21 |
+
trust_remote_code=True,
|
| 22 |
+
use_flash_attention=False # Disable flash attention
|
| 23 |
).to(device)
|
| 24 |
processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)
|
| 25 |
|
| 26 |
# Load CivitAI dataset (limited to 1000 samples)
|
| 27 |
+
print("Loading dataset...")
|
| 28 |
dataset = load_dataset("thefcraft/civitai-stable-diffusion-337k", split="train[:1000]")
|
| 29 |
df = pd.DataFrame(dataset)
|
| 30 |
+
print("Dataset loaded successfully!")
|
| 31 |
|
| 32 |
# Create cache for embeddings to improve performance
|
| 33 |
text_embedding_cache = {}
|
| 34 |
|
| 35 |
def get_image_embedding(image):
|
| 36 |
+
try:
|
| 37 |
+
inputs = processor(images=image, return_tensors="pt").to(device, torch_dtype)
|
| 38 |
+
with torch.no_grad():
|
| 39 |
+
outputs = model.get_image_features(**inputs)
|
| 40 |
+
return outputs.cpu().numpy()
|
| 41 |
+
except Exception as e:
|
| 42 |
+
print(f"Error in get_image_embedding: {str(e)}")
|
| 43 |
+
return None
|
| 44 |
|
| 45 |
def get_text_embedding(text):
|
| 46 |
+
try:
|
| 47 |
+
if text in text_embedding_cache:
|
| 48 |
+
return text_embedding_cache[text]
|
| 49 |
+
|
| 50 |
+
inputs = processor(text=text, return_tensors="pt").to(device, torch_dtype)
|
| 51 |
+
with torch.no_grad():
|
| 52 |
+
outputs = model.get_text_features(**inputs)
|
| 53 |
+
|
| 54 |
+
embedding = outputs.cpu().numpy()
|
| 55 |
+
text_embedding_cache[text] = embedding
|
| 56 |
+
return embedding
|
| 57 |
+
except Exception as e:
|
| 58 |
+
print(f"Error in get_text_embedding: {str(e)}")
|
| 59 |
+
return None
|
| 60 |
|
|
|
|
| 61 |
def precompute_embeddings():
|
| 62 |
print("Pre-computing text embeddings...")
|
| 63 |
for idx, row in df.iterrows():
|
|
|
|
| 68 |
print("Finished pre-computing embeddings")
|
| 69 |
|
| 70 |
def find_similar_images(uploaded_image, top_k=5):
|
|
|
|
| 71 |
query_embedding = get_image_embedding(uploaded_image)
|
| 72 |
+
if query_embedding is None:
|
| 73 |
+
return [], []
|
| 74 |
|
|
|
|
| 75 |
similarities = []
|
| 76 |
for idx, row in df.iterrows():
|
| 77 |
prompt_embedding = get_text_embedding(row['prompt'])
|
| 78 |
+
if prompt_embedding is not None:
|
| 79 |
+
similarity = cosine_similarity(query_embedding, prompt_embedding)[0][0]
|
| 80 |
+
similarities.append({
|
| 81 |
+
'similarity': similarity,
|
| 82 |
+
'model': row['Model'],
|
| 83 |
+
'prompt': row['prompt']
|
| 84 |
+
})
|
| 85 |
|
|
|
|
| 86 |
sorted_results = sorted(similarities, key=lambda x: x['similarity'], reverse=True)
|
| 87 |
top_models = []
|
| 88 |
top_prompts = []
|
|
|
|
| 107 |
if input_image is None:
|
| 108 |
return "Please upload an image.", "Please upload an image."
|
| 109 |
|
| 110 |
+
try:
|
| 111 |
+
if not isinstance(input_image, Image.Image):
|
| 112 |
+
input_image = Image.fromarray(input_image)
|
| 113 |
+
|
| 114 |
+
recommended_models, recommended_prompts = find_similar_images(input_image)
|
| 115 |
+
|
| 116 |
+
if not recommended_models or not recommended_prompts:
|
| 117 |
+
return "Error processing image.", "Error processing image."
|
| 118 |
+
|
| 119 |
+
models_text = "Recommended Models:\n" + "\n".join([f"{i+1}. {model}" for i, model in enumerate(recommended_models)])
|
| 120 |
+
prompts_text = "Recommended Prompts:\n" + "\n".join([f"{i+1}. {prompt}" for i, prompt in enumerate(recommended_prompts)])
|
| 121 |
+
|
| 122 |
+
return models_text, prompts_text
|
| 123 |
+
except Exception as e:
|
| 124 |
+
print(f"Error in process_image: {str(e)}")
|
| 125 |
+
return "Error processing image.", "Error processing image."
|
| 126 |
|
| 127 |
# Pre-compute embeddings when starting the application
|
| 128 |
+
try:
|
| 129 |
+
precompute_embeddings()
|
| 130 |
+
except Exception as e:
|
| 131 |
+
print(f"Error in precompute_embeddings: {str(e)}")
|
| 132 |
|
| 133 |
# Create Gradio interface
|
| 134 |
iface = gr.Interface(
|