Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,29 +1,24 @@
|
|
1 |
import torch
|
2 |
import gradio as gr
|
3 |
-
from transformers import
|
4 |
-
from peft import
|
5 |
-
|
6 |
-
# Load the PEFT model configuration and quantization settings
|
7 |
-
peft_model_id = "Prasi21/blip2-opt-2.7b-strep-throat-caption-adapters3"
|
8 |
-
config = PeftConfig.from_pretrained(peft_model_id)
|
9 |
-
config.base_model_name_or_path = "Prasi21/blip2-opt-2.7b-strep-throat-caption-adapters3"
|
10 |
-
|
11 |
-
# Enable 8-bit quantization for more efficient loading
|
12 |
-
quantization_config = BitsAndBytesConfig(load_in_8bit=True)
|
13 |
-
|
14 |
-
# Load the base model with quantization
|
15 |
-
model = Blip2ForConditionalGeneration.from_pretrained(
|
16 |
-
config.base_model_name_or_path,
|
17 |
-
quantization_config=quantization_config,
|
18 |
-
device_map="auto"
|
19 |
-
)
|
20 |
-
|
21 |
-
# Load the fine-tuned PEFT model
|
22 |
-
model = PeftModel.from_pretrained(model, peft_model_id)
|
23 |
|
24 |
# Load the processor
|
25 |
processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
|
26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
# Define the prediction function
|
28 |
def predict(image):
|
29 |
# Preprocess the image
|
|
|
1 |
import torch
|
2 |
import gradio as gr
|
3 |
+
from transformers import AutoProcessor, Blip2ForConditionalGeneration
|
4 |
+
from peft import LoraConfig, get_peft_model, PeftModel
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
6 |
# Load the processor
|
7 |
processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
|
8 |
|
9 |
+
# Load the base model from the original repository
|
10 |
+
base_model = Blip2ForConditionalGeneration.from_pretrained(
|
11 |
+
"ybelkada/blip2-opt-2.7b-fp16-sharded",
|
12 |
+
device_map="auto",
|
13 |
+
quantization_config=quantization_config
|
14 |
+
)
|
15 |
+
|
16 |
+
repo_id = "Prasi21/blip2-opt-2.7b-strep-throat-caption-adapters"
|
17 |
+
|
18 |
+
# Load the fine-tuned LoRA adapters from the Hugging Face Hub
|
19 |
+
model = PeftModel.from_pretrained(base_model, repo_id)
|
20 |
+
|
21 |
+
|
22 |
# Define the prediction function
|
23 |
def predict(image):
|
24 |
# Preprocess the image
|