Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,36 +1,27 @@
|
|
1 |
import gradio as gr
|
2 |
-
import spaces #
|
3 |
from transformers import pipeline
|
4 |
-
import torch
|
5 |
|
6 |
-
#
|
7 |
-
|
|
|
|
|
8 |
|
9 |
-
|
10 |
-
global generator
|
11 |
-
if generator is None:
|
12 |
-
try:
|
13 |
-
# If GPU is available, load on GPU (device=0)
|
14 |
-
if torch.cuda.is_available():
|
15 |
-
generator = pipeline("text-generation", model="EleutherAI/gpt-j-6B", device=0)
|
16 |
-
else:
|
17 |
-
generator = pipeline("text-generation", model="EleutherAI/gpt-j-6B", device=-1)
|
18 |
-
except Exception as e:
|
19 |
-
print("Error loading model on GPU, falling back to CPU:", e)
|
20 |
-
generator = pipeline("text-generation", model="EleutherAI/gpt-j-6B", device=-1)
|
21 |
-
return generator
|
22 |
-
|
23 |
-
@spaces.GPU # This decorator ensures ZeroGPU allocates a GPU when the function is called.
|
24 |
def expand_prompt(prompt, num_variants=5, max_length=100):
|
25 |
"""
|
26 |
-
Given a basic prompt, generate `num_variants` expanded
|
27 |
-
|
28 |
"""
|
29 |
-
|
30 |
-
|
|
|
|
|
|
|
31 |
expanded = [out["generated_text"].strip() for out in outputs]
|
32 |
return "\n\n".join(expanded)
|
33 |
|
|
|
34 |
iface = gr.Interface(
|
35 |
fn=expand_prompt,
|
36 |
inputs=gr.Textbox(lines=2, placeholder="Enter your basic prompt here...", label="Basic Prompt"),
|
@@ -38,9 +29,8 @@ iface = gr.Interface(
|
|
38 |
title="Prompt Expansion Generator",
|
39 |
description=(
|
40 |
"Enter a basic prompt and receive 5 creative, expanded prompt variants. "
|
41 |
-
"
|
42 |
-
"
|
43 |
-
"Simply copy the output for use with your downstream image-generation pipeline."
|
44 |
)
|
45 |
)
|
46 |
|
|
|
1 |
import gradio as gr
|
2 |
+
import spaces # ZeroGPU helper module
|
3 |
from transformers import pipeline
|
|
|
4 |
|
5 |
+
# Preload the text-generation model on CPU at startup.
|
6 |
+
# Model: EleutherAI/gpt-j-6B (https://huggingface.co/EleutherAI/gpt-j-6B)
|
7 |
+
# We load on CPU (device=-1) so that initialization is done before the GUI is up.
|
8 |
+
generator = pipeline("text-generation", model="EleutherAI/gpt-j-6B", device=-1)
|
9 |
|
10 |
+
@spaces.GPU # This decorator ensures ZeroGPU allocates a GPU only during this function call.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
def expand_prompt(prompt, num_variants=5, max_length=100):
|
12 |
"""
|
13 |
+
Given a basic prompt, generate `num_variants` expanded prompt variants.
|
14 |
+
Before generation, the model is moved to GPU (A100), and after generation it's moved back to CPU.
|
15 |
"""
|
16 |
+
# Move the model to GPU for generation.
|
17 |
+
generator.model.to("cuda")
|
18 |
+
outputs = generator(prompt, max_length=max_length, num_return_sequences=num_variants, do_sample=True)
|
19 |
+
# Move the model back to CPU after generation.
|
20 |
+
generator.model.to("cpu")
|
21 |
expanded = [out["generated_text"].strip() for out in outputs]
|
22 |
return "\n\n".join(expanded)
|
23 |
|
24 |
+
# Create a Gradio Interface
|
25 |
iface = gr.Interface(
|
26 |
fn=expand_prompt,
|
27 |
inputs=gr.Textbox(lines=2, placeholder="Enter your basic prompt here...", label="Basic Prompt"),
|
|
|
29 |
title="Prompt Expansion Generator",
|
30 |
description=(
|
31 |
"Enter a basic prompt and receive 5 creative, expanded prompt variants. "
|
32 |
+
"The model (EleutherAI/gpt-j-6B) is preloaded on CPU at startup and then moved to GPU (via ZeroGPU) only "
|
33 |
+
"when a prompt is submitted. Simply copy the output for use with your downstream image-generation pipeline."
|
|
|
34 |
)
|
35 |
)
|
36 |
|