Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -27,7 +27,7 @@ qwen_model = Qwen2VLForConditionalGeneration.from_pretrained(
|
|
27 |
qwen_processor = AutoProcessor.from_pretrained("prithivMLmods/JSONify-Flux", trust_remote_code=True)
|
28 |
|
29 |
# Prompt Enhancer
|
30 |
-
enhancer_long = pipeline("summarization", model="
|
31 |
|
32 |
MAX_SEED = np.iinfo(np.int32).max
|
33 |
MAX_IMAGE_SIZE = 2048
|
@@ -138,36 +138,33 @@ title = """<h1 align="center">FLUX.1-dev with Qwen2VL Captioner and Prompt Enhan
|
|
138 |
<p><center>
|
139 |
<a href="https://huggingface.co/black-forest-labs/FLUX.1-dev" target="_blank">[FLUX.1-dev Model]</a>
|
140 |
<a href="https://huggingface.co/prithivMLmods/JSONify-Flux" target="_blank">[JSONify Flux Model]</a>
|
141 |
-
<a href="https://huggingface.co/
|
142 |
<p align="center">Create long prompts from images or enhance your short prompts with prompt enhancer</p>
|
143 |
</center></p>
|
144 |
"""
|
145 |
|
146 |
with gr.Blocks(css=custom_css) as demo:
|
|
|
147 |
|
148 |
-
# Sidebar with About details
|
149 |
with gr.Sidebar(label="Parameters", open=True):
|
150 |
gr.Markdown(
|
151 |
"""
|
152 |
### About
|
153 |
|
154 |
-
####
|
155 |
-
FLUX.1 [dev] is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions.
|
156 |
-
FLUX.1
|
157 |
-
[FLUX.1-dev on Hugging Face](https://huggingface.co/black-forest-labs/FLUX.1-dev)
|
158 |
|
159 |
#### JSONify-Flux
|
160 |
-
JSONify-Flux is a multimodal image-text model
|
161 |
-
[JSONify-Flux
|
162 |
|
163 |
#### t5-Flan-Prompt-Enhance
|
164 |
-
t5-Flan-Prompt-Enhance is a prompt summarization model
|
165 |
-
[t5-Flan-Prompt-Enhance
|
166 |
"""
|
167 |
)
|
168 |
|
169 |
-
gr.HTML(title)
|
170 |
-
|
171 |
with gr.Row():
|
172 |
with gr.Column(scale=1):
|
173 |
with gr.Group(elem_classes="input-group"):
|
@@ -181,15 +178,15 @@ with gr.Blocks(css=custom_css) as demo:
|
|
181 |
width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
|
182 |
height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
|
183 |
guidance_scale = gr.Slider(label="Guidance Scale", minimum=1, maximum=15, step=0.1, value=3.5)
|
184 |
-
num_inference_steps = gr.Slider(label="Inference Steps", minimum=1, maximum=50, step=1, value=
|
185 |
|
186 |
-
generate_btn = gr.Button("Generate Image
|
187 |
|
188 |
with gr.Column(scale=1):
|
189 |
with gr.Group(elem_classes="output-group"):
|
190 |
-
output_image = gr.Image(label="
|
191 |
-
final_prompt = gr.Textbox(label="
|
192 |
-
used_seed = gr.Number(label="
|
193 |
|
194 |
generate_btn.click(
|
195 |
fn=process_workflow,
|
@@ -200,4 +197,4 @@ with gr.Blocks(css=custom_css) as demo:
|
|
200 |
outputs=[output_image, final_prompt, used_seed]
|
201 |
)
|
202 |
|
203 |
-
demo.launch(debug=True)
|
|
|
27 |
qwen_processor = AutoProcessor.from_pretrained("prithivMLmods/JSONify-Flux", trust_remote_code=True)
|
28 |
|
29 |
# Prompt Enhancer
|
30 |
+
enhancer_long = pipeline("summarization", model="prithivMLmods/t5-Flan-Prompt-Enhance", device=device)
|
31 |
|
32 |
MAX_SEED = np.iinfo(np.int32).max
|
33 |
MAX_IMAGE_SIZE = 2048
|
|
|
138 |
<p><center>
|
139 |
<a href="https://huggingface.co/black-forest-labs/FLUX.1-dev" target="_blank">[FLUX.1-dev Model]</a>
|
140 |
<a href="https://huggingface.co/prithivMLmods/JSONify-Flux" target="_blank">[JSONify Flux Model]</a>
|
141 |
+
<a href="https://huggingface.co/prithivMLmods/t5-Flan-Prompt-Enhance" target="_blank">[Prompt Enhancer t5]</a>
|
142 |
<p align="center">Create long prompts from images or enhance your short prompts with prompt enhancer</p>
|
143 |
</center></p>
|
144 |
"""
|
145 |
|
146 |
with gr.Blocks(css=custom_css) as demo:
|
147 |
+
gr.HTML(title)
|
148 |
|
|
|
149 |
with gr.Sidebar(label="Parameters", open=True):
|
150 |
gr.Markdown(
|
151 |
"""
|
152 |
### About
|
153 |
|
154 |
+
#### Flux.1-Dev
|
155 |
+
FLUX.1 [dev] is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions. FLUX.1 [dev] is an open-weight, guidance-distilled model for non-commercial applications. Directly distilled from FLUX.1 [pro], FLUX.1 [dev] obtains similar quality and prompt adherence capabilities, while being more efficient than a standard model of the same size.
|
156 |
+
[FLUX.1-dev Model](https://huggingface.co/black-forest-labs/FLUX.1-dev)
|
|
|
157 |
|
158 |
#### JSONify-Flux
|
159 |
+
JSONify-Flux is a multimodal image-text-text model trained on a dataset of FLUX-generated images with context-rich captions based on the Qwen2VL architecture. The JSON-based instruction has been manually removed to avoid JSON format captions.
|
160 |
+
[JSONify-Flux Model](https://huggingface.co/prithivMLmods/JSONify-Flux)
|
161 |
|
162 |
#### t5-Flan-Prompt-Enhance
|
163 |
+
t5-Flan-Prompt-Enhance is a prompt summarization model that enriches synthetic FLUX prompts with more detailed descriptions.
|
164 |
+
[t5-Flan-Prompt-Enhance Model](https://huggingface.co/prithivMLmods/t5-Flan-Prompt-Enhance)
|
165 |
"""
|
166 |
)
|
167 |
|
|
|
|
|
168 |
with gr.Row():
|
169 |
with gr.Column(scale=1):
|
170 |
with gr.Group(elem_classes="input-group"):
|
|
|
178 |
width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
|
179 |
height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
|
180 |
guidance_scale = gr.Slider(label="Guidance Scale", minimum=1, maximum=15, step=0.1, value=3.5)
|
181 |
+
num_inference_steps = gr.Slider(label="Inference Steps", minimum=1, maximum=50, step=1, value=32)
|
182 |
|
183 |
+
generate_btn = gr.Button("Generate Image & Prompt", elem_classes="submit-btn")
|
184 |
|
185 |
with gr.Column(scale=1):
|
186 |
with gr.Group(elem_classes="output-group"):
|
187 |
+
output_image = gr.Image(label="result", elem_id="gallery", show_label=False)
|
188 |
+
final_prompt = gr.Textbox(label="prompt")
|
189 |
+
used_seed = gr.Number(label="seed")
|
190 |
|
191 |
generate_btn.click(
|
192 |
fn=process_workflow,
|
|
|
197 |
outputs=[output_image, final_prompt, used_seed]
|
198 |
)
|
199 |
|
200 |
+
demo.launch(debug=True)
|