Spaces:
Running
on
L40S
Running
on
L40S
ryanzhangfan
commited on
Commit
•
db312d6
1
Parent(s):
9f2b36a
Update app.py
Browse files
app.py
CHANGED
@@ -44,6 +44,9 @@ EMU_GEN_HUB = "BAAI/Emu3-Gen"
|
|
44 |
EMU_CHAT_HUB = "BAAI/Emu3-Chat"
|
45 |
VQ_HUB = "BAAI/Emu3-VisionTokenizer"
|
46 |
|
|
|
|
|
|
|
47 |
# Prepare models and processors
|
48 |
# Emu3-Gen model and processor
|
49 |
gen_model = AutoModelForCausalLM.from_pretrained(
|
@@ -54,15 +57,6 @@ gen_model = AutoModelForCausalLM.from_pretrained(
|
|
54 |
trust_remote_code=True,
|
55 |
)
|
56 |
|
57 |
-
# Emu3-Chat model and processor
|
58 |
-
chat_model = AutoModelForCausalLM.from_pretrained(
|
59 |
-
EMU_CHAT_HUB,
|
60 |
-
device_map="cpu",
|
61 |
-
torch_dtype=torch.bfloat16,
|
62 |
-
attn_implementation="flash_attention_2",
|
63 |
-
trust_remote_code=True,
|
64 |
-
)
|
65 |
-
|
66 |
tokenizer = AutoTokenizer.from_pretrained(EMU_CHAT_HUB, trust_remote_code=True)
|
67 |
image_processor = AutoImageProcessor.from_pretrained(
|
68 |
VQ_HUB, trust_remote_code=True
|
@@ -70,16 +64,16 @@ image_processor = AutoImageProcessor.from_pretrained(
|
|
70 |
image_tokenizer = AutoModel.from_pretrained(
|
71 |
VQ_HUB, device_map="cpu", trust_remote_code=True
|
72 |
).eval()
|
73 |
-
processor = Emu3Processor(
|
74 |
-
image_processor, image_tokenizer, tokenizer
|
75 |
-
)
|
76 |
|
77 |
print(device)
|
78 |
gen_model.to(device)
|
79 |
-
chat_model.to(device)
|
80 |
image_tokenizer.to(device)
|
81 |
|
82 |
-
|
|
|
|
|
|
|
|
|
83 |
def generate_image(prompt):
|
84 |
POSITIVE_PROMPT = " masterpiece, film grained, best quality."
|
85 |
NEGATIVE_PROMPT = (
|
@@ -139,6 +133,48 @@ def generate_image(prompt):
|
|
139 |
return im
|
140 |
return None
|
141 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
142 |
@spaces.GPU
|
143 |
def vision_language_understanding(image, text):
|
144 |
inputs = processor(
|
@@ -176,19 +212,8 @@ def chat(history, user_input, user_image):
|
|
176 |
# Append the user input and response to the history
|
177 |
history = history + [(image2str(user_image) + "<br>" + user_input, response)]
|
178 |
else:
|
179 |
-
|
180 |
-
|
181 |
-
# Use Emu3-Gen for image generation
|
182 |
-
generated_image = generate_image(user_input)
|
183 |
-
if generated_image is not None:
|
184 |
-
# Append the user input and generated image to the history
|
185 |
-
history = history + [(user_input, image2str(generated_image))]
|
186 |
-
else:
|
187 |
-
# If image generation failed, respond with an error message
|
188 |
-
history = history + [
|
189 |
-
(user_input, "Sorry, I could not generate an image.")
|
190 |
-
]
|
191 |
-
# """
|
192 |
return history, history, gr.update(value=None)
|
193 |
|
194 |
def clear_input():
|
|
|
44 |
EMU_CHAT_HUB = "BAAI/Emu3-Chat"
|
45 |
VQ_HUB = "BAAI/Emu3-VisionTokenizer"
|
46 |
|
47 |
+
|
48 |
+
# uncomment to use gen model
|
49 |
+
"""
|
50 |
# Prepare models and processors
|
51 |
# Emu3-Gen model and processor
|
52 |
gen_model = AutoModelForCausalLM.from_pretrained(
|
|
|
57 |
trust_remote_code=True,
|
58 |
)
|
59 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
tokenizer = AutoTokenizer.from_pretrained(EMU_CHAT_HUB, trust_remote_code=True)
|
61 |
image_processor = AutoImageProcessor.from_pretrained(
|
62 |
VQ_HUB, trust_remote_code=True
|
|
|
64 |
image_tokenizer = AutoModel.from_pretrained(
|
65 |
VQ_HUB, device_map="cpu", trust_remote_code=True
|
66 |
).eval()
|
|
|
|
|
|
|
67 |
|
68 |
print(device)
|
69 |
gen_model.to(device)
|
|
|
70 |
image_tokenizer.to(device)
|
71 |
|
72 |
+
processor = Emu3Processor(
|
73 |
+
image_processor, image_tokenizer, tokenizer
|
74 |
+
)
|
75 |
+
|
76 |
+
@spaces.GPU(duration=300)
|
77 |
def generate_image(prompt):
|
78 |
POSITIVE_PROMPT = " masterpiece, film grained, best quality."
|
79 |
NEGATIVE_PROMPT = (
|
|
|
133 |
return im
|
134 |
return None
|
135 |
|
136 |
+
def chat(history, user_input, user_image):
|
137 |
+
if user_image is not None:
|
138 |
+
history = history + [("", "Sorry, gen model do not accept image input")]
|
139 |
+
else:
|
140 |
+
# Use Emu3-Gen for image generation
|
141 |
+
generated_image = generate_image(user_input)
|
142 |
+
if generated_image is not None:
|
143 |
+
# Append the user input and generated image to the history
|
144 |
+
history = history + [(user_input, image2str(generated_image))]
|
145 |
+
else:
|
146 |
+
# If image generation failed, respond with an error message
|
147 |
+
history = history + [
|
148 |
+
(user_input, "Sorry, I could not generate an image.")
|
149 |
+
]
|
150 |
+
return history, history, gr.update(value=None)
|
151 |
+
"""
|
152 |
+
|
153 |
+
# Emu3-Chat model and processor
|
154 |
+
chat_model = AutoModelForCausalLM.from_pretrained(
|
155 |
+
EMU_CHAT_HUB,
|
156 |
+
device_map="cpu",
|
157 |
+
torch_dtype=torch.bfloat16,
|
158 |
+
attn_implementation="flash_attention_2",
|
159 |
+
trust_remote_code=True,
|
160 |
+
)
|
161 |
+
|
162 |
+
tokenizer = AutoTokenizer.from_pretrained(EMU_CHAT_HUB, trust_remote_code=True)
|
163 |
+
image_processor = AutoImageProcessor.from_pretrained(
|
164 |
+
VQ_HUB, trust_remote_code=True
|
165 |
+
)
|
166 |
+
image_tokenizer = AutoModel.from_pretrained(
|
167 |
+
VQ_HUB, device_map="cpu", trust_remote_code=True
|
168 |
+
).eval()
|
169 |
+
|
170 |
+
print(device)
|
171 |
+
chat_model.to(device)
|
172 |
+
image_tokenizer.to(device)
|
173 |
+
|
174 |
+
processor = Emu3Processor(
|
175 |
+
image_processor, image_tokenizer, tokenizer
|
176 |
+
)
|
177 |
+
|
178 |
@spaces.GPU
|
179 |
def vision_language_understanding(image, text):
|
180 |
inputs = processor(
|
|
|
212 |
# Append the user input and response to the history
|
213 |
history = history + [(image2str(user_image) + "<br>" + user_input, response)]
|
214 |
else:
|
215 |
+
history = history + [(user_input, "Sorry, please specify a valid image for vl understanding.")]
|
216 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
217 |
return history, history, gr.update(value=None)
|
218 |
|
219 |
def clear_input():
|