Spaces:
Running
on
Zero
Running
on
Zero
update
Browse files- app.py +41 -26
- eagle_vl/serve/chat_utils.py +42 -20
- eagle_vl/serve/inference.py +32 -13
app.py
CHANGED
@@ -39,7 +39,7 @@ logger = configure_logger()
|
|
39 |
|
40 |
def parse_args():
|
41 |
parser = argparse.ArgumentParser()
|
42 |
-
parser.add_argument("--model", type=str, default="
|
43 |
parser.add_argument(
|
44 |
"--local-path",
|
45 |
type=str,
|
@@ -57,7 +57,7 @@ def fetch_model(model_name: str):
|
|
57 |
if args.local_path:
|
58 |
model_path = args.local_path
|
59 |
else:
|
60 |
-
model_path = f"
|
61 |
|
62 |
if model_name in DEPLOY_MODELS:
|
63 |
model_info = DEPLOY_MODELS[model_name]
|
@@ -100,6 +100,7 @@ def predict(
|
|
100 |
temperature,
|
101 |
max_length_tokens,
|
102 |
max_context_length_tokens,
|
|
|
103 |
chunk_size: int = 512,
|
104 |
):
|
105 |
"""
|
@@ -116,18 +117,7 @@ def predict(
|
|
116 |
max_context_length_tokens (int): The max context length tokens.
|
117 |
chunk_size (int): The chunk size.
|
118 |
"""
|
119 |
-
|
120 |
-
try:
|
121 |
-
logger.info("fetching model")
|
122 |
-
model, processor = fetch_model(args.model)
|
123 |
-
logger.info("model fetched")
|
124 |
-
if text == "":
|
125 |
-
yield chatbot, history, "Empty context."
|
126 |
-
return
|
127 |
-
except KeyError:
|
128 |
-
logger.info("no model found")
|
129 |
-
yield [[text, "No Model Found"]], [], "No Model Found"
|
130 |
-
return
|
131 |
|
132 |
if images is None:
|
133 |
images = []
|
@@ -136,15 +126,33 @@ def predict(
|
|
136 |
pil_images = []
|
137 |
for img_or_file in images:
|
138 |
try:
|
|
|
139 |
# load as pil image
|
140 |
if isinstance(images, Image.Image):
|
141 |
pil_images.append(img_or_file)
|
142 |
-
|
143 |
-
|
144 |
-
|
|
|
|
|
|
|
145 |
except Exception as e:
|
146 |
print(f"Error loading image: {e}")
|
147 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
148 |
# generate prompt
|
149 |
conversation = generate_prompt_with_history(
|
150 |
text,
|
@@ -166,6 +174,7 @@ def predict(
|
|
166 |
max_length=max_length_tokens,
|
167 |
temperature=temperature,
|
168 |
top_p=top_p,
|
|
|
169 |
):
|
170 |
full_response += x
|
171 |
response = strip_stop_words(full_response, stop_words)
|
@@ -174,12 +183,12 @@ def predict(
|
|
174 |
|
175 |
yield gradio_chatbot_output, to_gradio_history(conversation), "Generating..."
|
176 |
|
177 |
-
if last_image is not None:
|
178 |
-
|
179 |
-
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
|
184 |
logger.info("flushed result to gradio")
|
185 |
|
@@ -202,6 +211,7 @@ def retry(
|
|
202 |
temperature,
|
203 |
max_length_tokens,
|
204 |
max_context_length_tokens,
|
|
|
205 |
chunk_size: int = 512,
|
206 |
):
|
207 |
"""
|
@@ -226,6 +236,7 @@ def retry(
|
|
226 |
temperature,
|
227 |
max_length_tokens,
|
228 |
max_context_length_tokens,
|
|
|
229 |
chunk_size,
|
230 |
)
|
231 |
|
@@ -265,9 +276,10 @@ def build_demo(args: argparse.Namespace) -> gr.Blocks:
|
|
265 |
with gr.Column():
|
266 |
# add note no more than 2 images once
|
267 |
# gr.Markdown("Note: you can upload no more than 2 images once")
|
268 |
-
upload_images = gr.Files(file_types=["image"], show_label=True)
|
269 |
gallery = gr.Gallery(columns=[3], height="200px", show_label=True)
|
270 |
upload_images.change(preview_images, inputs=upload_images, outputs=gallery)
|
|
|
271 |
# Parameter Setting Tab for control the generation parameters
|
272 |
with gr.Tab(label="Parameter Setting"):
|
273 |
top_p = gr.Slider(minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=True, label="Top-p")
|
@@ -280,7 +292,9 @@ def build_demo(args: argparse.Namespace) -> gr.Blocks:
|
|
280 |
max_context_length_tokens = gr.Slider(
|
281 |
minimum=512, maximum=16384, value=4096, step=64, interactive=True, label="Max Context Length Tokens"
|
282 |
)
|
283 |
-
|
|
|
|
|
284 |
show_images = gr.HTML(visible=False)
|
285 |
|
286 |
gr.Examples(
|
@@ -298,6 +312,7 @@ def build_demo(args: argparse.Namespace) -> gr.Blocks:
|
|
298 |
temperature,
|
299 |
max_length_tokens,
|
300 |
max_context_length_tokens,
|
|
|
301 |
]
|
302 |
output_widgets = [chatbot, history, status_display]
|
303 |
|
@@ -336,7 +351,7 @@ def main(args: argparse.Namespace):
|
|
336 |
demo.queue().launch(
|
337 |
favicon_path=favicon_path,
|
338 |
server_name=args.ip,
|
339 |
-
server_port=args.port
|
340 |
)
|
341 |
|
342 |
|
|
|
39 |
|
40 |
def parse_args():
|
41 |
parser = argparse.ArgumentParser()
|
42 |
+
parser.add_argument("--model", type=str, default="Eagle2.5-VL-8B-Preview")
|
43 |
parser.add_argument(
|
44 |
"--local-path",
|
45 |
type=str,
|
|
|
57 |
if args.local_path:
|
58 |
model_path = args.local_path
|
59 |
else:
|
60 |
+
model_path = f"NVEagle/{args.model}"
|
61 |
|
62 |
if model_name in DEPLOY_MODELS:
|
63 |
model_info = DEPLOY_MODELS[model_name]
|
|
|
100 |
temperature,
|
101 |
max_length_tokens,
|
102 |
max_context_length_tokens,
|
103 |
+
video_nframes,
|
104 |
chunk_size: int = 512,
|
105 |
):
|
106 |
"""
|
|
|
117 |
max_context_length_tokens (int): The max context length tokens.
|
118 |
chunk_size (int): The chunk size.
|
119 |
"""
|
120 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
121 |
|
122 |
if images is None:
|
123 |
images = []
|
|
|
126 |
pil_images = []
|
127 |
for img_or_file in images:
|
128 |
try:
|
129 |
+
logger.info(f"img_or_file: {img_or_file}")
|
130 |
# load as pil image
|
131 |
if isinstance(images, Image.Image):
|
132 |
pil_images.append(img_or_file)
|
133 |
+
elif isinstance(img_or_file, str):
|
134 |
+
if img_or_file.endswith((".mp4", ".mov", ".avi", ".webm")):
|
135 |
+
pil_images.append(img_or_file)
|
136 |
+
else:
|
137 |
+
image = Image.open(img_or_file.name).convert("RGB")
|
138 |
+
pil_images.append(image)
|
139 |
except Exception as e:
|
140 |
print(f"Error loading image: {e}")
|
141 |
|
142 |
+
|
143 |
+
print("running the prediction function")
|
144 |
+
try:
|
145 |
+
logger.info("fetching model")
|
146 |
+
model, processor = fetch_model(args.model)
|
147 |
+
logger.info("model fetched")
|
148 |
+
if text == "":
|
149 |
+
yield chatbot, history, "Empty context."
|
150 |
+
return
|
151 |
+
except KeyError:
|
152 |
+
logger.info("no model found")
|
153 |
+
yield [[text, "No Model Found"]], [], "No Model Found"
|
154 |
+
return
|
155 |
+
|
156 |
# generate prompt
|
157 |
conversation = generate_prompt_with_history(
|
158 |
text,
|
|
|
174 |
max_length=max_length_tokens,
|
175 |
temperature=temperature,
|
176 |
top_p=top_p,
|
177 |
+
video_nframes=video_nframes,
|
178 |
):
|
179 |
full_response += x
|
180 |
response = strip_stop_words(full_response, stop_words)
|
|
|
183 |
|
184 |
yield gradio_chatbot_output, to_gradio_history(conversation), "Generating..."
|
185 |
|
186 |
+
# if last_image is not None:
|
187 |
+
# vg_image = parse_ref_bbox(response, last_image)
|
188 |
+
# if vg_image is not None:
|
189 |
+
# vg_base64 = pil_to_base64(vg_image, "vg", max_size=800, min_size=400)
|
190 |
+
# gradio_chatbot_output[-1][1] += vg_base64
|
191 |
+
# yield gradio_chatbot_output, to_gradio_history(conversation), "Generating..."
|
192 |
|
193 |
logger.info("flushed result to gradio")
|
194 |
|
|
|
211 |
temperature,
|
212 |
max_length_tokens,
|
213 |
max_context_length_tokens,
|
214 |
+
video_nframes,
|
215 |
chunk_size: int = 512,
|
216 |
):
|
217 |
"""
|
|
|
236 |
temperature,
|
237 |
max_length_tokens,
|
238 |
max_context_length_tokens,
|
239 |
+
video_nframes,
|
240 |
chunk_size,
|
241 |
)
|
242 |
|
|
|
276 |
with gr.Column():
|
277 |
# add note no more than 2 images once
|
278 |
# gr.Markdown("Note: you can upload no more than 2 images once")
|
279 |
+
upload_images = gr.Files(file_types=["image", "video"], show_label=True)
|
280 |
gallery = gr.Gallery(columns=[3], height="200px", show_label=True)
|
281 |
upload_images.change(preview_images, inputs=upload_images, outputs=gallery)
|
282 |
+
|
283 |
# Parameter Setting Tab for control the generation parameters
|
284 |
with gr.Tab(label="Parameter Setting"):
|
285 |
top_p = gr.Slider(minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=True, label="Top-p")
|
|
|
292 |
max_context_length_tokens = gr.Slider(
|
293 |
minimum=512, maximum=16384, value=4096, step=64, interactive=True, label="Max Context Length Tokens"
|
294 |
)
|
295 |
+
video_nframes = gr.Slider(
|
296 |
+
minimum=1, maximum=128, value=16, step=1, interactive=True, label="Video Nframes"
|
297 |
+
)
|
298 |
show_images = gr.HTML(visible=False)
|
299 |
|
300 |
gr.Examples(
|
|
|
312 |
temperature,
|
313 |
max_length_tokens,
|
314 |
max_context_length_tokens,
|
315 |
+
video_nframes
|
316 |
]
|
317 |
output_widgets = [chatbot, history, status_display]
|
318 |
|
|
|
351 |
demo.queue().launch(
|
352 |
favicon_path=favicon_path,
|
353 |
server_name=args.ip,
|
354 |
+
server_port=args.port,
|
355 |
)
|
356 |
|
357 |
|
eagle_vl/serve/chat_utils.py
CHANGED
@@ -13,7 +13,7 @@ import gradio as gr
|
|
13 |
import torch
|
14 |
import os
|
15 |
from .utils import pil_to_base64
|
16 |
-
|
17 |
IMAGE_TOKEN = "<image>"
|
18 |
logger = logging.getLogger("gradio_logger")
|
19 |
|
@@ -324,6 +324,7 @@ def convert_conversation_to_prompts(conversation: Conversation):
|
|
324 |
Convert the conversation to prompts.
|
325 |
"""
|
326 |
conv_prompts = []
|
|
|
327 |
last_image = None
|
328 |
|
329 |
messages = conversation.messages
|
@@ -342,34 +343,55 @@ def convert_conversation_to_prompts(conversation: Conversation):
|
|
342 |
|
343 |
|
344 |
def to_gradio_chatbot(conversation: Conversation) -> list:
|
345 |
-
"""Convert the conversation to gradio chatbot format."""
|
346 |
ret = []
|
347 |
for i, (_, msg) in enumerate(conversation.messages[conversation.offset :]):
|
|
|
348 |
if i % 2 == 0:
|
349 |
-
if
|
350 |
-
|
351 |
-
|
352 |
-
|
353 |
-
|
354 |
-
|
355 |
-
|
356 |
-
|
357 |
-
|
358 |
-
|
359 |
-
|
360 |
-
|
361 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
362 |
)
|
363 |
else:
|
364 |
-
|
|
|
365 |
|
366 |
-
|
367 |
-
|
368 |
-
|
369 |
-
|
|
|
370 |
|
|
|
371 |
ret.append([msg, None])
|
372 |
else:
|
|
|
373 |
ret[-1][-1] = msg
|
374 |
return ret
|
375 |
|
|
|
13 |
import torch
|
14 |
import os
|
15 |
from .utils import pil_to_base64
|
16 |
+
import mimetypes
|
17 |
IMAGE_TOKEN = "<image>"
|
18 |
logger = logging.getLogger("gradio_logger")
|
19 |
|
|
|
324 |
Convert the conversation to prompts.
|
325 |
"""
|
326 |
conv_prompts = []
|
327 |
+
|
328 |
last_image = None
|
329 |
|
330 |
messages = conversation.messages
|
|
|
343 |
|
344 |
|
345 |
def to_gradio_chatbot(conversation: Conversation) -> list:
|
346 |
+
"""Convert the conversation to gradio chatbot format, supporting images and video."""
|
347 |
ret = []
|
348 |
for i, (_, msg) in enumerate(conversation.messages[conversation.offset :]):
|
349 |
+
# User message
|
350 |
if i % 2 == 0:
|
351 |
+
if isinstance(msg, tuple):
|
352 |
+
msg_text, media = copy.deepcopy(msg)
|
353 |
+
media_str = ""
|
354 |
+
|
355 |
+
# Handle list of media items
|
356 |
+
if isinstance(media, list):
|
357 |
+
items = media
|
358 |
+
else:
|
359 |
+
items = [media]
|
360 |
+
|
361 |
+
for j, item in enumerate(items):
|
362 |
+
# If string path, determine type
|
363 |
+
if isinstance(item, str):
|
364 |
+
mime, _ = mimetypes.guess_type(item)
|
365 |
+
with open(item, "rb") as f:
|
366 |
+
data = f.read()
|
367 |
+
b64 = base64.b64encode(data).decode()
|
368 |
+
|
369 |
+
if mime and mime.startswith("image/"):
|
370 |
+
media_str += (
|
371 |
+
f'<img src="data:{mime};base64,{b64}" '
|
372 |
+
f'alt="user upload image_{j}" '
|
373 |
+
f'style="max-width:300px;height:auto;" />'
|
374 |
+
)
|
375 |
+
elif mime and mime.startswith("video/"):
|
376 |
+
media_str += (
|
377 |
+
f'<video controls '
|
378 |
+
f'style="max-width:300px;height:auto;" '
|
379 |
+
f'src="data:{mime};base64,{b64}"></video>'
|
380 |
)
|
381 |
else:
|
382 |
+
# Fallback to link
|
383 |
+
media_str += f'<a href="{item}" target="_blank">{item}</a>'
|
384 |
|
385 |
+
# If PIL image
|
386 |
+
else:
|
387 |
+
media_str += pil_to_base64(item, f"user upload image_{j}", max_size=800, min_size=400)
|
388 |
+
|
389 |
+
msg = media_str + msg_text
|
390 |
|
391 |
+
# Append user side
|
392 |
ret.append([msg, None])
|
393 |
else:
|
394 |
+
# Assistant side, fill previous tuple
|
395 |
ret[-1][-1] = msg
|
396 |
return ret
|
397 |
|
eagle_vl/serve/inference.py
CHANGED
@@ -12,7 +12,7 @@ from transformers import (
|
|
12 |
StoppingCriteriaList,
|
13 |
TextIteratorStreamer,
|
14 |
)
|
15 |
-
|
16 |
from .chat_utils import Conversation, get_conv_template
|
17 |
|
18 |
logger = logging.getLogger(__name__)
|
@@ -91,6 +91,7 @@ class StoppingCriteriaSub(StoppingCriteria):
|
|
91 |
def preprocess(
|
92 |
messages: list[dict],
|
93 |
processor,
|
|
|
94 |
):
|
95 |
"""
|
96 |
Build messages from the conversations and images.
|
@@ -110,12 +111,28 @@ def preprocess(
|
|
110 |
if "images" in message:
|
111 |
per_round_images = message["images"]
|
112 |
for image in per_round_images:
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
119 |
if 'content' in message:
|
120 |
record["content"].append(
|
121 |
{
|
@@ -148,12 +165,12 @@ def preprocess(
|
|
148 |
formatted_answer.count(processor.image_token) == 0
|
149 |
), f"there should be no {processor.image_token} in the assistant's reply, but got {messages}"
|
150 |
|
151 |
-
print(f"messages = {results}")
|
152 |
text = processor.apply_chat_template(results, add_generation_prompt=False)
|
153 |
-
print(f"raw text = {text}")
|
|
|
|
|
154 |
|
155 |
-
image_inputs, video_inputs = processor.process_vision_info(results)
|
156 |
-
|
157 |
inputs = processor(
|
158 |
images=image_inputs,
|
159 |
videos=video_inputs,
|
@@ -161,6 +178,7 @@ def preprocess(
|
|
161 |
return_tensors="pt",
|
162 |
padding=True,
|
163 |
truncation=True,
|
|
|
164 |
)
|
165 |
return inputs
|
166 |
|
@@ -176,10 +194,11 @@ def eagle_vl_generate(
|
|
176 |
temperature: float = 1.0,
|
177 |
top_p: float = 1.0,
|
178 |
chunk_size: int = -1,
|
|
|
179 |
):
|
180 |
# convert conversation to inputs
|
181 |
print(f"conversations = {conversations}")
|
182 |
-
inputs = preprocess(conversations, processor=processor)
|
183 |
inputs = inputs.to(model.device)
|
184 |
|
185 |
return generate(
|
@@ -202,7 +221,7 @@ def generate(
|
|
202 |
temperature: float = 0,
|
203 |
top_p: float = 0.95,
|
204 |
stop_words: List[str] = [],
|
205 |
-
chunk_size: int = -1
|
206 |
):
|
207 |
"""Stream the text output from the multimodality model with prompt and image inputs."""
|
208 |
tokenizer = processor.tokenizer
|
|
|
12 |
StoppingCriteriaList,
|
13 |
TextIteratorStreamer,
|
14 |
)
|
15 |
+
from PIL import Image
|
16 |
from .chat_utils import Conversation, get_conv_template
|
17 |
|
18 |
logger = logging.getLogger(__name__)
|
|
|
91 |
def preprocess(
|
92 |
messages: list[dict],
|
93 |
processor,
|
94 |
+
video_nframes: int = 16,
|
95 |
):
|
96 |
"""
|
97 |
Build messages from the conversations and images.
|
|
|
111 |
if "images" in message:
|
112 |
per_round_images = message["images"]
|
113 |
for image in per_round_images:
|
114 |
+
if isinstance(image, Image.Image):
|
115 |
+
record["content"].append(
|
116 |
+
{
|
117 |
+
"type": "image",
|
118 |
+
"image": image,
|
119 |
+
}
|
120 |
+
)
|
121 |
+
elif isinstance(image, str) and image.endswith((".jpeg", ".jpg", ".png", ".gif")):
|
122 |
+
record["content"].append(
|
123 |
+
{
|
124 |
+
"type": "image",
|
125 |
+
"image": image,
|
126 |
+
}
|
127 |
+
)
|
128 |
+
elif isinstance(image, str) and image.endswith((".mp4", ".mov", ".avi", ".webm")):
|
129 |
+
record["content"].append(
|
130 |
+
{
|
131 |
+
"type": "video",
|
132 |
+
"video": image,
|
133 |
+
"nframes": video_nframes,
|
134 |
+
}
|
135 |
+
)
|
136 |
if 'content' in message:
|
137 |
record["content"].append(
|
138 |
{
|
|
|
165 |
formatted_answer.count(processor.image_token) == 0
|
166 |
), f"there should be no {processor.image_token} in the assistant's reply, but got {messages}"
|
167 |
|
168 |
+
# print(f"messages = {results}")
|
169 |
text = processor.apply_chat_template(results, add_generation_prompt=False)
|
170 |
+
# print(f"raw text = {text}")
|
171 |
+
|
172 |
+
image_inputs, video_inputs, video_kwargs = processor.process_vision_info(results, return_video_kwargs=True)
|
173 |
|
|
|
|
|
174 |
inputs = processor(
|
175 |
images=image_inputs,
|
176 |
videos=video_inputs,
|
|
|
178 |
return_tensors="pt",
|
179 |
padding=True,
|
180 |
truncation=True,
|
181 |
+
videos_kwargs=video_kwargs,
|
182 |
)
|
183 |
return inputs
|
184 |
|
|
|
194 |
temperature: float = 1.0,
|
195 |
top_p: float = 1.0,
|
196 |
chunk_size: int = -1,
|
197 |
+
video_nframes: int = 16,
|
198 |
):
|
199 |
# convert conversation to inputs
|
200 |
print(f"conversations = {conversations}")
|
201 |
+
inputs = preprocess(conversations, processor=processor, video_nframes=video_nframes)
|
202 |
inputs = inputs.to(model.device)
|
203 |
|
204 |
return generate(
|
|
|
221 |
temperature: float = 0,
|
222 |
top_p: float = 0.95,
|
223 |
stop_words: List[str] = [],
|
224 |
+
chunk_size: int = -1
|
225 |
):
|
226 |
"""Stream the text output from the multimodality model with prompt and image inputs."""
|
227 |
tokenizer = processor.tokenizer
|