Spaces:
Running
on
Zero
Running
on
Zero
momergul
commited on
Commit
•
ab71bac
1
Parent(s):
d1a5104
Reverted to old form
Browse files
app.py
CHANGED
@@ -8,45 +8,7 @@ from typing import List, Tuple
|
|
8 |
|
9 |
from config_generator import generate_complete_game
|
10 |
from dataset import get_processor, joint_speaker_input, joint_listener_input, get_index_to_token
|
11 |
-
|
12 |
-
import torch
|
13 |
-
import transformers
|
14 |
-
from transformers import Idefics2ForConditionalGeneration
|
15 |
-
from peft import LoraConfig, get_peft_model
|
16 |
-
from joint_inference import IdeficsJointInferenceModel
|
17 |
-
|
18 |
-
# Initialize the model globally
|
19 |
-
repo = 'lil-lab/cogen'
|
20 |
-
checkpoint = "HuggingFaceM4/idefics2-8b"
|
21 |
-
model = Idefics2ForConditionalGeneration.from_pretrained(checkpoint, torch_dtype=torch.bfloat16)
|
22 |
-
|
23 |
-
target_modules=r'(.*(vision_model|modality_projection|perceiver_resampler).*(out_proj|fc1|fc2|down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$)|(.*(k_proj|q_proj|v_proj).*$)'
|
24 |
-
lora_config = LoraConfig(
|
25 |
-
r=16, lora_alpha=8,
|
26 |
-
lora_dropout=0.1,
|
27 |
-
target_modules=target_modules,
|
28 |
-
init_lora_weights="gaussian"
|
29 |
-
)
|
30 |
-
model = get_peft_model(model, lora_config, adapter_name="initial")
|
31 |
-
model.load_adapter(repo, "initial", revision="r0_full")
|
32 |
-
|
33 |
-
# Add other adapter
|
34 |
-
new_targets = set()
|
35 |
-
for n, p in model.named_parameters():
|
36 |
-
if 'lora' in n:
|
37 |
-
new_targets.add(n[17:n.find('lora')-1])
|
38 |
-
new_targets = list(new_targets)
|
39 |
-
|
40 |
-
lora_config = LoraConfig(
|
41 |
-
r=16, lora_alpha=8,
|
42 |
-
lora_dropout=0.1,
|
43 |
-
target_modules=new_targets,
|
44 |
-
init_lora_weights="gaussian"
|
45 |
-
)
|
46 |
-
model.add_adapter('final', lora_config)
|
47 |
-
model.load_adapter(repo, "final", revision="r3_full")
|
48 |
-
model = IdeficsJointInferenceModel(0.5, 0, model=model).cuda()
|
49 |
-
model.eval()
|
50 |
|
51 |
css="""
|
52 |
.radio-group .wrap {
|
@@ -110,6 +72,7 @@ def get_model_response(
|
|
110 |
def get_speaker_response(model, images, input_tokens, attn_mask, image_attn_mask, label, image_paths, processor, img_dir, index_to_token, adapter_name):
|
111 |
if model.model.active_adapter != adapter_name:
|
112 |
model.model.set_adapter(adapter_name)
|
|
|
113 |
with torch.no_grad():
|
114 |
captions, _, _, _, _ = model.generate(
|
115 |
images.cuda(), input_tokens.cuda(), attn_mask.cuda(), image_attn_mask.cuda(), label.cuda(),
|
@@ -124,6 +87,7 @@ def get_listener_response(model, images, l_input_tokens, l_attn_mask, l_image_at
|
|
124 |
s_input_tokens, s_attn_mask, s_image_attn_mask, s_target_mask, s_target_label, image_paths, adapter_name):
|
125 |
if model.model.active_adapter != adapter_name:
|
126 |
model.model.set_adapter(adapter_name)
|
|
|
127 |
with torch.no_grad():
|
128 |
_, _, joint_log_probs = model.comprehension_side([
|
129 |
images.cuda(), l_input_tokens.cuda(), l_attn_mask.cuda(), l_image_attn_mask.cuda(), index_to_token,
|
@@ -155,7 +119,7 @@ def initialize_interaction(model_iteration):
|
|
155 |
|
156 |
return new_history
|
157 |
|
158 |
-
def progress_game(user_message, processor, index_to_token, current_state):
|
159 |
# First get the game state
|
160 |
turn = current_state['turn']
|
161 |
image_role_pairs = current_state['image_role_pairs']
|
@@ -293,6 +257,7 @@ def create_app():
|
|
293 |
)
|
294 |
|
295 |
send_btn = gr.Button("Send", interactive=False)
|
|
|
296 |
processor = get_processor()
|
297 |
index_to_token = get_index_to_token()
|
298 |
|
@@ -316,6 +281,7 @@ def create_app():
|
|
316 |
gr.update(interactive=not human_listener), gr.update(interactive=human_listener), gr.update(interactive=True), gr.update(interactive=False), current_history
|
317 |
|
318 |
def send_message(message, radio_choice, current_state):
|
|
|
319 |
nonlocal processor
|
320 |
nonlocal index_to_token
|
321 |
|
@@ -326,7 +292,7 @@ def create_app():
|
|
326 |
|
327 |
# Regular game progress
|
328 |
user_output = message if radio_choice is None else radio_choice
|
329 |
-
images, conversation, role, turn, acc_message, current_state = progress_game(user_output, processor, index_to_token, current_state)
|
330 |
human_listener = role == "Listener"
|
331 |
return [(f"tangram_pngs/{img}", f"Image {i+1}") for i, img in enumerate(images)], "\n".join(conversation), role, turn, \
|
332 |
acc_message, gr.update(interactive=not human_listener, value=""), gr.update(interactive=human_listener, value=None), \
|
|
|
8 |
|
9 |
from config_generator import generate_complete_game
|
10 |
from dataset import get_processor, joint_speaker_input, joint_listener_input, get_index_to_token
|
11 |
+
from models import get_model
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
css="""
|
14 |
.radio-group .wrap {
|
|
|
72 |
def get_speaker_response(model, images, input_tokens, attn_mask, image_attn_mask, label, image_paths, processor, img_dir, index_to_token, adapter_name):
|
73 |
if model.model.active_adapter != adapter_name:
|
74 |
model.model.set_adapter(adapter_name)
|
75 |
+
model = model.cuda()
|
76 |
with torch.no_grad():
|
77 |
captions, _, _, _, _ = model.generate(
|
78 |
images.cuda(), input_tokens.cuda(), attn_mask.cuda(), image_attn_mask.cuda(), label.cuda(),
|
|
|
87 |
s_input_tokens, s_attn_mask, s_image_attn_mask, s_target_mask, s_target_label, image_paths, adapter_name):
|
88 |
if model.model.active_adapter != adapter_name:
|
89 |
model.model.set_adapter(adapter_name)
|
90 |
+
model = model.cuda()
|
91 |
with torch.no_grad():
|
92 |
_, _, joint_log_probs = model.comprehension_side([
|
93 |
images.cuda(), l_input_tokens.cuda(), l_attn_mask.cuda(), l_image_attn_mask.cuda(), index_to_token,
|
|
|
119 |
|
120 |
return new_history
|
121 |
|
122 |
+
def progress_game(user_message, model, processor, index_to_token, current_state):
|
123 |
# First get the game state
|
124 |
turn = current_state['turn']
|
125 |
image_role_pairs = current_state['image_role_pairs']
|
|
|
257 |
)
|
258 |
|
259 |
send_btn = gr.Button("Send", interactive=False)
|
260 |
+
model = get_model()
|
261 |
processor = get_processor()
|
262 |
index_to_token = get_index_to_token()
|
263 |
|
|
|
281 |
gr.update(interactive=not human_listener), gr.update(interactive=human_listener), gr.update(interactive=True), gr.update(interactive=False), current_history
|
282 |
|
283 |
def send_message(message, radio_choice, current_state):
|
284 |
+
nonlocal model
|
285 |
nonlocal processor
|
286 |
nonlocal index_to_token
|
287 |
|
|
|
292 |
|
293 |
# Regular game progress
|
294 |
user_output = message if radio_choice is None else radio_choice
|
295 |
+
images, conversation, role, turn, acc_message, current_state = progress_game(user_output, model, processor, index_to_token, current_state)
|
296 |
human_listener = role == "Listener"
|
297 |
return [(f"tangram_pngs/{img}", f"Image {i+1}") for i, img in enumerate(images)], "\n".join(conversation), role, turn, \
|
298 |
acc_message, gr.update(interactive=not human_listener, value=""), gr.update(interactive=human_listener, value=None), \
|