Spaces:
Running
Running
Commit
·
d2662cc
1
Parent(s):
4a3f789
fix
Browse files- vms/config.py +10 -10
- vms/ui/app_ui.py +30 -2
- vms/ui/monitoring/services/monitoring.py +1 -0
- vms/ui/monitoring/tabs/general_tab.py +1 -1
- vms/ui/project/services/captioning.py +1 -0
- vms/ui/project/services/previewing.py +123 -56
- vms/ui/project/services/splitting.py +1 -0
- vms/ui/project/services/training.py +26 -7
- vms/ui/project/tabs/preview_tab.py +147 -106
- vms/ui/project/tabs/train_tab.py +112 -48
vms/config.py
CHANGED
@@ -61,7 +61,7 @@ JPEG_QUALITY = int(os.environ.get('JPEG_QUALITY', '97'))
|
|
61 |
MODEL_TYPES = {
|
62 |
"HunyuanVideo": "hunyuan_video",
|
63 |
"LTX-Video": "ltx_video",
|
64 |
-
"Wan
|
65 |
}
|
66 |
|
67 |
# Training types
|
@@ -70,8 +70,8 @@ TRAINING_TYPES = {
|
|
70 |
"Full Finetune": "full-finetune"
|
71 |
}
|
72 |
|
73 |
-
# Model
|
74 |
-
|
75 |
"wan": {
|
76 |
"Wan-AI/Wan2.1-T2V-1.3B-Diffusers": {
|
77 |
"name": "Wan 2.1 T2V 1.3B (text-only, smaller)",
|
@@ -342,7 +342,7 @@ class TrainingConfig:
|
|
342 |
|
343 |
# Optional arguments follow
|
344 |
revision: Optional[str] = None
|
345 |
-
|
346 |
cache_dir: Optional[str] = None
|
347 |
|
348 |
# Dataset arguments
|
@@ -415,7 +415,7 @@ class TrainingConfig:
|
|
415 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
416 |
lr=2e-5,
|
417 |
gradient_checkpointing=True,
|
418 |
-
id_token=
|
419 |
gradient_accumulation_steps=1,
|
420 |
lora_rank=DEFAULT_LORA_RANK,
|
421 |
lora_alpha=DEFAULT_LORA_ALPHA,
|
@@ -437,7 +437,7 @@ class TrainingConfig:
|
|
437 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
438 |
lr=DEFAULT_LEARNING_RATE,
|
439 |
gradient_checkpointing=True,
|
440 |
-
id_token=
|
441 |
gradient_accumulation_steps=4,
|
442 |
lora_rank=DEFAULT_LORA_RANK,
|
443 |
lora_alpha=DEFAULT_LORA_ALPHA,
|
@@ -459,7 +459,7 @@ class TrainingConfig:
|
|
459 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
460 |
lr=1e-5,
|
461 |
gradient_checkpointing=True,
|
462 |
-
id_token=
|
463 |
gradient_accumulation_steps=1,
|
464 |
video_resolution_buckets=buckets or SMALL_TRAINING_BUCKETS,
|
465 |
caption_dropout_p=DEFAULT_CAPTION_DROPOUT_P,
|
@@ -479,7 +479,7 @@ class TrainingConfig:
|
|
479 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
480 |
lr=5e-5,
|
481 |
gradient_checkpointing=True,
|
482 |
-
id_token=None,
|
483 |
gradient_accumulation_steps=1,
|
484 |
lora_rank=32,
|
485 |
lora_alpha=32,
|
@@ -502,8 +502,8 @@ class TrainingConfig:
|
|
502 |
args.extend(["--pretrained_model_name_or_path", self.pretrained_model_name_or_path])
|
503 |
if self.revision:
|
504 |
args.extend(["--revision", self.revision])
|
505 |
-
if self.
|
506 |
-
args.extend(["--variant", self.
|
507 |
if self.cache_dir:
|
508 |
args.extend(["--cache_dir", self.cache_dir])
|
509 |
|
|
|
61 |
MODEL_TYPES = {
|
62 |
"HunyuanVideo": "hunyuan_video",
|
63 |
"LTX-Video": "ltx_video",
|
64 |
+
"Wan": "wan"
|
65 |
}
|
66 |
|
67 |
# Training types
|
|
|
70 |
"Full Finetune": "full-finetune"
|
71 |
}
|
72 |
|
73 |
+
# Model versions for each model type
|
74 |
+
MODEL_VERSIONS = {
|
75 |
"wan": {
|
76 |
"Wan-AI/Wan2.1-T2V-1.3B-Diffusers": {
|
77 |
"name": "Wan 2.1 T2V 1.3B (text-only, smaller)",
|
|
|
342 |
|
343 |
# Optional arguments follow
|
344 |
revision: Optional[str] = None
|
345 |
+
version: Optional[str] = None
|
346 |
cache_dir: Optional[str] = None
|
347 |
|
348 |
# Dataset arguments
|
|
|
415 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
416 |
lr=2e-5,
|
417 |
gradient_checkpointing=True,
|
418 |
+
id_token=None,
|
419 |
gradient_accumulation_steps=1,
|
420 |
lora_rank=DEFAULT_LORA_RANK,
|
421 |
lora_alpha=DEFAULT_LORA_ALPHA,
|
|
|
437 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
438 |
lr=DEFAULT_LEARNING_RATE,
|
439 |
gradient_checkpointing=True,
|
440 |
+
id_token=None,
|
441 |
gradient_accumulation_steps=4,
|
442 |
lora_rank=DEFAULT_LORA_RANK,
|
443 |
lora_alpha=DEFAULT_LORA_ALPHA,
|
|
|
459 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
460 |
lr=1e-5,
|
461 |
gradient_checkpointing=True,
|
462 |
+
id_token=None,
|
463 |
gradient_accumulation_steps=1,
|
464 |
video_resolution_buckets=buckets or SMALL_TRAINING_BUCKETS,
|
465 |
caption_dropout_p=DEFAULT_CAPTION_DROPOUT_P,
|
|
|
479 |
train_steps=DEFAULT_NB_TRAINING_STEPS,
|
480 |
lr=5e-5,
|
481 |
gradient_checkpointing=True,
|
482 |
+
id_token=None,
|
483 |
gradient_accumulation_steps=1,
|
484 |
lora_rank=32,
|
485 |
lora_alpha=32,
|
|
|
502 |
args.extend(["--pretrained_model_name_or_path", self.pretrained_model_name_or_path])
|
503 |
if self.revision:
|
504 |
args.extend(["--revision", self.revision])
|
505 |
+
if self.version:
|
506 |
+
args.extend(["--variant", self.version])
|
507 |
if self.cache_dir:
|
508 |
args.extend(["--cache_dir", self.cache_dir])
|
509 |
|
vms/ui/app_ui.py
CHANGED
@@ -8,7 +8,7 @@ from typing import Any, Optional, Dict, List, Union, Tuple
|
|
8 |
from vms.config import (
|
9 |
STORAGE_PATH, VIDEOS_TO_SPLIT_PATH, STAGING_PATH, OUTPUT_PATH,
|
10 |
TRAINING_PATH, LOG_FILE_PATH, TRAINING_PRESETS, TRAINING_VIDEOS_PATH, MODEL_PATH, OUTPUT_PATH,
|
11 |
-
MODEL_TYPES, SMALL_TRAINING_BUCKETS, TRAINING_TYPES,
|
12 |
DEFAULT_NB_TRAINING_STEPS, DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS,
|
13 |
DEFAULT_BATCH_SIZE, DEFAULT_CAPTION_DROPOUT_P,
|
14 |
DEFAULT_LEARNING_RATE,
|
@@ -220,6 +220,7 @@ class AppUI:
|
|
220 |
self.project_tabs["train_tab"].components["pause_resume_btn"],
|
221 |
self.project_tabs["train_tab"].components["training_preset"],
|
222 |
self.project_tabs["train_tab"].components["model_type"],
|
|
|
223 |
self.project_tabs["train_tab"].components["training_type"],
|
224 |
self.project_tabs["train_tab"].components["lora_rank"],
|
225 |
self.project_tabs["train_tab"].components["lora_alpha"],
|
@@ -378,6 +379,20 @@ class AppUI:
|
|
378 |
model_type_val = list(MODEL_TYPES.keys())[0]
|
379 |
logger.warning(f"Invalid model type '{model_type_val}', using default: {model_type_val}")
|
380 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
381 |
# Ensure training_type is a valid display name
|
382 |
training_type_val = ui_state.get("training_type", list(TRAINING_TYPES.keys())[0])
|
383 |
if training_type_val not in TRAINING_TYPES:
|
@@ -436,6 +451,7 @@ class AppUI:
|
|
436 |
delete_checkpoints_btn,
|
437 |
training_preset,
|
438 |
model_type_val,
|
|
|
439 |
training_type_val,
|
440 |
lora_rank_val,
|
441 |
lora_alpha_val,
|
@@ -453,10 +469,22 @@ class AppUI:
|
|
453 |
"""Initialize UI components from saved state"""
|
454 |
ui_state = self.load_ui_values()
|
455 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
456 |
# Return values in order matching the outputs in app.load
|
457 |
return (
|
458 |
ui_state.get("training_preset", list(TRAINING_PRESETS.keys())[0]),
|
459 |
-
|
|
|
460 |
ui_state.get("training_type", list(TRAINING_TYPES.keys())[0]),
|
461 |
ui_state.get("lora_rank", DEFAULT_LORA_RANK_STR),
|
462 |
ui_state.get("lora_alpha", DEFAULT_LORA_ALPHA_STR),
|
|
|
8 |
from vms.config import (
|
9 |
STORAGE_PATH, VIDEOS_TO_SPLIT_PATH, STAGING_PATH, OUTPUT_PATH,
|
10 |
TRAINING_PATH, LOG_FILE_PATH, TRAINING_PRESETS, TRAINING_VIDEOS_PATH, MODEL_PATH, OUTPUT_PATH,
|
11 |
+
MODEL_TYPES, SMALL_TRAINING_BUCKETS, TRAINING_TYPES, MODEL_VERSIONS,
|
12 |
DEFAULT_NB_TRAINING_STEPS, DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS,
|
13 |
DEFAULT_BATCH_SIZE, DEFAULT_CAPTION_DROPOUT_P,
|
14 |
DEFAULT_LEARNING_RATE,
|
|
|
220 |
self.project_tabs["train_tab"].components["pause_resume_btn"],
|
221 |
self.project_tabs["train_tab"].components["training_preset"],
|
222 |
self.project_tabs["train_tab"].components["model_type"],
|
223 |
+
self.project_tabs["train_tab"].components["model_version"],
|
224 |
self.project_tabs["train_tab"].components["training_type"],
|
225 |
self.project_tabs["train_tab"].components["lora_rank"],
|
226 |
self.project_tabs["train_tab"].components["lora_alpha"],
|
|
|
379 |
model_type_val = list(MODEL_TYPES.keys())[0]
|
380 |
logger.warning(f"Invalid model type '{model_type_val}', using default: {model_type_val}")
|
381 |
|
382 |
+
# Get model_version value
|
383 |
+
model_version_val = ""
|
384 |
+
# First get the internal model type for the currently selected model
|
385 |
+
model_internal_type = MODEL_TYPES.get(model_type_val)
|
386 |
+
if model_internal_type and model_internal_type in MODEL_VERSIONS:
|
387 |
+
# If there's a saved model_version and it's valid for this model type
|
388 |
+
if "model_version" in ui_state and ui_state["model_version"] in MODEL_VERSIONS.get(model_internal_type, {}):
|
389 |
+
model_version_val = ui_state["model_version"]
|
390 |
+
else:
|
391 |
+
# Otherwise use the first available version
|
392 |
+
versions = list(MODEL_VERSIONS.get(model_internal_type, {}).keys())
|
393 |
+
if versions:
|
394 |
+
model_version_val = versions[0]
|
395 |
+
|
396 |
# Ensure training_type is a valid display name
|
397 |
training_type_val = ui_state.get("training_type", list(TRAINING_TYPES.keys())[0])
|
398 |
if training_type_val not in TRAINING_TYPES:
|
|
|
451 |
delete_checkpoints_btn,
|
452 |
training_preset,
|
453 |
model_type_val,
|
454 |
+
model_version_val,
|
455 |
training_type_val,
|
456 |
lora_rank_val,
|
457 |
lora_alpha_val,
|
|
|
469 |
"""Initialize UI components from saved state"""
|
470 |
ui_state = self.load_ui_values()
|
471 |
|
472 |
+
# Get model type and determine the default model version if not specified
|
473 |
+
model_type = ui_state.get("model_type", list(MODEL_TYPES.keys())[0])
|
474 |
+
model_internal_type = MODEL_TYPES.get(model_type)
|
475 |
+
|
476 |
+
# Get model_version, defaulting to first available version if not set
|
477 |
+
model_version = ui_state.get("model_version", "")
|
478 |
+
if not model_version and model_internal_type and model_internal_type in MODEL_VERSIONS:
|
479 |
+
versions = list(MODEL_VERSIONS.get(model_internal_type, {}).keys())
|
480 |
+
if versions:
|
481 |
+
model_version = versions[0]
|
482 |
+
|
483 |
# Return values in order matching the outputs in app.load
|
484 |
return (
|
485 |
ui_state.get("training_preset", list(TRAINING_PRESETS.keys())[0]),
|
486 |
+
model_type,
|
487 |
+
model_version,
|
488 |
ui_state.get("training_type", list(TRAINING_TYPES.keys())[0]),
|
489 |
ui_state.get("lora_rank", DEFAULT_LORA_RANK_STR),
|
490 |
ui_state.get("lora_alpha", DEFAULT_LORA_ALPHA_STR),
|
vms/ui/monitoring/services/monitoring.py
CHANGED
@@ -22,6 +22,7 @@ import matplotlib.pyplot as plt
|
|
22 |
import numpy as np
|
23 |
|
24 |
logger = logging.getLogger(__name__)
|
|
|
25 |
|
26 |
class MonitoringService:
|
27 |
"""Service for monitoring system resources and performance"""
|
|
|
22 |
import numpy as np
|
23 |
|
24 |
logger = logging.getLogger(__name__)
|
25 |
+
logger.setLevel(logging.INFO)
|
26 |
|
27 |
class MonitoringService:
|
28 |
"""Service for monitoring system resources and performance"""
|
vms/ui/monitoring/tabs/general_tab.py
CHANGED
@@ -17,7 +17,7 @@ from vms.config import STORAGE_PATH
|
|
17 |
from vms.ui.monitoring.utils import get_folder_size, human_readable_size
|
18 |
|
19 |
logger = logging.getLogger(__name__)
|
20 |
-
|
21 |
|
22 |
class GeneralTab(BaseTab):
|
23 |
"""Monitor tab for general system resource monitoring"""
|
|
|
17 |
from vms.ui.monitoring.utils import get_folder_size, human_readable_size
|
18 |
|
19 |
logger = logging.getLogger(__name__)
|
20 |
+
logger.setLevel(logging.INFO)
|
21 |
|
22 |
class GeneralTab(BaseTab):
|
23 |
"""Monitor tab for general system resource monitoring"""
|
vms/ui/project/services/captioning.py
CHANGED
@@ -21,6 +21,7 @@ from vms.config import TRAINING_VIDEOS_PATH, STAGING_PATH, PRELOAD_CAPTIONING_MO
|
|
21 |
from vms.utils import extract_scene_info, is_image_file, is_video_file, copy_files_to_training_dir, prepare_finetrainers_dataset
|
22 |
|
23 |
logger = logging.getLogger(__name__)
|
|
|
24 |
|
25 |
@dataclass
|
26 |
class CaptioningProgress:
|
|
|
21 |
from vms.utils import extract_scene_info, is_image_file, is_video_file, copy_files_to_training_dir, prepare_finetrainers_dataset
|
22 |
|
23 |
logger = logging.getLogger(__name__)
|
24 |
+
logger.setLevel(logging.INFO)
|
25 |
|
26 |
@dataclass
|
27 |
class CaptioningProgress:
|
vms/ui/project/services/previewing.py
CHANGED
@@ -6,17 +6,20 @@ Handles the video generation logic and model integration
|
|
6 |
|
7 |
import logging
|
8 |
import tempfile
|
|
|
|
|
9 |
from pathlib import Path
|
10 |
from typing import Dict, Any, List, Optional, Tuple, Callable
|
11 |
import time
|
12 |
|
13 |
from vms.config import (
|
14 |
OUTPUT_PATH, STORAGE_PATH, MODEL_TYPES, TRAINING_PATH,
|
15 |
-
DEFAULT_PROMPT_PREFIX,
|
16 |
)
|
17 |
from vms.utils import format_time
|
18 |
|
19 |
logger = logging.getLogger(__name__)
|
|
|
20 |
|
21 |
class PreviewingService:
|
22 |
"""Handles the video generation logic and model integration"""
|
@@ -48,14 +51,14 @@ class PreviewingService:
|
|
48 |
logger.error(f"Error finding LoRA weights: {e}")
|
49 |
return None
|
50 |
|
51 |
-
def
|
52 |
-
"""Get available model
|
53 |
-
return
|
54 |
|
55 |
def generate_video(
|
56 |
self,
|
57 |
model_type: str,
|
58 |
-
|
59 |
prompt: str,
|
60 |
negative_prompt: str,
|
61 |
prompt_prefix: str,
|
@@ -66,13 +69,15 @@ class PreviewingService:
|
|
66 |
flow_shift: float,
|
67 |
lora_weight: float,
|
68 |
inference_steps: int,
|
69 |
-
|
70 |
-
|
|
|
71 |
conditioning_image: Optional[str] = None
|
72 |
) -> Tuple[Optional[str], str, str]:
|
73 |
"""Generate a video using the trained model"""
|
74 |
try:
|
75 |
log_messages = []
|
|
|
76 |
|
77 |
def log(msg: str):
|
78 |
log_messages.append(msg)
|
@@ -102,33 +107,46 @@ class PreviewingService:
|
|
102 |
if not internal_model_type:
|
103 |
return None, f"Error: Invalid model type {model_type}", log(f"Error: Invalid model type {model_type}")
|
104 |
|
105 |
-
# Check if model
|
106 |
-
|
107 |
-
if
|
108 |
-
#
|
109 |
-
|
110 |
-
|
111 |
-
log(f"Warning:
|
|
|
|
|
|
|
|
|
112 |
else:
|
113 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
114 |
if internal_model_type == "wan":
|
115 |
-
|
116 |
elif internal_model_type == "ltx_video":
|
117 |
-
|
118 |
elif internal_model_type == "hunyuan_video":
|
119 |
-
|
120 |
-
log(f"
|
121 |
|
122 |
# Check if this is an image-to-video model but no image was provided
|
123 |
-
|
124 |
-
if
|
125 |
-
return None, "Error: This model requires a conditioning image", log("Error: This model
|
126 |
|
127 |
log(f"Generating video with model type: {internal_model_type}")
|
128 |
-
log(f"Using model
|
129 |
log(f"Using LoRA weights from: {lora_path}")
|
130 |
log(f"Resolution: {width}x{height}, Frames: {num_frames}, FPS: {fps}")
|
131 |
log(f"Guidance Scale: {guidance_scale}, Flow Shift: {flow_shift}, LoRA Weight: {lora_weight}")
|
|
|
132 |
log(f"Prompt: {full_prompt}")
|
133 |
log(f"Negative Prompt: {negative_prompt}")
|
134 |
|
@@ -137,22 +155,22 @@ class PreviewingService:
|
|
137 |
return self.generate_wan_video(
|
138 |
full_prompt, negative_prompt, width, height, num_frames,
|
139 |
guidance_scale, flow_shift, lora_path, lora_weight,
|
140 |
-
inference_steps, enable_cpu_offload, fps, log,
|
141 |
-
|
142 |
)
|
143 |
elif internal_model_type == "ltx_video":
|
144 |
return self.generate_ltx_video(
|
145 |
full_prompt, negative_prompt, width, height, num_frames,
|
146 |
guidance_scale, flow_shift, lora_path, lora_weight,
|
147 |
-
inference_steps, enable_cpu_offload, fps, log,
|
148 |
-
|
149 |
)
|
150 |
elif internal_model_type == "hunyuan_video":
|
151 |
return self.generate_hunyuan_video(
|
152 |
full_prompt, negative_prompt, width, height, num_frames,
|
153 |
guidance_scale, flow_shift, lora_path, lora_weight,
|
154 |
-
inference_steps, enable_cpu_offload, fps, log,
|
155 |
-
|
156 |
)
|
157 |
else:
|
158 |
return None, f"Error: Unsupported model type {internal_model_type}", log(f"Error: Unsupported model type {internal_model_type}")
|
@@ -173,16 +191,18 @@ class PreviewingService:
|
|
173 |
lora_path: str,
|
174 |
lora_weight: float,
|
175 |
inference_steps: int,
|
176 |
-
|
177 |
-
|
178 |
-
|
179 |
-
|
|
|
180 |
conditioning_image: Optional[str] = None
|
181 |
) -> Tuple[Optional[str], str, str]:
|
182 |
"""Generate video using Wan model"""
|
183 |
|
184 |
try:
|
185 |
import torch
|
|
|
186 |
from diffusers import AutoencoderKLWan, WanPipeline
|
187 |
from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
|
188 |
from diffusers.utils import export_to_video
|
@@ -192,14 +212,26 @@ class PreviewingService:
|
|
192 |
start_time = torch.cuda.Event(enable_timing=True)
|
193 |
end_time = torch.cuda.Event(enable_timing=True)
|
194 |
|
195 |
-
|
196 |
log_fn("Importing Wan model components...")
|
197 |
|
198 |
-
|
199 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
200 |
|
201 |
-
log_fn(f"Loading
|
202 |
-
|
|
|
|
|
|
|
203 |
|
204 |
log_fn(f"Configuring scheduler with flow_shift={flow_shift}...")
|
205 |
pipe.scheduler = UniPCMultistepScheduler.from_config(
|
@@ -213,11 +245,13 @@ class PreviewingService:
|
|
213 |
if enable_cpu_offload:
|
214 |
log_fn("Enabling model CPU offload...")
|
215 |
pipe.enable_model_cpu_offload()
|
216 |
-
|
217 |
log_fn(f"Loading LoRA weights from {lora_path} with weight {lora_weight}...")
|
218 |
pipe.load_lora_weights(lora_path)
|
219 |
-
pipe.fuse_lora(lora_weight)
|
220 |
|
|
|
|
|
|
|
221 |
# Create temporary file for the output
|
222 |
with tempfile.NamedTemporaryFile(suffix='.mp4', delete=False) as temp_file:
|
223 |
output_path = temp_file.name
|
@@ -226,7 +260,7 @@ class PreviewingService:
|
|
226 |
start_time.record()
|
227 |
|
228 |
# Check if this is an image-to-video model
|
229 |
-
is_i2v = "I2V" in
|
230 |
|
231 |
if is_i2v and conditioning_image:
|
232 |
log_fn(f"Loading conditioning image from {conditioning_image}...")
|
@@ -243,6 +277,7 @@ class PreviewingService:
|
|
243 |
num_frames=num_frames,
|
244 |
guidance_scale=guidance_scale,
|
245 |
num_inference_steps=inference_steps,
|
|
|
246 |
).frames[0]
|
247 |
else:
|
248 |
log_fn("Generating video with text-only conditioning...")
|
@@ -254,6 +289,7 @@ class PreviewingService:
|
|
254 |
num_frames=num_frames,
|
255 |
guidance_scale=guidance_scale,
|
256 |
num_inference_steps=inference_steps,
|
|
|
257 |
).frames[0]
|
258 |
|
259 |
end_time.record()
|
@@ -274,11 +310,12 @@ class PreviewingService:
|
|
274 |
return output_path, "Video generated successfully!", log_fn(f"Generation completed in {format_time(generation_time)}")
|
275 |
|
276 |
except Exception as e:
|
|
|
277 |
log_fn(f"Error generating video with Wan: {str(e)}")
|
278 |
# Clean up CUDA memory
|
279 |
torch.cuda.empty_cache()
|
280 |
return None, f"Error: {str(e)}", log_fn(f"Exception occurred: {str(e)}")
|
281 |
-
|
282 |
def generate_ltx_video(
|
283 |
self,
|
284 |
prompt: str,
|
@@ -291,27 +328,41 @@ class PreviewingService:
|
|
291 |
lora_path: str,
|
292 |
lora_weight: float,
|
293 |
inference_steps: int,
|
294 |
-
|
295 |
-
|
296 |
-
|
297 |
-
|
|
|
298 |
conditioning_image: Optional[str] = None
|
299 |
) -> Tuple[Optional[str], str, str]:
|
300 |
"""Generate video using LTX model"""
|
301 |
|
302 |
try:
|
303 |
import torch
|
|
|
304 |
from diffusers import LTXPipeline
|
305 |
from diffusers.utils import export_to_video
|
306 |
from PIL import Image
|
307 |
|
308 |
start_time = torch.cuda.Event(enable_timing=True)
|
309 |
end_time = torch.cuda.Event(enable_timing=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
310 |
|
311 |
log_fn("Importing LTX model components...")
|
312 |
|
313 |
-
log_fn(f"Loading pipeline from {
|
314 |
-
pipe = LTXPipeline.from_pretrained(
|
315 |
|
316 |
log_fn("Moving pipeline to CUDA device...")
|
317 |
pipe.to("cuda")
|
@@ -342,6 +393,7 @@ class PreviewingService:
|
|
342 |
decode_timestep=0.03,
|
343 |
decode_noise_scale=0.025,
|
344 |
num_inference_steps=inference_steps,
|
|
|
345 |
).frames[0]
|
346 |
|
347 |
end_time.record()
|
@@ -379,10 +431,11 @@ class PreviewingService:
|
|
379 |
lora_path: str,
|
380 |
lora_weight: float,
|
381 |
inference_steps: int,
|
382 |
-
|
383 |
-
|
384 |
-
|
385 |
-
|
|
|
386 |
conditioning_image: Optional[str] = None
|
387 |
) -> Tuple[Optional[str], str, str]:
|
388 |
"""Generate video using HunyuanVideo model"""
|
@@ -390,24 +443,37 @@ class PreviewingService:
|
|
390 |
|
391 |
try:
|
392 |
import torch
|
|
|
393 |
from diffusers import HunyuanVideoPipeline, HunyuanVideoTransformer3DModel, AutoencoderKLHunyuanVideo
|
394 |
from diffusers.utils import export_to_video
|
395 |
|
396 |
start_time = torch.cuda.Event(enable_timing=True)
|
397 |
end_time = torch.cuda.Event(enable_timing=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
398 |
|
399 |
log_fn("Importing HunyuanVideo model components...")
|
400 |
|
401 |
-
log_fn(f"Loading transformer from {
|
402 |
transformer = HunyuanVideoTransformer3DModel.from_pretrained(
|
403 |
-
|
404 |
subfolder="transformer",
|
405 |
torch_dtype=torch.bfloat16
|
406 |
)
|
407 |
|
408 |
-
log_fn(f"Loading pipeline from {
|
409 |
pipe = HunyuanVideoPipeline.from_pretrained(
|
410 |
-
|
411 |
transformer=transformer,
|
412 |
torch_dtype=torch.float16
|
413 |
)
|
@@ -446,6 +512,7 @@ class PreviewingService:
|
|
446 |
guidance_scale=guidance_scale,
|
447 |
true_cfg_scale=1.0,
|
448 |
num_inference_steps=inference_steps,
|
|
|
449 |
).frames[0]
|
450 |
|
451 |
end_time.record()
|
|
|
6 |
|
7 |
import logging
|
8 |
import tempfile
|
9 |
+
import traceback
|
10 |
+
import random
|
11 |
from pathlib import Path
|
12 |
from typing import Dict, Any, List, Optional, Tuple, Callable
|
13 |
import time
|
14 |
|
15 |
from vms.config import (
|
16 |
OUTPUT_PATH, STORAGE_PATH, MODEL_TYPES, TRAINING_PATH,
|
17 |
+
DEFAULT_PROMPT_PREFIX, MODEL_VERSIONS
|
18 |
)
|
19 |
from vms.utils import format_time
|
20 |
|
21 |
logger = logging.getLogger(__name__)
|
22 |
+
logger.setLevel(logging.INFO)
|
23 |
|
24 |
class PreviewingService:
|
25 |
"""Handles the video generation logic and model integration"""
|
|
|
51 |
logger.error(f"Error finding LoRA weights: {e}")
|
52 |
return None
|
53 |
|
54 |
+
def get_model_versions(self, model_type: str) -> Dict[str, Dict[str, str]]:
|
55 |
+
"""Get available model versions for the given model type"""
|
56 |
+
return MODEL_VERSIONS.get(model_type, {})
|
57 |
|
58 |
def generate_video(
|
59 |
self,
|
60 |
model_type: str,
|
61 |
+
model_version: str,
|
62 |
prompt: str,
|
63 |
negative_prompt: str,
|
64 |
prompt_prefix: str,
|
|
|
69 |
flow_shift: float,
|
70 |
lora_weight: float,
|
71 |
inference_steps: int,
|
72 |
+
seed: int = -1,
|
73 |
+
enable_cpu_offload: bool = True,
|
74 |
+
fps: int = 16,
|
75 |
conditioning_image: Optional[str] = None
|
76 |
) -> Tuple[Optional[str], str, str]:
|
77 |
"""Generate a video using the trained model"""
|
78 |
try:
|
79 |
log_messages = []
|
80 |
+
print("generate_video")
|
81 |
|
82 |
def log(msg: str):
|
83 |
log_messages.append(msg)
|
|
|
107 |
if not internal_model_type:
|
108 |
return None, f"Error: Invalid model type {model_type}", log(f"Error: Invalid model type {model_type}")
|
109 |
|
110 |
+
# Check if model version is valid
|
111 |
+
# This section uses model_version directly from parameter
|
112 |
+
if model_version:
|
113 |
+
# Verify that the specified model_version exists in our versions
|
114 |
+
versions = self.get_model_versions(internal_model_type)
|
115 |
+
if model_version not in versions:
|
116 |
+
log(f"Warning: Specified model version '{model_version}' is not recognized")
|
117 |
+
# Fall back to default version for this model
|
118 |
+
if len(versions) > 0:
|
119 |
+
model_version = next(iter(versions.keys()))
|
120 |
+
log(f"Using default model version instead: {model_version}")
|
121 |
else:
|
122 |
+
log(f"Using specified model version: {model_version}")
|
123 |
+
else:
|
124 |
+
# No model version specified, use default
|
125 |
+
versions = self.get_model_versions(internal_model_type)
|
126 |
+
if len(versions) > 0:
|
127 |
+
model_version = next(iter(versions.keys()))
|
128 |
+
log(f"No model version specified, using default: {model_version}")
|
129 |
+
else:
|
130 |
+
# Fall back to hardcoded defaults if no versions defined
|
131 |
if internal_model_type == "wan":
|
132 |
+
model_version = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
|
133 |
elif internal_model_type == "ltx_video":
|
134 |
+
model_version = "Lightricks/LTX-Video"
|
135 |
elif internal_model_type == "hunyuan_video":
|
136 |
+
model_version = "hunyuanvideo-community/HunyuanVideo"
|
137 |
+
log(f"No versions defined for model type, using default: {model_version}")
|
138 |
|
139 |
# Check if this is an image-to-video model but no image was provided
|
140 |
+
model_version_info = versions.get(model_version, {})
|
141 |
+
if model_version_info.get("type") == "image-to-video" and not conditioning_image:
|
142 |
+
return None, "Error: This model requires a conditioning image", log("Error: This model version requires a conditioning image but none was provided")
|
143 |
|
144 |
log(f"Generating video with model type: {internal_model_type}")
|
145 |
+
log(f"Using model version: {model_version}")
|
146 |
log(f"Using LoRA weights from: {lora_path}")
|
147 |
log(f"Resolution: {width}x{height}, Frames: {num_frames}, FPS: {fps}")
|
148 |
log(f"Guidance Scale: {guidance_scale}, Flow Shift: {flow_shift}, LoRA Weight: {lora_weight}")
|
149 |
+
log(f"Generation Seed: {seed}")
|
150 |
log(f"Prompt: {full_prompt}")
|
151 |
log(f"Negative Prompt: {negative_prompt}")
|
152 |
|
|
|
155 |
return self.generate_wan_video(
|
156 |
full_prompt, negative_prompt, width, height, num_frames,
|
157 |
guidance_scale, flow_shift, lora_path, lora_weight,
|
158 |
+
inference_steps, seed, enable_cpu_offload, fps, log,
|
159 |
+
model_version, conditioning_image
|
160 |
)
|
161 |
elif internal_model_type == "ltx_video":
|
162 |
return self.generate_ltx_video(
|
163 |
full_prompt, negative_prompt, width, height, num_frames,
|
164 |
guidance_scale, flow_shift, lora_path, lora_weight,
|
165 |
+
inference_steps, seed, enable_cpu_offload, fps, log,
|
166 |
+
model_version, conditioning_image
|
167 |
)
|
168 |
elif internal_model_type == "hunyuan_video":
|
169 |
return self.generate_hunyuan_video(
|
170 |
full_prompt, negative_prompt, width, height, num_frames,
|
171 |
guidance_scale, flow_shift, lora_path, lora_weight,
|
172 |
+
inference_steps, seed, enable_cpu_offload, fps, log,
|
173 |
+
model_version, conditioning_image
|
174 |
)
|
175 |
else:
|
176 |
return None, f"Error: Unsupported model type {internal_model_type}", log(f"Error: Unsupported model type {internal_model_type}")
|
|
|
191 |
lora_path: str,
|
192 |
lora_weight: float,
|
193 |
inference_steps: int,
|
194 |
+
seed: int = -1,
|
195 |
+
enable_cpu_offload: bool = True,
|
196 |
+
fps: int = 16,
|
197 |
+
log_fn: Callable = print,
|
198 |
+
model_version: str = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
|
199 |
conditioning_image: Optional[str] = None
|
200 |
) -> Tuple[Optional[str], str, str]:
|
201 |
"""Generate video using Wan model"""
|
202 |
|
203 |
try:
|
204 |
import torch
|
205 |
+
import numpy as np
|
206 |
from diffusers import AutoencoderKLWan, WanPipeline
|
207 |
from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
|
208 |
from diffusers.utils import export_to_video
|
|
|
212 |
start_time = torch.cuda.Event(enable_timing=True)
|
213 |
end_time = torch.cuda.Event(enable_timing=True)
|
214 |
|
215 |
+
print("Initializing wan generation..")
|
216 |
log_fn("Importing Wan model components...")
|
217 |
|
218 |
+
# Set up random seed
|
219 |
+
if seed == -1:
|
220 |
+
seed = random.randint(0, 2**32 - 1)
|
221 |
+
log_fn(f"Using randomly generated seed: {seed}")
|
222 |
+
|
223 |
+
# Set random seeds for reproducibility
|
224 |
+
random.seed(seed)
|
225 |
+
np.random.seed(seed)
|
226 |
+
torch.manual_seed(seed)
|
227 |
+
generator = torch.Generator(device="cuda")
|
228 |
+
generator = generator.manual_seed(seed)
|
229 |
|
230 |
+
log_fn(f"Loading VAE from {model_version}...")
|
231 |
+
vae = AutoencoderKLWan.from_pretrained(model_version, subfolder="vae", torch_dtype=torch.float32)
|
232 |
+
|
233 |
+
log_fn(f"Loading transformer from {model_version}...")
|
234 |
+
pipe = WanPipeline.from_pretrained(model_version, vae=vae, torch_dtype=torch.bfloat16)
|
235 |
|
236 |
log_fn(f"Configuring scheduler with flow_shift={flow_shift}...")
|
237 |
pipe.scheduler = UniPCMultistepScheduler.from_config(
|
|
|
245 |
if enable_cpu_offload:
|
246 |
log_fn("Enabling model CPU offload...")
|
247 |
pipe.enable_model_cpu_offload()
|
248 |
+
|
249 |
log_fn(f"Loading LoRA weights from {lora_path} with weight {lora_weight}...")
|
250 |
pipe.load_lora_weights(lora_path)
|
|
|
251 |
|
252 |
+
# TODO: Set the lora scale directly instead of using fuse_lora
|
253 |
+
#pipe._lora_scale = lora_weight
|
254 |
+
|
255 |
# Create temporary file for the output
|
256 |
with tempfile.NamedTemporaryFile(suffix='.mp4', delete=False) as temp_file:
|
257 |
output_path = temp_file.name
|
|
|
260 |
start_time.record()
|
261 |
|
262 |
# Check if this is an image-to-video model
|
263 |
+
is_i2v = "I2V" in model_version
|
264 |
|
265 |
if is_i2v and conditioning_image:
|
266 |
log_fn(f"Loading conditioning image from {conditioning_image}...")
|
|
|
277 |
num_frames=num_frames,
|
278 |
guidance_scale=guidance_scale,
|
279 |
num_inference_steps=inference_steps,
|
280 |
+
generator=generator,
|
281 |
).frames[0]
|
282 |
else:
|
283 |
log_fn("Generating video with text-only conditioning...")
|
|
|
289 |
num_frames=num_frames,
|
290 |
guidance_scale=guidance_scale,
|
291 |
num_inference_steps=inference_steps,
|
292 |
+
generator=generator,
|
293 |
).frames[0]
|
294 |
|
295 |
end_time.record()
|
|
|
310 |
return output_path, "Video generated successfully!", log_fn(f"Generation completed in {format_time(generation_time)}")
|
311 |
|
312 |
except Exception as e:
|
313 |
+
traceback.print_exc()
|
314 |
log_fn(f"Error generating video with Wan: {str(e)}")
|
315 |
# Clean up CUDA memory
|
316 |
torch.cuda.empty_cache()
|
317 |
return None, f"Error: {str(e)}", log_fn(f"Exception occurred: {str(e)}")
|
318 |
+
|
319 |
def generate_ltx_video(
|
320 |
self,
|
321 |
prompt: str,
|
|
|
328 |
lora_path: str,
|
329 |
lora_weight: float,
|
330 |
inference_steps: int,
|
331 |
+
seed: int = -1,
|
332 |
+
enable_cpu_offload: bool = True,
|
333 |
+
fps: int = 16,
|
334 |
+
log_fn: Callable = print,
|
335 |
+
model_version: str = "Lightricks/LTX-Video",
|
336 |
conditioning_image: Optional[str] = None
|
337 |
) -> Tuple[Optional[str], str, str]:
|
338 |
"""Generate video using LTX model"""
|
339 |
|
340 |
try:
|
341 |
import torch
|
342 |
+
import numpy as np
|
343 |
from diffusers import LTXPipeline
|
344 |
from diffusers.utils import export_to_video
|
345 |
from PIL import Image
|
346 |
|
347 |
start_time = torch.cuda.Event(enable_timing=True)
|
348 |
end_time = torch.cuda.Event(enable_timing=True)
|
349 |
+
|
350 |
+
# Set up random seed
|
351 |
+
if seed == -1:
|
352 |
+
seed = random.randint(0, 2**32 - 1)
|
353 |
+
log_fn(f"Using randomly generated seed: {seed}")
|
354 |
+
|
355 |
+
# Set random seeds for reproducibility
|
356 |
+
random.seed(seed)
|
357 |
+
np.random.seed(seed)
|
358 |
+
torch.manual_seed(seed)
|
359 |
+
generator = torch.Generator(device="cuda")
|
360 |
+
generator = generator.manual_seed(seed)
|
361 |
|
362 |
log_fn("Importing LTX model components...")
|
363 |
|
364 |
+
log_fn(f"Loading pipeline from {model_version}...")
|
365 |
+
pipe = LTXPipeline.from_pretrained(model_version, torch_dtype=torch.bfloat16)
|
366 |
|
367 |
log_fn("Moving pipeline to CUDA device...")
|
368 |
pipe.to("cuda")
|
|
|
393 |
decode_timestep=0.03,
|
394 |
decode_noise_scale=0.025,
|
395 |
num_inference_steps=inference_steps,
|
396 |
+
generator=generator,
|
397 |
).frames[0]
|
398 |
|
399 |
end_time.record()
|
|
|
431 |
lora_path: str,
|
432 |
lora_weight: float,
|
433 |
inference_steps: int,
|
434 |
+
seed: int = -1,
|
435 |
+
enable_cpu_offload: bool = True,
|
436 |
+
fps: int = 16,
|
437 |
+
log_fn: Callable = print,
|
438 |
+
model_version: str = "hunyuanvideo-community/HunyuanVideo",
|
439 |
conditioning_image: Optional[str] = None
|
440 |
) -> Tuple[Optional[str], str, str]:
|
441 |
"""Generate video using HunyuanVideo model"""
|
|
|
443 |
|
444 |
try:
|
445 |
import torch
|
446 |
+
import numpy as np
|
447 |
from diffusers import HunyuanVideoPipeline, HunyuanVideoTransformer3DModel, AutoencoderKLHunyuanVideo
|
448 |
from diffusers.utils import export_to_video
|
449 |
|
450 |
start_time = torch.cuda.Event(enable_timing=True)
|
451 |
end_time = torch.cuda.Event(enable_timing=True)
|
452 |
+
|
453 |
+
# Set up random seed
|
454 |
+
if seed == -1:
|
455 |
+
seed = random.randint(0, 2**32 - 1)
|
456 |
+
log_fn(f"Using randomly generated seed: {seed}")
|
457 |
+
|
458 |
+
# Set random seeds for reproducibility
|
459 |
+
random.seed(seed)
|
460 |
+
np.random.seed(seed)
|
461 |
+
torch.manual_seed(seed)
|
462 |
+
generator = torch.Generator(device="cuda")
|
463 |
+
generator = generator.manual_seed(seed)
|
464 |
|
465 |
log_fn("Importing HunyuanVideo model components...")
|
466 |
|
467 |
+
log_fn(f"Loading transformer from {model_version}...")
|
468 |
transformer = HunyuanVideoTransformer3DModel.from_pretrained(
|
469 |
+
model_version,
|
470 |
subfolder="transformer",
|
471 |
torch_dtype=torch.bfloat16
|
472 |
)
|
473 |
|
474 |
+
log_fn(f"Loading pipeline from {model_version}...")
|
475 |
pipe = HunyuanVideoPipeline.from_pretrained(
|
476 |
+
model_version,
|
477 |
transformer=transformer,
|
478 |
torch_dtype=torch.float16
|
479 |
)
|
|
|
512 |
guidance_scale=guidance_scale,
|
513 |
true_cfg_scale=1.0,
|
514 |
num_inference_steps=inference_steps,
|
515 |
+
generator=generator,
|
516 |
).frames[0]
|
517 |
|
518 |
end_time.record()
|
vms/ui/project/services/splitting.py
CHANGED
@@ -16,6 +16,7 @@ from vms.config import TRAINING_PATH, STORAGE_PATH, TRAINING_VIDEOS_PATH, VIDEOS
|
|
16 |
from vms.utils import remove_black_bars, extract_scene_info, is_video_file, is_image_file, add_prefix_to_caption
|
17 |
|
18 |
logger = logging.getLogger(__name__)
|
|
|
19 |
|
20 |
class SplittingService:
|
21 |
def __init__(self):
|
|
|
16 |
from vms.utils import remove_black_bars, extract_scene_info, is_video_file, is_image_file, add_prefix_to_caption
|
17 |
|
18 |
logger = logging.getLogger(__name__)
|
19 |
+
logger.setLevel(logging.INFO)
|
20 |
|
21 |
class SplittingService:
|
22 |
def __init__(self):
|
vms/ui/project/services/training.py
CHANGED
@@ -23,7 +23,7 @@ from huggingface_hub import upload_folder, create_repo
|
|
23 |
from vms.config import (
|
24 |
TrainingConfig, TRAINING_PRESETS, LOG_FILE_PATH, TRAINING_VIDEOS_PATH,
|
25 |
STORAGE_PATH, TRAINING_PATH, MODEL_PATH, OUTPUT_PATH, HF_API_TOKEN,
|
26 |
-
MODEL_TYPES, TRAINING_TYPES,
|
27 |
DEFAULT_NB_TRAINING_STEPS, DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS,
|
28 |
DEFAULT_BATCH_SIZE, DEFAULT_CAPTION_DROPOUT_P,
|
29 |
DEFAULT_LEARNING_RATE,
|
@@ -50,6 +50,7 @@ from vms.utils import (
|
|
50 |
)
|
51 |
|
52 |
logger = logging.getLogger(__name__)
|
|
|
53 |
|
54 |
class TrainingService:
|
55 |
def __init__(self, app=None):
|
@@ -134,6 +135,7 @@ class TrainingService:
|
|
134 |
validated_values = {}
|
135 |
default_state = {
|
136 |
"model_type": list(MODEL_TYPES.keys())[0],
|
|
|
137 |
"training_type": list(TRAINING_TYPES.keys())[0],
|
138 |
"lora_rank": DEFAULT_LORA_RANK_STR,
|
139 |
"lora_alpha": DEFAULT_LORA_ALPHA_STR,
|
@@ -213,6 +215,7 @@ class TrainingService:
|
|
213 |
ui_state_file = OUTPUT_PATH / "ui_state.json"
|
214 |
default_state = {
|
215 |
"model_type": list(MODEL_TYPES.keys())[0],
|
|
|
216 |
"training_type": list(TRAINING_TYPES.keys())[0],
|
217 |
"lora_rank": DEFAULT_LORA_RANK_STR,
|
218 |
"lora_alpha": DEFAULT_LORA_ALPHA_STR,
|
@@ -255,7 +258,7 @@ class TrainingService:
|
|
255 |
if "model_type" in saved_state and " (LoRA)" in saved_state["model_type"]:
|
256 |
saved_state["model_type"] = saved_state["model_type"].replace(" (LoRA)", "")
|
257 |
logger.info(f"Removed (LoRA) suffix from saved model type: {saved_state['model_type']}")
|
258 |
-
|
259 |
# Convert numeric values to appropriate types
|
260 |
if "train_steps" in saved_state:
|
261 |
try:
|
@@ -302,6 +305,18 @@ class TrainingService:
|
|
302 |
if not model_found:
|
303 |
merged_state["model_type"] = default_state["model_type"]
|
304 |
logger.warning(f"Invalid model type in saved state, using default")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
305 |
|
306 |
# Validate training_type is in available choices
|
307 |
if merged_state["training_type"] not in TRAINING_TYPES:
|
@@ -545,6 +560,7 @@ class TrainingService:
|
|
545 |
repo_id: str,
|
546 |
preset_name: str,
|
547 |
training_type: str = DEFAULT_TRAINING_TYPE,
|
|
|
548 |
resume_from_checkpoint: Optional[str] = None,
|
549 |
num_gpus: int = DEFAULT_NUM_GPUS,
|
550 |
precomputation_items: int = DEFAULT_PRECOMPUTATION_ITEMS,
|
@@ -869,6 +885,7 @@ class TrainingService:
|
|
869 |
# Save session info including repo_id for later hub upload
|
870 |
self.save_session({
|
871 |
"model_type": model_type,
|
|
|
872 |
"training_type": training_type,
|
873 |
"lora_rank": lora_rank,
|
874 |
"lora_alpha": lora_alpha,
|
@@ -1039,6 +1056,7 @@ class TrainingService:
|
|
1039 |
last_session = {
|
1040 |
"params": {
|
1041 |
"model_type": MODEL_TYPES.get(ui_state.get("model_type", list(MODEL_TYPES.keys())[0])),
|
|
|
1042 |
"training_type": TRAINING_TYPES.get(ui_state.get("training_type", list(TRAINING_TYPES.keys())[0])),
|
1043 |
"lora_rank": ui_state.get("lora_rank", DEFAULT_LORA_RANK_STR),
|
1044 |
"lora_alpha": ui_state.get("lora_alpha", DEFAULT_LORA_ALPHA_STR),
|
@@ -1102,8 +1120,9 @@ class TrainingService:
|
|
1102 |
# Add UI updates to restore the training parameters in the UI
|
1103 |
# This shows the user what values are being used for the resumed training
|
1104 |
ui_updates.update({
|
1105 |
-
"model_type": model_type_display,
|
1106 |
-
"
|
|
|
1107 |
"lora_rank": params.get('lora_rank', DEFAULT_LORA_RANK_STR),
|
1108 |
"lora_alpha": params.get('lora_alpha', DEFAULT_LORA_ALPHA_STR),
|
1109 |
"train_steps": params.get('train_steps', DEFAULT_NB_TRAINING_STEPS),
|
@@ -1122,19 +1141,19 @@ class TrainingService:
|
|
1122 |
# Use the internal model_type for the actual training
|
1123 |
# But keep model_type_display for the UI
|
1124 |
result = self.start_training(
|
1125 |
-
model_type=
|
1126 |
lora_rank=params.get('lora_rank', DEFAULT_LORA_RANK_STR),
|
1127 |
lora_alpha=params.get('lora_alpha', DEFAULT_LORA_ALPHA_STR),
|
1128 |
train_size=params.get('train_steps', DEFAULT_NB_TRAINING_STEPS),
|
1129 |
batch_size=params.get('batch_size', DEFAULT_BATCH_SIZE),
|
1130 |
learning_rate=params.get('learning_rate', DEFAULT_LEARNING_RATE),
|
1131 |
save_iterations=params.get('save_iterations', DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS),
|
|
|
1132 |
repo_id=params.get('repo_id', ''),
|
1133 |
preset_name=params.get('preset_name', list(TRAINING_PRESETS.keys())[0]),
|
1134 |
training_type=training_type_internal,
|
1135 |
resume_from_checkpoint=str(latest_checkpoint)
|
1136 |
)
|
1137 |
-
|
1138 |
# Set buttons for active training
|
1139 |
ui_updates.update({
|
1140 |
"start_btn": {"interactive": False, "variant": "secondary", "value": "Continue Training"},
|
@@ -1142,7 +1161,7 @@ class TrainingService:
|
|
1142 |
"delete_checkpoints_btn": {"interactive": False, "variant": "stop", "value": "Delete All Checkpoints"},
|
1143 |
"pause_resume_btn": {"interactive": False, "variant": "secondary", "visible": False}
|
1144 |
})
|
1145 |
-
|
1146 |
return {
|
1147 |
"status": "recovered",
|
1148 |
"message": f"Training resumed from checkpoint {checkpoint_step}",
|
|
|
23 |
from vms.config import (
|
24 |
TrainingConfig, TRAINING_PRESETS, LOG_FILE_PATH, TRAINING_VIDEOS_PATH,
|
25 |
STORAGE_PATH, TRAINING_PATH, MODEL_PATH, OUTPUT_PATH, HF_API_TOKEN,
|
26 |
+
MODEL_TYPES, TRAINING_TYPES, MODEL_VERSIONS,
|
27 |
DEFAULT_NB_TRAINING_STEPS, DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS,
|
28 |
DEFAULT_BATCH_SIZE, DEFAULT_CAPTION_DROPOUT_P,
|
29 |
DEFAULT_LEARNING_RATE,
|
|
|
50 |
)
|
51 |
|
52 |
logger = logging.getLogger(__name__)
|
53 |
+
logger.setLevel(logging.INFO)
|
54 |
|
55 |
class TrainingService:
|
56 |
def __init__(self, app=None):
|
|
|
135 |
validated_values = {}
|
136 |
default_state = {
|
137 |
"model_type": list(MODEL_TYPES.keys())[0],
|
138 |
+
"model_version": "",
|
139 |
"training_type": list(TRAINING_TYPES.keys())[0],
|
140 |
"lora_rank": DEFAULT_LORA_RANK_STR,
|
141 |
"lora_alpha": DEFAULT_LORA_ALPHA_STR,
|
|
|
215 |
ui_state_file = OUTPUT_PATH / "ui_state.json"
|
216 |
default_state = {
|
217 |
"model_type": list(MODEL_TYPES.keys())[0],
|
218 |
+
"model_version": "",
|
219 |
"training_type": list(TRAINING_TYPES.keys())[0],
|
220 |
"lora_rank": DEFAULT_LORA_RANK_STR,
|
221 |
"lora_alpha": DEFAULT_LORA_ALPHA_STR,
|
|
|
258 |
if "model_type" in saved_state and " (LoRA)" in saved_state["model_type"]:
|
259 |
saved_state["model_type"] = saved_state["model_type"].replace(" (LoRA)", "")
|
260 |
logger.info(f"Removed (LoRA) suffix from saved model type: {saved_state['model_type']}")
|
261 |
+
|
262 |
# Convert numeric values to appropriate types
|
263 |
if "train_steps" in saved_state:
|
264 |
try:
|
|
|
305 |
if not model_found:
|
306 |
merged_state["model_type"] = default_state["model_type"]
|
307 |
logger.warning(f"Invalid model type in saved state, using default")
|
308 |
+
|
309 |
+
# Validate model_version is appropriate for model_type
|
310 |
+
if "model_type" in merged_state and "model_version" in merged_state:
|
311 |
+
model_internal_type = MODEL_TYPES.get(merged_state["model_type"])
|
312 |
+
if model_internal_type:
|
313 |
+
valid_versions = MODEL_VERSIONS.get(model_internal_type, {}).keys()
|
314 |
+
if merged_state["model_version"] not in valid_versions:
|
315 |
+
# Set to default for this model type
|
316 |
+
from vms.ui.project.tabs.train_tab import TrainTab
|
317 |
+
train_tab = TrainTab(None) # Temporary instance just for the helper method
|
318 |
+
merged_state["model_version"] = train_tab.get_default_model_version(saved_state["model_type"])
|
319 |
+
logger.warning(f"Invalid model version for {merged_state['model_type']}, using default")
|
320 |
|
321 |
# Validate training_type is in available choices
|
322 |
if merged_state["training_type"] not in TRAINING_TYPES:
|
|
|
560 |
repo_id: str,
|
561 |
preset_name: str,
|
562 |
training_type: str = DEFAULT_TRAINING_TYPE,
|
563 |
+
model_version: str = "",
|
564 |
resume_from_checkpoint: Optional[str] = None,
|
565 |
num_gpus: int = DEFAULT_NUM_GPUS,
|
566 |
precomputation_items: int = DEFAULT_PRECOMPUTATION_ITEMS,
|
|
|
885 |
# Save session info including repo_id for later hub upload
|
886 |
self.save_session({
|
887 |
"model_type": model_type,
|
888 |
+
"model_version": model_version,
|
889 |
"training_type": training_type,
|
890 |
"lora_rank": lora_rank,
|
891 |
"lora_alpha": lora_alpha,
|
|
|
1056 |
last_session = {
|
1057 |
"params": {
|
1058 |
"model_type": MODEL_TYPES.get(ui_state.get("model_type", list(MODEL_TYPES.keys())[0])),
|
1059 |
+
"model_version": ui_state.get("model_version", ""),
|
1060 |
"training_type": TRAINING_TYPES.get(ui_state.get("training_type", list(TRAINING_TYPES.keys())[0])),
|
1061 |
"lora_rank": ui_state.get("lora_rank", DEFAULT_LORA_RANK_STR),
|
1062 |
"lora_alpha": ui_state.get("lora_alpha", DEFAULT_LORA_ALPHA_STR),
|
|
|
1120 |
# Add UI updates to restore the training parameters in the UI
|
1121 |
# This shows the user what values are being used for the resumed training
|
1122 |
ui_updates.update({
|
1123 |
+
"model_type": model_type_display,
|
1124 |
+
"model_version": params.get('model_version', ''),
|
1125 |
+
"training_type": training_type_display,
|
1126 |
"lora_rank": params.get('lora_rank', DEFAULT_LORA_RANK_STR),
|
1127 |
"lora_alpha": params.get('lora_alpha', DEFAULT_LORA_ALPHA_STR),
|
1128 |
"train_steps": params.get('train_steps', DEFAULT_NB_TRAINING_STEPS),
|
|
|
1141 |
# Use the internal model_type for the actual training
|
1142 |
# But keep model_type_display for the UI
|
1143 |
result = self.start_training(
|
1144 |
+
model_type=model_internal_type,
|
1145 |
lora_rank=params.get('lora_rank', DEFAULT_LORA_RANK_STR),
|
1146 |
lora_alpha=params.get('lora_alpha', DEFAULT_LORA_ALPHA_STR),
|
1147 |
train_size=params.get('train_steps', DEFAULT_NB_TRAINING_STEPS),
|
1148 |
batch_size=params.get('batch_size', DEFAULT_BATCH_SIZE),
|
1149 |
learning_rate=params.get('learning_rate', DEFAULT_LEARNING_RATE),
|
1150 |
save_iterations=params.get('save_iterations', DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS),
|
1151 |
+
model_version=params.get('model_version', ''),
|
1152 |
repo_id=params.get('repo_id', ''),
|
1153 |
preset_name=params.get('preset_name', list(TRAINING_PRESETS.keys())[0]),
|
1154 |
training_type=training_type_internal,
|
1155 |
resume_from_checkpoint=str(latest_checkpoint)
|
1156 |
)
|
|
|
1157 |
# Set buttons for active training
|
1158 |
ui_updates.update({
|
1159 |
"start_btn": {"interactive": False, "variant": "secondary", "value": "Continue Training"},
|
|
|
1161 |
"delete_checkpoints_btn": {"interactive": False, "variant": "stop", "value": "Delete All Checkpoints"},
|
1162 |
"pause_resume_btn": {"interactive": False, "variant": "secondary", "visible": False}
|
1163 |
})
|
1164 |
+
|
1165 |
return {
|
1166 |
"status": "recovered",
|
1167 |
"message": f"Training resumed from checkpoint {checkpoint_step}",
|
vms/ui/project/tabs/preview_tab.py
CHANGED
@@ -4,16 +4,18 @@ Preview tab for Video Model Studio UI
|
|
4 |
|
5 |
import gradio as gr
|
6 |
import logging
|
|
|
7 |
from pathlib import Path
|
8 |
from typing import Dict, Any, List, Optional, Tuple
|
9 |
import time
|
10 |
|
11 |
from vms.utils import BaseTab
|
12 |
from vms.config import (
|
13 |
-
MODEL_TYPES, DEFAULT_PROMPT_PREFIX
|
14 |
)
|
15 |
|
16 |
logger = logging.getLogger(__name__)
|
|
|
17 |
|
18 |
class PreviewTab(BaseTab):
|
19 |
"""Preview tab for testing trained models"""
|
@@ -49,25 +51,35 @@ class PreviewTab(BaseTab):
|
|
49 |
placeholder="Prefix to add to all prompts",
|
50 |
value=DEFAULT_PROMPT_PREFIX
|
51 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
|
53 |
with gr.Row():
|
54 |
# Get the currently selected model type from training tab if possible
|
55 |
default_model = self.get_default_model_type()
|
56 |
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
|
|
71 |
|
72 |
# Add image input for image-to-video models
|
73 |
self.components["conditioning_image"] = gr.Image(
|
@@ -177,36 +189,55 @@ class PreviewTab(BaseTab):
|
|
177 |
|
178 |
return tab
|
179 |
|
180 |
-
def
|
181 |
-
"""Get model
|
182 |
# Convert UI display name to internal name
|
183 |
internal_type = MODEL_TYPES.get(model_type)
|
184 |
if not internal_type:
|
185 |
return []
|
186 |
|
187 |
-
# Get
|
188 |
-
|
189 |
-
if not
|
190 |
return []
|
191 |
|
192 |
# Format choices with display name and description
|
193 |
choices = []
|
194 |
-
for model_id, info in
|
195 |
choices.append(f"{model_id} - {info.get('name', '')}")
|
196 |
|
197 |
return choices
|
198 |
|
199 |
-
def
|
200 |
-
"""Get default model
|
201 |
-
choices = self.
|
202 |
if choices:
|
203 |
return choices[0]
|
204 |
return ""
|
205 |
-
|
206 |
def get_default_model_type(self) -> str:
|
207 |
-
"""Get the
|
208 |
try:
|
209 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
210 |
ui_state = self.app.training.load_ui_state()
|
211 |
model_type = ui_state.get("model_type")
|
212 |
|
@@ -214,7 +245,7 @@ class PreviewTab(BaseTab):
|
|
214 |
if model_type in MODEL_TYPES:
|
215 |
return model_type
|
216 |
|
217 |
-
# If we couldn't get a valid model type, try to get it from the training tab
|
218 |
if hasattr(self.app, 'tabs') and 'train_tab' in self.app.tabs:
|
219 |
train_tab = self.app.tabs['train_tab']
|
220 |
if hasattr(train_tab, 'components') and 'model_type' in train_tab.components:
|
@@ -225,31 +256,31 @@ class PreviewTab(BaseTab):
|
|
225 |
# Fallback to first model type
|
226 |
return list(MODEL_TYPES.keys())[0]
|
227 |
except Exception as e:
|
228 |
-
logger.warning(f"Failed to get default model type: {e}")
|
229 |
return list(MODEL_TYPES.keys())[0]
|
230 |
|
231 |
-
def extract_model_id(self,
|
232 |
-
"""Extract model ID from
|
233 |
-
if " - " in
|
234 |
-
return
|
235 |
-
return
|
236 |
|
237 |
-
def
|
238 |
-
"""Get the
|
239 |
# Convert UI display name to internal name
|
240 |
internal_type = MODEL_TYPES.get(model_type)
|
241 |
if not internal_type:
|
242 |
return "text-to-video"
|
243 |
|
244 |
-
# Extract model_id from
|
245 |
-
model_id = self.extract_model_id(
|
246 |
|
247 |
-
# Get
|
248 |
-
|
249 |
-
|
250 |
|
251 |
-
# Return the
|
252 |
-
return
|
253 |
|
254 |
def connect_events(self) -> None:
|
255 |
"""Connect event handlers to UI components"""
|
@@ -264,23 +295,23 @@ class PreviewTab(BaseTab):
|
|
264 |
]
|
265 |
)
|
266 |
|
267 |
-
# Update
|
268 |
if hasattr(self.app, 'tabs_component') and self.app.tabs_component is not None:
|
269 |
self.app.tabs_component.select(
|
270 |
-
fn=self.
|
271 |
inputs=[],
|
272 |
outputs=[
|
273 |
self.components["model_type"],
|
274 |
-
self.components["
|
275 |
]
|
276 |
)
|
277 |
|
278 |
-
# Update
|
279 |
-
self.components["
|
280 |
-
fn=self.
|
281 |
inputs=[
|
282 |
self.components["model_type"],
|
283 |
-
self.components["
|
284 |
],
|
285 |
outputs=[
|
286 |
self.components["conditioning_image"]
|
@@ -305,13 +336,13 @@ class PreviewTab(BaseTab):
|
|
305 |
self.components["lora_weight"],
|
306 |
self.components["inference_steps"],
|
307 |
self.components["enable_cpu_offload"],
|
308 |
-
self.components["
|
309 |
]
|
310 |
)
|
311 |
|
312 |
# Save preview UI state when values change
|
313 |
for component_name in [
|
314 |
-
"prompt", "negative_prompt", "prompt_prefix", "
|
315 |
"width", "height", "num_frames", "fps", "guidance_scale", "flow_shift",
|
316 |
"lora_weight", "inference_steps", "enable_cpu_offload"
|
317 |
]:
|
@@ -327,7 +358,7 @@ class PreviewTab(BaseTab):
|
|
327 |
fn=self.generate_video,
|
328 |
inputs=[
|
329 |
self.components["model_type"],
|
330 |
-
self.components["
|
331 |
self.components["prompt"],
|
332 |
self.components["negative_prompt"],
|
333 |
self.components["prompt_prefix"],
|
@@ -349,22 +380,41 @@ class PreviewTab(BaseTab):
|
|
349 |
]
|
350 |
)
|
351 |
|
352 |
-
def
|
353 |
-
"""Update UI based on the selected model
|
354 |
-
|
355 |
|
356 |
# Show conditioning image input only for image-to-video models
|
357 |
-
show_conditioning_image =
|
358 |
|
359 |
return {
|
360 |
self.components["conditioning_image"]: gr.Image(visible=show_conditioning_image)
|
361 |
}
|
362 |
|
363 |
-
def
|
364 |
-
"""Sync model type with training tab when preview tab is selected and update
|
365 |
model_type = self.get_default_model_type()
|
366 |
-
|
367 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
368 |
|
369 |
def update_resolution(self, preset: str) -> Tuple[int, int, float]:
|
370 |
"""Update resolution and flow shift based on preset"""
|
@@ -385,11 +435,11 @@ class PreviewTab(BaseTab):
|
|
385 |
# Get model type (can't be changed in UI)
|
386 |
model_type = self.get_default_model_type()
|
387 |
|
388 |
-
# If
|
389 |
-
|
390 |
-
|
391 |
-
if
|
392 |
-
|
393 |
|
394 |
return (
|
395 |
preview_state.get("prompt", ""),
|
@@ -404,7 +454,7 @@ class PreviewTab(BaseTab):
|
|
404 |
preview_state.get("lora_weight", 0.7),
|
405 |
preview_state.get("inference_steps", 30),
|
406 |
preview_state.get("enable_cpu_offload", True),
|
407 |
-
|
408 |
)
|
409 |
except Exception as e:
|
410 |
logger.error(f"Error loading preview state: {e}")
|
@@ -414,7 +464,7 @@ class PreviewTab(BaseTab):
|
|
414 |
"worst quality, low quality, blurry, jittery, distorted, ugly, deformed, disfigured, messy background",
|
415 |
DEFAULT_PROMPT_PREFIX,
|
416 |
832, 480, 49, 16, 5.0, 3.0, 0.7, 30, True,
|
417 |
-
self.
|
418 |
)
|
419 |
|
420 |
def save_preview_state_value(self, value: Any) -> None:
|
@@ -456,7 +506,7 @@ class PreviewTab(BaseTab):
|
|
456 |
def generate_video(
|
457 |
self,
|
458 |
model_type: str,
|
459 |
-
|
460 |
prompt: str,
|
461 |
negative_prompt: str,
|
462 |
prompt_prefix: str,
|
@@ -473,13 +523,14 @@ class PreviewTab(BaseTab):
|
|
473 |
) -> Tuple[Optional[str], str, str]:
|
474 |
"""Handler for generate button click, delegates to preview service"""
|
475 |
# Save all the parameters to preview state before generating
|
|
|
476 |
try:
|
477 |
state = self.app.training.load_ui_state()
|
478 |
if "preview" not in state:
|
479 |
state["preview"] = {}
|
480 |
|
481 |
-
# Extract model ID from
|
482 |
-
|
483 |
|
484 |
# Update all values
|
485 |
preview_state = {
|
@@ -487,7 +538,7 @@ class PreviewTab(BaseTab):
|
|
487 |
"negative_prompt": negative_prompt,
|
488 |
"prompt_prefix": prompt_prefix,
|
489 |
"model_type": model_type,
|
490 |
-
"
|
491 |
"width": width,
|
492 |
"height": height,
|
493 |
"num_frames": num_frames,
|
@@ -504,40 +555,30 @@ class PreviewTab(BaseTab):
|
|
504 |
except Exception as e:
|
505 |
logger.error(f"Error saving preview state before generation: {e}")
|
506 |
|
507 |
-
#
|
508 |
-
|
509 |
-
yield None, "Starting generation...", ""
|
510 |
-
time.sleep(0.1)
|
511 |
|
512 |
-
#
|
513 |
-
|
514 |
|
515 |
-
#
|
516 |
-
|
517 |
-
|
518 |
-
|
519 |
-
|
520 |
-
|
521 |
-
|
522 |
-
|
523 |
-
|
524 |
-
|
525 |
-
|
526 |
-
|
527 |
-
|
528 |
-
|
529 |
-
|
530 |
-
|
531 |
-
|
532 |
-
|
533 |
-
inference_steps=inference_steps,
|
534 |
-
enable_cpu_offload=enable_cpu_offload,
|
535 |
-
fps=fps,
|
536 |
-
conditioning_image=conditioning_image
|
537 |
-
)
|
538 |
-
|
539 |
-
# Return final result
|
540 |
-
return result
|
541 |
|
542 |
-
# Return
|
543 |
-
return
|
|
|
4 |
|
5 |
import gradio as gr
|
6 |
import logging
|
7 |
+
import json
|
8 |
from pathlib import Path
|
9 |
from typing import Dict, Any, List, Optional, Tuple
|
10 |
import time
|
11 |
|
12 |
from vms.utils import BaseTab
|
13 |
from vms.config import (
|
14 |
+
OUTPUT_PATH, MODEL_TYPES, DEFAULT_PROMPT_PREFIX, MODEL_VERSIONS
|
15 |
)
|
16 |
|
17 |
logger = logging.getLogger(__name__)
|
18 |
+
logger.setLevel(logging.INFO)
|
19 |
|
20 |
class PreviewTab(BaseTab):
|
21 |
"""Preview tab for testing trained models"""
|
|
|
51 |
placeholder="Prefix to add to all prompts",
|
52 |
value=DEFAULT_PROMPT_PREFIX
|
53 |
)
|
54 |
+
|
55 |
+
self.components["seed"] = gr.Slider(
|
56 |
+
label="Generation Seed (-1 for random)",
|
57 |
+
minimum=-1,
|
58 |
+
maximum=2147483647, # 2^31 - 1
|
59 |
+
step=1,
|
60 |
+
value=-1,
|
61 |
+
info="Set to -1 for random seed or specific value for reproducible results"
|
62 |
+
)
|
63 |
|
64 |
with gr.Row():
|
65 |
# Get the currently selected model type from training tab if possible
|
66 |
default_model = self.get_default_model_type()
|
67 |
|
68 |
+
with gr.Column():
|
69 |
+
# Make model_type read-only (disabled), as it must match what was trained
|
70 |
+
self.components["model_type"] = gr.Dropdown(
|
71 |
+
choices=list(MODEL_TYPES.keys()),
|
72 |
+
label="Model Type (from training)",
|
73 |
+
value=default_model,
|
74 |
+
interactive=False
|
75 |
+
)
|
76 |
+
|
77 |
+
# Add model version selection based on model type
|
78 |
+
self.components["model_version"] = gr.Dropdown(
|
79 |
+
label="Model Version",
|
80 |
+
choices=self.get_model_version_choices(default_model),
|
81 |
+
value=self.get_default_model_version(default_model)
|
82 |
+
)
|
83 |
|
84 |
# Add image input for image-to-video models
|
85 |
self.components["conditioning_image"] = gr.Image(
|
|
|
189 |
|
190 |
return tab
|
191 |
|
192 |
+
def get_model_version_choices(self, model_type: str) -> List[str]:
|
193 |
+
"""Get model version choices based on model type"""
|
194 |
# Convert UI display name to internal name
|
195 |
internal_type = MODEL_TYPES.get(model_type)
|
196 |
if not internal_type:
|
197 |
return []
|
198 |
|
199 |
+
# Get versions from preview service
|
200 |
+
versions = self.app.previewing.get_model_versions(internal_type)
|
201 |
+
if not versions:
|
202 |
return []
|
203 |
|
204 |
# Format choices with display name and description
|
205 |
choices = []
|
206 |
+
for model_id, info in versions.items():
|
207 |
choices.append(f"{model_id} - {info.get('name', '')}")
|
208 |
|
209 |
return choices
|
210 |
|
211 |
+
def get_default_model_version(self, model_type: str) -> str:
|
212 |
+
"""Get default model version for the model type"""
|
213 |
+
choices = self.get_model_version_choices(model_type)
|
214 |
if choices:
|
215 |
return choices[0]
|
216 |
return ""
|
217 |
+
|
218 |
def get_default_model_type(self) -> str:
|
219 |
+
"""Get the model type from the latest training session"""
|
220 |
try:
|
221 |
+
# First check the session.json which contains the actual training data
|
222 |
+
session_file = OUTPUT_PATH / "session.json"
|
223 |
+
if session_file.exists():
|
224 |
+
with open(session_file, 'r') as f:
|
225 |
+
session_data = json.load(f)
|
226 |
+
|
227 |
+
# Get the internal model type from the session parameters
|
228 |
+
if "params" in session_data and "model_type" in session_data["params"]:
|
229 |
+
internal_model_type = session_data["params"]["model_type"]
|
230 |
+
|
231 |
+
# Convert internal model type to display name
|
232 |
+
for display_name, internal_name in MODEL_TYPES.items():
|
233 |
+
if internal_name == internal_model_type:
|
234 |
+
logger.info(f"Using model type '{display_name}' from session file")
|
235 |
+
return display_name
|
236 |
+
|
237 |
+
# If we couldn't map it, log a warning
|
238 |
+
logger.warning(f"Could not map internal model type '{internal_model_type}' to a display name")
|
239 |
+
|
240 |
+
# If we couldn't get it from session.json, try to get it from UI state
|
241 |
ui_state = self.app.training.load_ui_state()
|
242 |
model_type = ui_state.get("model_type")
|
243 |
|
|
|
245 |
if model_type in MODEL_TYPES:
|
246 |
return model_type
|
247 |
|
248 |
+
# If we still couldn't get a valid model type, try to get it from the training tab
|
249 |
if hasattr(self.app, 'tabs') and 'train_tab' in self.app.tabs:
|
250 |
train_tab = self.app.tabs['train_tab']
|
251 |
if hasattr(train_tab, 'components') and 'model_type' in train_tab.components:
|
|
|
256 |
# Fallback to first model type
|
257 |
return list(MODEL_TYPES.keys())[0]
|
258 |
except Exception as e:
|
259 |
+
logger.warning(f"Failed to get default model type from session: {e}")
|
260 |
return list(MODEL_TYPES.keys())[0]
|
261 |
|
262 |
+
def extract_model_id(self, model_version_choice: str) -> str:
|
263 |
+
"""Extract model ID from model version choice string"""
|
264 |
+
if " - " in model_version_choice:
|
265 |
+
return model_version_choice.split(" - ")[0].strip()
|
266 |
+
return model_version_choice
|
267 |
|
268 |
+
def get_model_version_type(self, model_type: str, model_version: str) -> str:
|
269 |
+
"""Get the model version type (text-to-video or image-to-video)"""
|
270 |
# Convert UI display name to internal name
|
271 |
internal_type = MODEL_TYPES.get(model_type)
|
272 |
if not internal_type:
|
273 |
return "text-to-video"
|
274 |
|
275 |
+
# Extract model_id from model version choice
|
276 |
+
model_id = self.extract_model_id(model_version)
|
277 |
|
278 |
+
# Get versions from preview service
|
279 |
+
versions = self.app.previewing.get_model_versions(internal_type)
|
280 |
+
model_version_info = versions.get(model_id, {})
|
281 |
|
282 |
+
# Return the model version type or default to text-to-video
|
283 |
+
return model_version_info.get("type", "text-to-video")
|
284 |
|
285 |
def connect_events(self) -> None:
|
286 |
"""Connect event handlers to UI components"""
|
|
|
295 |
]
|
296 |
)
|
297 |
|
298 |
+
# Update model_version choices when model_type changes or tab is selected
|
299 |
if hasattr(self.app, 'tabs_component') and self.app.tabs_component is not None:
|
300 |
self.app.tabs_component.select(
|
301 |
+
fn=self.sync_model_type_and_verions,
|
302 |
inputs=[],
|
303 |
outputs=[
|
304 |
self.components["model_type"],
|
305 |
+
self.components["model_version"]
|
306 |
]
|
307 |
)
|
308 |
|
309 |
+
# Update model version-specific UI elements when version changes
|
310 |
+
self.components["model_version"].change(
|
311 |
+
fn=self.update_model_version_ui,
|
312 |
inputs=[
|
313 |
self.components["model_type"],
|
314 |
+
self.components["model_version"]
|
315 |
],
|
316 |
outputs=[
|
317 |
self.components["conditioning_image"]
|
|
|
336 |
self.components["lora_weight"],
|
337 |
self.components["inference_steps"],
|
338 |
self.components["enable_cpu_offload"],
|
339 |
+
self.components["model_version"]
|
340 |
]
|
341 |
)
|
342 |
|
343 |
# Save preview UI state when values change
|
344 |
for component_name in [
|
345 |
+
"prompt", "negative_prompt", "prompt_prefix", "model_version", "resolution_preset",
|
346 |
"width", "height", "num_frames", "fps", "guidance_scale", "flow_shift",
|
347 |
"lora_weight", "inference_steps", "enable_cpu_offload"
|
348 |
]:
|
|
|
358 |
fn=self.generate_video,
|
359 |
inputs=[
|
360 |
self.components["model_type"],
|
361 |
+
self.components["model_version"],
|
362 |
self.components["prompt"],
|
363 |
self.components["negative_prompt"],
|
364 |
self.components["prompt_prefix"],
|
|
|
380 |
]
|
381 |
)
|
382 |
|
383 |
+
def update_model_version_ui(self, model_type: str, model_version: str) -> Dict[str, Any]:
|
384 |
+
"""Update UI based on the selected model version"""
|
385 |
+
model_version_type = self.get_model_version_type(model_type, model_version)
|
386 |
|
387 |
# Show conditioning image input only for image-to-video models
|
388 |
+
show_conditioning_image = model_version_type == "image-to-video"
|
389 |
|
390 |
return {
|
391 |
self.components["conditioning_image"]: gr.Image(visible=show_conditioning_image)
|
392 |
}
|
393 |
|
394 |
+
def sync_model_type_and_verions(self) -> Tuple[str, str]:
|
395 |
+
"""Sync model type with training tab when preview tab is selected and update model version choices"""
|
396 |
model_type = self.get_default_model_type()
|
397 |
+
model_version = ""
|
398 |
+
|
399 |
+
# Try to get model_version from session or UI state
|
400 |
+
ui_state = self.app.training.load_ui_state()
|
401 |
+
preview_state = ui_state.get("preview", {})
|
402 |
+
model_version = preview_state.get("model_version", "")
|
403 |
+
|
404 |
+
if not model_version:
|
405 |
+
# Format it as a display choice
|
406 |
+
internal_type = MODEL_TYPES.get(model_type)
|
407 |
+
if internal_type and internal_type in MODEL_VERSIONS:
|
408 |
+
first_version = next(iter(MODEL_VERSIONS[internal_type].keys()), "")
|
409 |
+
if first_version:
|
410 |
+
model_version_info = MODEL_VERSIONS[internal_type][first_version]
|
411 |
+
model_version = f"{first_version} - {model_version_info.get('name', '')}"
|
412 |
+
|
413 |
+
# If we couldn't get it, use default
|
414 |
+
if not model_version:
|
415 |
+
model_version = self.get_default_model_version(model_type)
|
416 |
+
|
417 |
+
return model_type, model_version
|
418 |
|
419 |
def update_resolution(self, preset: str) -> Tuple[int, int, float]:
|
420 |
"""Update resolution and flow shift based on preset"""
|
|
|
435 |
# Get model type (can't be changed in UI)
|
436 |
model_type = self.get_default_model_type()
|
437 |
|
438 |
+
# If model_version not in choices for current model_type, use default
|
439 |
+
model_version = preview_state.get("model_version", "")
|
440 |
+
model_version_choices = self.get_model_version_choices(model_type)
|
441 |
+
if model_version not in model_version_choices and model_version_choices:
|
442 |
+
model_version = model_version_choices[0]
|
443 |
|
444 |
return (
|
445 |
preview_state.get("prompt", ""),
|
|
|
454 |
preview_state.get("lora_weight", 0.7),
|
455 |
preview_state.get("inference_steps", 30),
|
456 |
preview_state.get("enable_cpu_offload", True),
|
457 |
+
model_version
|
458 |
)
|
459 |
except Exception as e:
|
460 |
logger.error(f"Error loading preview state: {e}")
|
|
|
464 |
"worst quality, low quality, blurry, jittery, distorted, ugly, deformed, disfigured, messy background",
|
465 |
DEFAULT_PROMPT_PREFIX,
|
466 |
832, 480, 49, 16, 5.0, 3.0, 0.7, 30, True,
|
467 |
+
self.get_default_model_version(self.get_default_model_type())
|
468 |
)
|
469 |
|
470 |
def save_preview_state_value(self, value: Any) -> None:
|
|
|
506 |
def generate_video(
|
507 |
self,
|
508 |
model_type: str,
|
509 |
+
model_version: str,
|
510 |
prompt: str,
|
511 |
negative_prompt: str,
|
512 |
prompt_prefix: str,
|
|
|
523 |
) -> Tuple[Optional[str], str, str]:
|
524 |
"""Handler for generate button click, delegates to preview service"""
|
525 |
# Save all the parameters to preview state before generating
|
526 |
+
print("preview_tab: generate_video() has been called")
|
527 |
try:
|
528 |
state = self.app.training.load_ui_state()
|
529 |
if "preview" not in state:
|
530 |
state["preview"] = {}
|
531 |
|
532 |
+
# Extract model ID from model version choice
|
533 |
+
model_version_id = self.extract_model_id(model_version)
|
534 |
|
535 |
# Update all values
|
536 |
preview_state = {
|
|
|
538 |
"negative_prompt": negative_prompt,
|
539 |
"prompt_prefix": prompt_prefix,
|
540 |
"model_type": model_type,
|
541 |
+
"model_version": model_version,
|
542 |
"width": width,
|
543 |
"height": height,
|
544 |
"num_frames": num_frames,
|
|
|
555 |
except Exception as e:
|
556 |
logger.error(f"Error saving preview state before generation: {e}")
|
557 |
|
558 |
+
# Extract model ID from model version choice string
|
559 |
+
model_version_id = self.extract_model_id(model_version)
|
|
|
|
|
560 |
|
561 |
+
# Initial UI update
|
562 |
+
video_path, status, log = None, "Initializing generation...", "Starting video generation process..."
|
563 |
|
564 |
+
# Start actual generation
|
565 |
+
result = self.app.previewing.generate_video(
|
566 |
+
model_type=model_type,
|
567 |
+
model_version=model_version_id,
|
568 |
+
prompt=prompt,
|
569 |
+
negative_prompt=negative_prompt,
|
570 |
+
prompt_prefix=prompt_prefix,
|
571 |
+
width=width,
|
572 |
+
height=height,
|
573 |
+
num_frames=num_frames,
|
574 |
+
guidance_scale=guidance_scale,
|
575 |
+
flow_shift=flow_shift,
|
576 |
+
lora_weight=lora_weight,
|
577 |
+
inference_steps=inference_steps,
|
578 |
+
enable_cpu_offload=enable_cpu_offload,
|
579 |
+
fps=fps,
|
580 |
+
conditioning_image=conditioning_image
|
581 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
582 |
|
583 |
+
# Return final result
|
584 |
+
return result
|
vms/ui/project/tabs/train_tab.py
CHANGED
@@ -5,12 +5,15 @@ Train tab for Video Model Studio UI with improved task progress display
|
|
5 |
import gradio as gr
|
6 |
import logging
|
7 |
import os
|
|
|
8 |
from typing import Dict, Any, List, Optional, Tuple
|
9 |
from pathlib import Path
|
10 |
|
11 |
from vms.utils import BaseTab
|
12 |
from vms.config import (
|
13 |
-
|
|
|
|
|
14 |
DEFAULT_NB_TRAINING_STEPS, DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS,
|
15 |
DEFAULT_BATCH_SIZE, DEFAULT_CAPTION_DROPOUT_P,
|
16 |
DEFAULT_LEARNING_RATE,
|
@@ -53,12 +56,27 @@ class TrainTab(BaseTab):
|
|
53 |
|
54 |
with gr.Row():
|
55 |
with gr.Column():
|
|
|
|
|
|
|
56 |
self.components["model_type"] = gr.Dropdown(
|
57 |
choices=list(MODEL_TYPES.keys()),
|
58 |
label="Model Type",
|
59 |
-
value=
|
|
|
60 |
)
|
61 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
self.components["training_type"] = gr.Dropdown(
|
63 |
choices=list(TRAINING_TYPES.keys()),
|
64 |
label="Training Type",
|
@@ -198,45 +216,36 @@ class TrainTab(BaseTab):
|
|
198 |
|
199 |
def connect_events(self) -> None:
|
200 |
"""Connect event handlers to UI components"""
|
201 |
-
# Model type change event
|
202 |
-
def update_model_info(model, training_type):
|
203 |
-
params = self.get_default_params(MODEL_TYPES[model], TRAINING_TYPES[training_type])
|
204 |
-
info = self.get_model_info(model, training_type)
|
205 |
-
show_lora_params = training_type == list(TRAINING_TYPES.keys())[0] # Show if LoRA Finetune
|
206 |
-
|
207 |
-
return {
|
208 |
-
self.components["model_info"]: info,
|
209 |
-
self.components["train_steps"]: params["train_steps"],
|
210 |
-
self.components["batch_size"]: params["batch_size"],
|
211 |
-
self.components["learning_rate"]: params["learning_rate"],
|
212 |
-
self.components["save_iterations"]: params["save_iterations"],
|
213 |
-
self.components["lora_params_row"]: gr.Row(visible=show_lora_params)
|
214 |
-
}
|
215 |
-
|
216 |
self.components["model_type"].change(
|
|
|
|
|
|
|
|
|
217 |
fn=lambda v: self.app.update_ui_state(model_type=v),
|
218 |
inputs=[self.components["model_type"]],
|
219 |
outputs=[]
|
220 |
).then(
|
221 |
-
|
|
|
222 |
inputs=[self.components["model_type"], self.components["training_type"]],
|
223 |
-
outputs=[
|
224 |
-
self.components["model_info"],
|
225 |
-
self.components["train_steps"],
|
226 |
-
self.components["batch_size"],
|
227 |
-
self.components["learning_rate"],
|
228 |
-
self.components["save_iterations"],
|
229 |
-
self.components["lora_params_row"]
|
230 |
-
]
|
231 |
)
|
232 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
233 |
# Training type change event
|
234 |
self.components["training_type"].change(
|
235 |
fn=lambda v: self.app.update_ui_state(training_type=v),
|
236 |
inputs=[self.components["training_type"]],
|
237 |
outputs=[]
|
238 |
).then(
|
239 |
-
fn=update_model_info,
|
240 |
inputs=[self.components["model_type"], self.components["training_type"]],
|
241 |
outputs=[
|
242 |
self.components["model_info"],
|
@@ -248,7 +257,6 @@ class TrainTab(BaseTab):
|
|
248 |
]
|
249 |
)
|
250 |
|
251 |
-
|
252 |
# Add in the connect_events() method:
|
253 |
self.components["num_gpus"].change(
|
254 |
fn=lambda v: self.app.update_ui_state(num_gpus=v),
|
@@ -326,7 +334,9 @@ class TrainTab(BaseTab):
|
|
326 |
self.components["lora_params_row"],
|
327 |
self.components["num_gpus"],
|
328 |
self.components["precomputation_items"],
|
329 |
-
self.components["lr_warmup_steps"]
|
|
|
|
|
330 |
]
|
331 |
)
|
332 |
|
@@ -336,6 +346,7 @@ class TrainTab(BaseTab):
|
|
336 |
inputs=[
|
337 |
self.components["training_preset"],
|
338 |
self.components["model_type"],
|
|
|
339 |
self.components["training_type"],
|
340 |
self.components["lora_rank"],
|
341 |
self.components["lora_alpha"],
|
@@ -383,9 +394,19 @@ class TrainTab(BaseTab):
|
|
383 |
fn=lambda: self.app.training.delete_all_checkpoints(),
|
384 |
outputs=[self.components["status_box"]]
|
385 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
386 |
|
387 |
def handle_training_start(
|
388 |
-
self, preset, model_type,
|
|
|
|
|
389 |
):
|
390 |
"""Handle training start with proper log parser reset and checkpoint detection"""
|
391 |
# Safely reset log parser if it exists
|
@@ -396,9 +417,6 @@ class TrainTab(BaseTab):
|
|
396 |
from ..utils import TrainingLogParser
|
397 |
self.app.log_parser = TrainingLogParser()
|
398 |
|
399 |
-
# Initialize progress
|
400 |
-
#progress(0, desc="Initializing training")
|
401 |
-
|
402 |
# Check for latest checkpoint
|
403 |
checkpoints = list(OUTPUT_PATH.glob("checkpoint-*"))
|
404 |
resume_from = None
|
@@ -408,10 +426,6 @@ class TrainTab(BaseTab):
|
|
408 |
latest_checkpoint = max(checkpoints, key=os.path.getmtime)
|
409 |
resume_from = str(latest_checkpoint)
|
410 |
logger.info(f"Found checkpoint at {resume_from}, will resume training")
|
411 |
-
#progress(0.05, desc=f"Resuming from checkpoint {Path(resume_from).name}")
|
412 |
-
else:
|
413 |
-
#progress(0.05, desc="Starting new training run")
|
414 |
-
pass
|
415 |
|
416 |
# Convert model_type display name to internal name
|
417 |
model_internal_type = MODEL_TYPES.get(model_type)
|
@@ -432,9 +446,6 @@ class TrainTab(BaseTab):
|
|
432 |
precomputation_items = int(self.components["precomputation_items"].value)
|
433 |
lr_warmup_steps = int(self.components["lr_warmup_steps"].value)
|
434 |
|
435 |
-
# Progress update
|
436 |
-
#progress(0.1, desc="Preparing dataset")
|
437 |
-
|
438 |
# Start training (it will automatically use the checkpoint if provided)
|
439 |
try:
|
440 |
return self.app.training.start_training(
|
@@ -448,6 +459,7 @@ class TrainTab(BaseTab):
|
|
448 |
repo_id,
|
449 |
preset_name=preset,
|
450 |
training_type=training_internal_type,
|
|
|
451 |
resume_from_checkpoint=resume_from,
|
452 |
num_gpus=num_gpus,
|
453 |
precomputation_items=precomputation_items,
|
@@ -458,6 +470,52 @@ class TrainTab(BaseTab):
|
|
458 |
logger.exception("Error starting training")
|
459 |
return f"Error starting training: {str(e)}", f"Exception: {str(e)}\n\nCheck the logs for more details."
|
460 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
461 |
def get_model_info(self, model_type: str, training_type: str) -> str:
|
462 |
"""Get information about the selected model type and training method"""
|
463 |
if model_type == "HunyuanVideo":
|
@@ -483,14 +541,14 @@ class TrainTab(BaseTab):
|
|
483 |
else:
|
484 |
return base_info + "\n- Required VRAM: ~21GB minimum\n- Full model size: ~8GB"
|
485 |
|
486 |
-
elif model_type == "Wan
|
487 |
-
base_info = """### Wan
|
488 |
-
- Recommended batch size:
|
489 |
-
- Typical training time:
|
490 |
- Default resolution: 49x512x768"""
|
491 |
|
492 |
if training_type == "LoRA Finetune":
|
493 |
-
return base_info + "\n- Required VRAM:
|
494 |
else:
|
495 |
return base_info + "\n- **Full finetune not recommended due to VRAM requirements**"
|
496 |
|
@@ -601,6 +659,10 @@ class TrainTab(BaseTab):
|
|
601 |
precomputation_items_val = current_state.get("precomputation_items") if current_state.get("precomputation_items") != preset.get("precomputation_items", DEFAULT_PRECOMPUTATION_ITEMS) else preset.get("precomputation_items", DEFAULT_PRECOMPUTATION_ITEMS)
|
602 |
lr_warmup_steps_val = current_state.get("lr_warmup_steps") if current_state.get("lr_warmup_steps") != preset.get("lr_warmup_steps", DEFAULT_NB_LR_WARMUP_STEPS) else preset.get("lr_warmup_steps", DEFAULT_NB_LR_WARMUP_STEPS)
|
603 |
|
|
|
|
|
|
|
|
|
604 |
# Return values in the same order as the output components
|
605 |
return (
|
606 |
model_display_name,
|
@@ -615,9 +677,11 @@ class TrainTab(BaseTab):
|
|
615 |
gr.Row(visible=show_lora_params),
|
616 |
num_gpus_val,
|
617 |
precomputation_items_val,
|
618 |
-
lr_warmup_steps_val
|
|
|
619 |
)
|
620 |
-
|
|
|
621 |
def get_latest_status_message_and_logs(self) -> Tuple[str, str, str]:
|
622 |
"""Get latest status message, log content, and status code in a safer way"""
|
623 |
state = self.app.training.get_status()
|
|
|
5 |
import gradio as gr
|
6 |
import logging
|
7 |
import os
|
8 |
+
import json
|
9 |
from typing import Dict, Any, List, Optional, Tuple
|
10 |
from pathlib import Path
|
11 |
|
12 |
from vms.utils import BaseTab
|
13 |
from vms.config import (
|
14 |
+
OUTPUT_PATH, ASK_USER_TO_DUPLICATE_SPACE,
|
15 |
+
SMALL_TRAINING_BUCKETS,
|
16 |
+
TRAINING_PRESETS, TRAINING_TYPES, MODEL_TYPES, MODEL_VERSIONS,
|
17 |
DEFAULT_NB_TRAINING_STEPS, DEFAULT_SAVE_CHECKPOINT_EVERY_N_STEPS,
|
18 |
DEFAULT_BATCH_SIZE, DEFAULT_CAPTION_DROPOUT_P,
|
19 |
DEFAULT_LEARNING_RATE,
|
|
|
56 |
|
57 |
with gr.Row():
|
58 |
with gr.Column():
|
59 |
+
# Get the default model type from the first preset
|
60 |
+
default_model_type = list(MODEL_TYPES.keys())[0]
|
61 |
+
|
62 |
self.components["model_type"] = gr.Dropdown(
|
63 |
choices=list(MODEL_TYPES.keys()),
|
64 |
label="Model Type",
|
65 |
+
value=default_model_type,
|
66 |
+
interactive=True
|
67 |
)
|
68 |
+
|
69 |
+
# Get model versions for the default model type
|
70 |
+
default_model_versions = self.get_model_version_choices(default_model_type)
|
71 |
+
default_model_version = self.get_default_model_version(default_model_type)
|
72 |
+
|
73 |
+
self.components["model_version"] = gr.Dropdown(
|
74 |
+
choices=default_model_versions,
|
75 |
+
label="Model Version",
|
76 |
+
value=default_model_version,
|
77 |
+
interactive=True
|
78 |
+
)
|
79 |
+
|
80 |
self.components["training_type"] = gr.Dropdown(
|
81 |
choices=list(TRAINING_TYPES.keys()),
|
82 |
label="Training Type",
|
|
|
216 |
|
217 |
def connect_events(self) -> None:
|
218 |
"""Connect event handlers to UI components"""
|
219 |
+
# Model type change event - Update model version dropdown choices
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
220 |
self.components["model_type"].change(
|
221 |
+
fn=self.update_model_versions,
|
222 |
+
inputs=[self.components["model_type"]],
|
223 |
+
outputs=[self.components["model_version"]]
|
224 |
+
).then(
|
225 |
fn=lambda v: self.app.update_ui_state(model_type=v),
|
226 |
inputs=[self.components["model_type"]],
|
227 |
outputs=[]
|
228 |
).then(
|
229 |
+
# Use get_model_info instead of update_model_info
|
230 |
+
fn=self.get_model_info,
|
231 |
inputs=[self.components["model_type"], self.components["training_type"]],
|
232 |
+
outputs=[self.components["model_info"]]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
233 |
)
|
234 |
|
235 |
+
# Model version change event
|
236 |
+
self.components["model_version"].change(
|
237 |
+
fn=lambda v: self.app.update_ui_state(model_version=v),
|
238 |
+
inputs=[self.components["model_version"]],
|
239 |
+
outputs=[]
|
240 |
+
)
|
241 |
+
|
242 |
# Training type change event
|
243 |
self.components["training_type"].change(
|
244 |
fn=lambda v: self.app.update_ui_state(training_type=v),
|
245 |
inputs=[self.components["training_type"]],
|
246 |
outputs=[]
|
247 |
).then(
|
248 |
+
fn=self.update_model_info,
|
249 |
inputs=[self.components["model_type"], self.components["training_type"]],
|
250 |
outputs=[
|
251 |
self.components["model_info"],
|
|
|
257 |
]
|
258 |
)
|
259 |
|
|
|
260 |
# Add in the connect_events() method:
|
261 |
self.components["num_gpus"].change(
|
262 |
fn=lambda v: self.app.update_ui_state(num_gpus=v),
|
|
|
334 |
self.components["lora_params_row"],
|
335 |
self.components["num_gpus"],
|
336 |
self.components["precomputation_items"],
|
337 |
+
self.components["lr_warmup_steps"],
|
338 |
+
# Add model_version to the outputs
|
339 |
+
self.components["model_version"]
|
340 |
]
|
341 |
)
|
342 |
|
|
|
346 |
inputs=[
|
347 |
self.components["training_preset"],
|
348 |
self.components["model_type"],
|
349 |
+
self.components["model_version"], # Add model_version to the inputs
|
350 |
self.components["training_type"],
|
351 |
self.components["lora_rank"],
|
352 |
self.components["lora_alpha"],
|
|
|
394 |
fn=lambda: self.app.training.delete_all_checkpoints(),
|
395 |
outputs=[self.components["status_box"]]
|
396 |
)
|
397 |
+
|
398 |
+
def update_model_versions(self, model_type: str) -> Dict:
|
399 |
+
"""Update model version choices based on selected model type"""
|
400 |
+
model_versions = self.get_model_version_choices(model_type)
|
401 |
+
default_version = self.get_default_model_version(model_type)
|
402 |
+
|
403 |
+
# Update the model_version dropdown with new choices and default value
|
404 |
+
return gr.Dropdown(choices=model_versions, value=default_version)
|
405 |
|
406 |
def handle_training_start(
|
407 |
+
self, preset, model_type, model_version, training_type,
|
408 |
+
lora_rank, lora_alpha, train_steps, batch_size, learning_rate,
|
409 |
+
save_iterations, repo_id, progress=gr.Progress()
|
410 |
):
|
411 |
"""Handle training start with proper log parser reset and checkpoint detection"""
|
412 |
# Safely reset log parser if it exists
|
|
|
417 |
from ..utils import TrainingLogParser
|
418 |
self.app.log_parser = TrainingLogParser()
|
419 |
|
|
|
|
|
|
|
420 |
# Check for latest checkpoint
|
421 |
checkpoints = list(OUTPUT_PATH.glob("checkpoint-*"))
|
422 |
resume_from = None
|
|
|
426 |
latest_checkpoint = max(checkpoints, key=os.path.getmtime)
|
427 |
resume_from = str(latest_checkpoint)
|
428 |
logger.info(f"Found checkpoint at {resume_from}, will resume training")
|
|
|
|
|
|
|
|
|
429 |
|
430 |
# Convert model_type display name to internal name
|
431 |
model_internal_type = MODEL_TYPES.get(model_type)
|
|
|
446 |
precomputation_items = int(self.components["precomputation_items"].value)
|
447 |
lr_warmup_steps = int(self.components["lr_warmup_steps"].value)
|
448 |
|
|
|
|
|
|
|
449 |
# Start training (it will automatically use the checkpoint if provided)
|
450 |
try:
|
451 |
return self.app.training.start_training(
|
|
|
459 |
repo_id,
|
460 |
preset_name=preset,
|
461 |
training_type=training_internal_type,
|
462 |
+
model_version=model_version, # Pass the model version from dropdown
|
463 |
resume_from_checkpoint=resume_from,
|
464 |
num_gpus=num_gpus,
|
465 |
precomputation_items=precomputation_items,
|
|
|
470 |
logger.exception("Error starting training")
|
471 |
return f"Error starting training: {str(e)}", f"Exception: {str(e)}\n\nCheck the logs for more details."
|
472 |
|
473 |
+
def get_model_version_choices(self, model_type: str) -> List[str]:
|
474 |
+
"""Get model version choices based on model type"""
|
475 |
+
# Convert UI display name to internal name
|
476 |
+
internal_type = MODEL_TYPES.get(model_type)
|
477 |
+
if not internal_type or internal_type not in MODEL_VERSIONS:
|
478 |
+
return []
|
479 |
+
|
480 |
+
# Get versions and return them as choices
|
481 |
+
versions = MODEL_VERSIONS.get(internal_type, {})
|
482 |
+
return list(versions.keys())
|
483 |
+
|
484 |
+
def get_default_model_version(self, model_type: str) -> str:
|
485 |
+
"""Get default model version for the given model type"""
|
486 |
+
# Convert UI display name to internal name
|
487 |
+
internal_type = MODEL_TYPES.get(model_type)
|
488 |
+
if not internal_type or internal_type not in MODEL_VERSIONS:
|
489 |
+
return ""
|
490 |
+
|
491 |
+
# Get the first version available for this model type
|
492 |
+
versions = MODEL_VERSIONS.get(internal_type, {})
|
493 |
+
if versions:
|
494 |
+
return next(iter(versions.keys()))
|
495 |
+
|
496 |
+
return ""
|
497 |
+
|
498 |
+
def update_model_info(self, model_type: str, training_type: str) -> Dict:
|
499 |
+
"""Update model info and related UI components based on model type and training type"""
|
500 |
+
# Get model info text
|
501 |
+
model_info = self.get_model_info(model_type, training_type)
|
502 |
+
|
503 |
+
# Get default parameters for this model type and training type
|
504 |
+
params = self.get_default_params(MODEL_TYPES.get(model_type), TRAINING_TYPES.get(training_type))
|
505 |
+
|
506 |
+
# Check if LoRA params should be visible
|
507 |
+
show_lora_params = training_type == "LoRA Finetune"
|
508 |
+
|
509 |
+
# Return updates for UI components
|
510 |
+
return {
|
511 |
+
self.components["model_info"]: model_info,
|
512 |
+
self.components["train_steps"]: params["train_steps"],
|
513 |
+
self.components["batch_size"]: params["batch_size"],
|
514 |
+
self.components["learning_rate"]: params["learning_rate"],
|
515 |
+
self.components["save_iterations"]: params["save_iterations"],
|
516 |
+
self.components["lora_params_row"]: gr.Row(visible=show_lora_params)
|
517 |
+
}
|
518 |
+
|
519 |
def get_model_info(self, model_type: str, training_type: str) -> str:
|
520 |
"""Get information about the selected model type and training method"""
|
521 |
if model_type == "HunyuanVideo":
|
|
|
541 |
else:
|
542 |
return base_info + "\n- Required VRAM: ~21GB minimum\n- Full model size: ~8GB"
|
543 |
|
544 |
+
elif model_type == "Wan":
|
545 |
+
base_info = """### Wan
|
546 |
+
- Recommended batch size: 1-4
|
547 |
+
- Typical training time: 1-3 hours
|
548 |
- Default resolution: 49x512x768"""
|
549 |
|
550 |
if training_type == "LoRA Finetune":
|
551 |
+
return base_info + "\n- Required VRAM: ~16GB minimum\n- Default LoRA rank: 32 (~120 MB)"
|
552 |
else:
|
553 |
return base_info + "\n- **Full finetune not recommended due to VRAM requirements**"
|
554 |
|
|
|
659 |
precomputation_items_val = current_state.get("precomputation_items") if current_state.get("precomputation_items") != preset.get("precomputation_items", DEFAULT_PRECOMPUTATION_ITEMS) else preset.get("precomputation_items", DEFAULT_PRECOMPUTATION_ITEMS)
|
660 |
lr_warmup_steps_val = current_state.get("lr_warmup_steps") if current_state.get("lr_warmup_steps") != preset.get("lr_warmup_steps", DEFAULT_NB_LR_WARMUP_STEPS) else preset.get("lr_warmup_steps", DEFAULT_NB_LR_WARMUP_STEPS)
|
661 |
|
662 |
+
# Get the appropriate model version for the selected model type
|
663 |
+
model_versions = self.get_model_version_choices(model_display_name)
|
664 |
+
default_model_version = self.get_default_model_version(model_display_name)
|
665 |
+
|
666 |
# Return values in the same order as the output components
|
667 |
return (
|
668 |
model_display_name,
|
|
|
677 |
gr.Row(visible=show_lora_params),
|
678 |
num_gpus_val,
|
679 |
precomputation_items_val,
|
680 |
+
lr_warmup_steps_val,
|
681 |
+
gr.Dropdown(choices=model_versions, value=default_model_version)
|
682 |
)
|
683 |
+
|
684 |
+
|
685 |
def get_latest_status_message_and_logs(self) -> Tuple[str, str, str]:
|
686 |
"""Get latest status message, log content, and status code in a safer way"""
|
687 |
state = self.app.training.get_status()
|