VideoModelStudio / vms /ui /video_trainer_ui.py
jbilcke-hf's picture
jbilcke-hf HF Staff
fix for Finetrainers
66c6879
raw
history blame
15.3 kB
import platform
import gradio as gr
from pathlib import Path
import logging
import asyncio
from typing import Any, Optional, Dict, List, Union, Tuple
from ..services import TrainingService, CaptioningService, SplittingService, ImportService
from ..config import (
STORAGE_PATH, VIDEOS_TO_SPLIT_PATH, STAGING_PATH, OUTPUT_PATH,
TRAINING_PATH, LOG_FILE_PATH, TRAINING_PRESETS, TRAINING_VIDEOS_PATH, MODEL_PATH, OUTPUT_PATH,
MODEL_TYPES, SMALL_TRAINING_BUCKETS
)
from ..utils import count_media_files, format_media_title, TrainingLogParser
from ..tabs import ImportTab, SplitTab, CaptionTab, TrainTab, ManageTab
logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)
httpx_logger = logging.getLogger('httpx')
httpx_logger.setLevel(logging.WARN)
class VideoTrainerUI:
def __init__(self):
"""Initialize services and tabs"""
# Initialize core services
self.trainer = TrainingService()
self.splitter = SplittingService()
self.importer = ImportService()
self.captioner = CaptioningService()
# Recovery status from any interrupted training
recovery_result = self.trainer.recover_interrupted_training()
# Add null check for recovery_result
if recovery_result is None:
recovery_result = {"status": "unknown", "ui_updates": {}}
self.recovery_status = recovery_result.get("status", "unknown")
self.ui_updates = recovery_result.get("ui_updates", {})
# Initialize log parser
self.log_parser = TrainingLogParser()
# Shared state for tabs
self.state = {
"recovery_result": recovery_result
}
# Initialize tabs dictionary (will be populated in create_ui)
self.tabs = {}
self.tabs_component = None
# Log recovery status
logger.info(f"Initialization complete. Recovery status: {self.recovery_status}")
def create_ui(self):
"""Create the main Gradio UI"""
with gr.Blocks(title="🎥 Video Model Studio") as app:
gr.Markdown("# 🎥 Video Model Studio")
# Create main tabs component
with gr.Tabs() as self.tabs_component:
# Initialize tab objects
self.tabs["import_tab"] = ImportTab(self)
self.tabs["split_tab"] = SplitTab(self)
self.tabs["caption_tab"] = CaptionTab(self)
self.tabs["train_tab"] = TrainTab(self)
self.tabs["manage_tab"] = ManageTab(self)
# Create tab UI components
for tab_id, tab_obj in self.tabs.items():
tab_obj.create(self.tabs_component)
# Connect event handlers
for tab_id, tab_obj in self.tabs.items():
tab_obj.connect_events()
# Add app-level timers for auto-refresh functionality
self._add_timers()
# Initialize app state on load
app.load(
fn=self.initialize_app_state,
outputs=[
self.tabs["split_tab"].components["video_list"],
self.tabs["caption_tab"].components["training_dataset"],
self.tabs["train_tab"].components["start_btn"],
self.tabs["train_tab"].components["stop_btn"],
self.tabs["train_tab"].components["pause_resume_btn"],
self.tabs["train_tab"].components["training_preset"],
self.tabs["train_tab"].components["model_type"],
self.tabs["train_tab"].components["lora_rank"],
self.tabs["train_tab"].components["lora_alpha"],
self.tabs["train_tab"].components["num_epochs"],
self.tabs["train_tab"].components["batch_size"],
self.tabs["train_tab"].components["learning_rate"],
self.tabs["train_tab"].components["save_iterations"]
]
)
return app
def _add_timers(self):
"""Add auto-refresh timers to the UI"""
# Status update timer (every 1 second)
status_timer = gr.Timer(value=1)
# Use a safer approach - check if the component exists before using it
outputs = [
self.tabs["train_tab"].components["status_box"],
self.tabs["train_tab"].components["log_box"],
self.tabs["train_tab"].components["start_btn"],
self.tabs["train_tab"].components["stop_btn"]
]
# Add delete_checkpoints_btn only if it exists
if "delete_checkpoints_btn" in self.tabs["train_tab"].components:
outputs.append(self.tabs["train_tab"].components["delete_checkpoints_btn"])
else:
# Add pause_resume_btn as fallback
outputs.append(self.tabs["train_tab"].components["pause_resume_btn"])
status_timer.tick(
fn=self.tabs["train_tab"].get_latest_status_message_logs_and_button_labels,
outputs=outputs
)
# Dataset refresh timer (every 5 seconds)
dataset_timer = gr.Timer(value=5)
dataset_timer.tick(
fn=self.refresh_dataset,
outputs=[
self.tabs["split_tab"].components["video_list"],
self.tabs["caption_tab"].components["training_dataset"]
]
)
# Titles update timer (every 6 seconds)
titles_timer = gr.Timer(value=6)
titles_timer.tick(
fn=self.update_titles,
outputs=[
self.tabs["split_tab"].components["split_title"],
self.tabs["caption_tab"].components["caption_title"],
self.tabs["train_tab"].components["train_title"]
]
)
def initialize_app_state(self):
"""Initialize all app state in one function to ensure correct output count"""
# Get dataset info
video_list = self.tabs["split_tab"].list_unprocessed_videos()
training_dataset = self.tabs["caption_tab"].list_training_files_to_caption()
# Get button states based on recovery status
button_states = self.get_initial_button_states()
start_btn = button_states[0]
stop_btn = button_states[1]
delete_checkpoints_btn = button_states[2] # This replaces pause_resume_btn in the response tuple
# Get UI form values - possibly from the recovery
if self.recovery_status in ["recovered", "ready_to_recover", "running"] and "ui_updates" in self.state["recovery_result"]:
recovery_ui = self.state["recovery_result"]["ui_updates"]
# If we recovered training parameters from the original session
ui_state = {}
# Handle model_type specifically - could be internal or display name
if "model_type" in recovery_ui:
model_type_value = recovery_ui["model_type"]
# If it's an internal name, convert to display name
if model_type_value not in MODEL_TYPES:
# Find the display name for this internal model type
for display_name, internal_name in MODEL_TYPES.items():
if internal_name == model_type_value:
model_type_value = display_name
logger.info(f"Converted internal model type '{recovery_ui['model_type']}' to display name '{model_type_value}'")
break
ui_state["model_type"] = model_type_value
# Copy other parameters
for param in ["lora_rank", "lora_alpha", "num_epochs",
"batch_size", "learning_rate", "save_iterations", "training_preset"]:
if param in recovery_ui:
ui_state[param] = recovery_ui[param]
# Merge with existing UI state if needed
if ui_state:
current_state = self.load_ui_values()
current_state.update(ui_state)
self.trainer.save_ui_state(current_state)
logger.info(f"Updated UI state from recovery: {ui_state}")
# Load values (potentially with recovery updates applied)
ui_state = self.load_ui_values()
# Ensure model_type is a display name, not internal name
model_type_val = ui_state.get("model_type", list(MODEL_TYPES.keys())[0])
if model_type_val not in MODEL_TYPES:
# Convert from internal to display name
for display_name, internal_name in MODEL_TYPES.items():
if internal_name == model_type_val:
model_type_val = display_name
break
training_preset = ui_state.get("training_preset", list(TRAINING_PRESETS.keys())[0])
lora_rank_val = ui_state.get("lora_rank", "128")
lora_alpha_val = ui_state.get("lora_alpha", "128")
num_epochs_val = int(ui_state.get("num_epochs", 70))
batch_size_val = int(ui_state.get("batch_size", 1))
learning_rate_val = float(ui_state.get("learning_rate", 3e-5))
save_iterations_val = int(ui_state.get("save_iterations", 500))
# Return all values in the exact order expected by outputs
return (
video_list,
training_dataset,
start_btn,
stop_btn,
delete_checkpoints_btn,
training_preset,
model_type_val,
lora_rank_val,
lora_alpha_val,
num_epochs_val,
batch_size_val,
learning_rate_val,
save_iterations_val
)
def initialize_ui_from_state(self):
"""Initialize UI components from saved state"""
ui_state = self.load_ui_values()
# Return values in order matching the outputs in app.load
return (
ui_state.get("training_preset", list(TRAINING_PRESETS.keys())[0]),
ui_state.get("model_type", list(MODEL_TYPES.keys())[0]),
ui_state.get("lora_rank", "128"),
ui_state.get("lora_alpha", "128"),
ui_state.get("num_epochs", 70),
ui_state.get("batch_size", 1),
ui_state.get("learning_rate", 3e-5),
ui_state.get("save_iterations", 500)
)
def update_ui_state(self, **kwargs):
"""Update UI state with new values"""
current_state = self.trainer.load_ui_state()
current_state.update(kwargs)
self.trainer.save_ui_state(current_state)
# Don't return anything to avoid Gradio warnings
return None
def load_ui_values(self):
"""Load UI state values for initializing form fields"""
ui_state = self.trainer.load_ui_state()
# Ensure proper type conversion for numeric values
ui_state["lora_rank"] = ui_state.get("lora_rank", "128")
ui_state["lora_alpha"] = ui_state.get("lora_alpha", "128")
ui_state["num_epochs"] = int(ui_state.get("num_epochs", 70))
ui_state["batch_size"] = int(ui_state.get("batch_size", 1))
ui_state["learning_rate"] = float(ui_state.get("learning_rate", 3e-5))
ui_state["save_iterations"] = int(ui_state.get("save_iterations", 500))
return ui_state
# Add this new method to get initial button states:
def get_initial_button_states(self):
"""Get the initial states for training buttons based on recovery status"""
recovery_result = self.state.get("recovery_result") or self.trainer.recover_interrupted_training()
ui_updates = recovery_result.get("ui_updates", {})
# Check for checkpoints to determine start button text
has_checkpoints = len(list(OUTPUT_PATH.glob("checkpoint-*"))) > 0
# Default button states if recovery didn't provide any
if not ui_updates or not ui_updates.get("start_btn"):
is_training = self.trainer.is_training_running()
if is_training:
# Active training detected
start_btn_props = {"interactive": False, "variant": "secondary", "value": "Continue Training" if has_checkpoints else "Start Training"}
stop_btn_props = {"interactive": True, "variant": "primary", "value": "Stop at Last Checkpoint"}
delete_btn_props = {"interactive": False, "variant": "stop", "value": "Delete All Checkpoints"}
else:
# No active training
start_btn_props = {"interactive": True, "variant": "primary", "value": "Continue Training" if has_checkpoints else "Start Training"}
stop_btn_props = {"interactive": False, "variant": "secondary", "value": "Stop at Last Checkpoint"}
delete_btn_props = {"interactive": has_checkpoints, "variant": "stop", "value": "Delete All Checkpoints"}
else:
# Use button states from recovery
start_btn_props = ui_updates.get("start_btn", {"interactive": True, "variant": "primary", "value": "Start Training"})
stop_btn_props = ui_updates.get("stop_btn", {"interactive": False, "variant": "secondary", "value": "Stop at Last Checkpoint"})
delete_btn_props = ui_updates.get("delete_checkpoints_btn", {"interactive": has_checkpoints, "variant": "stop", "value": "Delete All Checkpoints"})
# Return button states in the correct order
return (
gr.Button(**start_btn_props),
gr.Button(**stop_btn_props),
gr.Button(**delete_btn_props)
)
def update_titles(self) -> Tuple[Any]:
"""Update all dynamic titles with current counts
Returns:
Dict of Gradio updates
"""
# Count files for splitting
split_videos, _, split_size = count_media_files(VIDEOS_TO_SPLIT_PATH)
split_title = format_media_title(
"split", split_videos, 0, split_size
)
# Count files for captioning
caption_videos, caption_images, caption_size = count_media_files(STAGING_PATH)
caption_title = format_media_title(
"caption", caption_videos, caption_images, caption_size
)
# Count files for training
train_videos, train_images, train_size = count_media_files(TRAINING_VIDEOS_PATH)
train_title = format_media_title(
"train", train_videos, train_images, train_size
)
return (
gr.Markdown(value=split_title),
gr.Markdown(value=caption_title),
gr.Markdown(value=f"{train_title} available for training")
)
def refresh_dataset(self):
"""Refresh all dynamic lists and training state"""
video_list = self.tabs["split_tab"].list_unprocessed_videos()
training_dataset = self.tabs["caption_tab"].list_training_files_to_caption()
return (
video_list,
training_dataset
)