Merge pull request #2354 from bmaltais/dev

v24.0.5

.release

@@ -1 +1 @@
-v24.0.4
+v24.0.5

README.md

@@ -42,7 +42,10 @@ The GUI allows you to set the training parameters and generate and run the requi
   - [SDXL training](#sdxl-training)
   - [Masked loss](#masked-loss)
   - [Change History](#change-history)
-    - [2024/04/25 (v24.0.4)](#20240425-v2404)
+    - [2024/04/19 (v24.0.5)](#20240419-v2405)
+    - [2024/04/18 (v24.0.4)](#20240418-v2404)
+      - [What's Changed](#whats-changed)
+      - [New Contributors](#new-contributors)
     - [2024/04/24 (v24.0.3)](#20240424-v2403)
     - [2024/04/24 (v24.0.2)](#20240424-v2402)
     - [2024/04/17 (v24.0.1)](#20240417-v2401)
@@ -409,9 +412,25 @@ ControlNet dataset is used to specify the mask. The mask images should be the RG
 
 ## Change History
 
-### 2024/04/25 (v24.0.4)
+### 2024/04/19 (v24.0.5)
 
-- ...
+- fdds
+
+### 2024/04/18 (v24.0.4)
+
+#### What's Changed
+
+- Fix options.md heading by @bmaltais in <https://github.com/bmaltais/kohya_ss/pull/2337>
+- Use correct file extensions when browsing for model file by @b-fission in <https://github.com/bmaltais/kohya_ss/pull/2323>
+- Add argument for Gradio's `root_path` to enable reverse proxy support by @hlky in <https://github.com/bmaltais/kohya_ss/pull/2333>
+- 2325 quotes wrapping python path cause subprocess cant find target in v2403 by @bmaltais in <https://github.com/bmaltais/kohya_ss/pull/2338>
+- 2330 another seemingly new data validation leads to unusable configs 2403 by @bmaltais in <https://github.com/bmaltais/kohya_ss/pull/2339>
+- Fix bad Lora parameters by @bmaltais in <https://github.com/bmaltais/kohya_ss/pull/2341>
+
+#### New Contributors
+
+- @b-fission made their first contribution in <https://github.com/bmaltais/kohya_ss/pull/2323>
+- @hlky made their first contribution in <https://github.com/bmaltais/kohya_ss/pull/2333>
 
 ### 2024/04/24 (v24.0.3)
 

kohya_gui/blip2_caption_gui.py

@@ -120,7 +120,7 @@ def generate_caption(
         output_file_path = os.path.splitext(file_path)[0] + caption_file_ext
 
         # Write the generated text to the output file
-        with open(output_file_path, "w") as output_file:
+        with open(output_file_path, "w", encoding="utf-8") as output_file:
             output_file.write(generated_text)
 
         # Log the image file path with a message about the fact that the caption was generated

kohya_gui/class_advanced_training.py

@@ -510,10 +510,11 @@ def list_state_dirs(path):
                 value=self.config.get("advanced.max_data_loader_n_workers", 0),
             )
         with gr.Row():
-            self.use_wandb = gr.Checkbox(
-                label="WANDB Logging",
-                value=self.config.get("advanced.use_wandb", False),
-                info="If unchecked, tensorboard will be used as the default for logging.",
+            self.log_with = gr.Dropdown(
+                label="Logging",
+                choices=["","wandb", "tensorboard","all"],
+                value="",
+                info="Loggers to use, tensorboard will be used as the default.",
             )
             self.wandb_api_key = gr.Textbox(
                 label="WANDB API Key",

kohya_gui/class_basic_training.py

@@ -98,15 +98,15 @@ def init_training_controls(self) -> None:
                 step=1,
                 # precision=0,
                 minimum=0,
-                value=self.config.get("basic.max_train_epochs", 1600),
+                value=self.config.get("basic.max_train_epochs", 0),
             )
             # Initialize the maximum train steps input
             self.max_train_steps = gr.Number(
                 label="Max train steps",
                 info="Overrides # training steps. 0 = no override",
                 step=1,
                 # precision=0,
-                value=self.config.get("basic.max_train_steps", 0),
+                value=self.config.get("basic.max_train_steps", 1600),
             )
             # Initialize the save every N epochs input
             self.save_every_n_epochs = gr.Number(

kohya_gui/class_gui_config.py

@@ -45,7 +45,7 @@ def save_config(self, config: dict, config_file_path: str = "./config.toml"):
         - config (dict): The configuration data to save.
         """
         # Write the configuration data to the TOML file
-        with open(f"{config_file_path}", "w") as f:
+        with open(f"{config_file_path}", "w", encoding="utf-8") as f:
             toml.dump(config, f)
 
     def get(self, key: str, default=None):

kohya_gui/class_sample_images.py

@@ -30,7 +30,7 @@ def create_prompt_file(sample_prompts, output_dir):
     """
     sample_prompts_path = os.path.join(output_dir, "prompt.txt")
 
-    with open(sample_prompts_path, "w") as f:
+    with open(sample_prompts_path, "w", encoding="utf-8") as f:
         f.write(sample_prompts)
 
     return sample_prompts_path

kohya_gui/class_tensorboard.py

@@ -94,9 +94,20 @@ def stop_tensorboard(self):
         return self.get_button_states(started=False)
 
     def gradio_interface(self):
+        try:
+            os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0'
+
+            import tensorflow # Attempt to import tensorflow to check if it is installed
+
+            visibility = True
+
+        except ImportError:
+            self.log.error("tensorflow is not installed, hiding the tensorboard button...")
+            visibility = False
+
         with gr.Row():
             button_start_tensorboard = gr.Button(
-                value="Start tensorboard", elem_id="myTensorButton"
+                value="Start tensorboard", elem_id="myTensorButton", visible=visibility
             )
             button_stop_tensorboard = gr.Button(
                 value="Stop tensorboard",

kohya_gui/common_gui.py

@@ -331,7 +331,6 @@ def update_my_data(my_data):
         "lr_warmup",
         "max_data_loader_n_workers",
         "max_train_epochs",
-        "max_train_steps",
         "save_every_n_epochs",
         "seed",
     ]:
@@ -352,6 +351,17 @@ def update_my_data(my_data):
             except ValueError:
                 # Handle the case where the string is not a valid float
                 my_data[key] = int(1)
+
+    for key in [
+        "max_train_steps",
+    ]:
+        value = my_data.get(key)
+        if value is not None:
+            try:
+                my_data[key] = int(value)
+            except ValueError:
+                # Handle the case where the string is not a valid float
+                my_data[key] = int(1600)
 
     # Convert values to int if they are strings
     for key in ["max_token_length"]:
@@ -406,7 +416,20 @@ def update_my_data(my_data):
             my_data["xformers"] = "xformers"
         else:
             my_data["xformers"] = "none"
-
+
+    # Convert use_wandb to log_with="wandb" if it is set to True
+    for key in ["use_wandb"]:
+        value = my_data.get(key)
+        if value is not None:
+            try:
+                if value == "True":
+                    my_data["log_with"] = "wandb"
+            except ValueError:
+                # Handle the case where the string is not a valid float
+                pass
+
+        my_data.pop(key, None)
+
     return my_data
 
 
@@ -743,13 +766,13 @@ def add_pre_postfix(
         # Check if the caption file does not exist
         if not os.path.exists(caption_file_path):
             # Create a new caption file with the specified prefix and/or postfix
-            with open(caption_file_path, "w", encoding="utf8") as f:
+            with open(caption_file_path, "w", encoding="utf-8") as f:
                 # Determine the separator based on whether both prefix and postfix are provided
                 separator = " " if prefix and postfix else ""
                 f.write(f"{prefix}{separator}{postfix}")
         else:
             # Open the existing caption file for reading and writing
-            with open(caption_file_path, "r+", encoding="utf8") as f:
+            with open(caption_file_path, "r+", encoding="utf-8") as f:
                 # Read the content of the caption file, stripping any trailing whitespace
                 content = f.read().rstrip()
                 # Move the file pointer to the beginning of the file
@@ -850,11 +873,11 @@ def find_replace(
         file_path = os.path.join(folder_path, caption_file)
         # Read and replace text
         try:
-            with open(file_path, "r", errors="ignore") as f:
+            with open(file_path, "r", errors="ignore", encoding="utf-8") as f:
                 content = f.read().replace(search_text, replace_text)
 
             # Write the updated content back to the file
-            with open(file_path, "w") as f:
+            with open(file_path, "w", encoding="utf-8") as f:
                 f.write(content)
         except Exception as e:
             log.error(f"Error processing file {file_path}: {e}")
@@ -1218,7 +1241,7 @@ def SaveConfigFile(
         log.info(f"Creating folder {folder_path} for the configuration file...")
 
     # Save the data to the specified JSON file
-    with open(file_path, "w") as file:
+    with open(file_path, "w", encoding="utf-8") as file:
         json.dump(variables, file, indent=2)
 
 
@@ -1242,7 +1265,7 @@ def save_to_file(content):
 
     # Append content to the specified file
     try:
-        with open(file_path, "a") as file:
+        with open(file_path, "a", encoding="utf-8") as file:
             file.write(content + "\n")
     except IOError as e:
         print(f"Error: Could not write to file - {e}")
@@ -1443,7 +1466,7 @@ def is_file_writable(file_path: str) -> bool:
 
     try:
         # Attempt to open the file in append mode to check if it can be written to
-        with open(file_path, "a"):
+        with open(file_path, "a", encoding="utf-8"):
             pass
         # If the file can be opened, it is considered writable
         return True
@@ -1463,7 +1486,7 @@ def print_command_and_toml(run_cmd, tmpfilename):
 
     log.info(f"Showing toml config file: {tmpfilename}")
     print("")
-    with open(tmpfilename, "r") as toml_file:
+    with open(tmpfilename, "r", encoding="utf-8") as toml_file:
         log.info(toml_file.read())
     log.info(f"end of toml config file: {tmpfilename}")
 

kohya_gui/dreambooth_gui.py

@@ -159,7 +159,7 @@ def save_configuration(
     save_every_n_steps,
     save_last_n_steps,
     save_last_n_steps_state,
-    use_wandb,
+    log_with,
     wandb_api_key,
     wandb_run_name,
     log_tracker_name,
@@ -317,7 +317,7 @@ def open_configuration(
     save_every_n_steps,
     save_last_n_steps,
     save_last_n_steps_state,
-    use_wandb,
+    log_with,
     wandb_api_key,
     wandb_run_name,
     log_tracker_name,
@@ -350,7 +350,7 @@ def open_configuration(
 
     if not file_path == "" and not file_path == None:
         # load variables from JSON file
-        with open(file_path, "r") as f:
+        with open(file_path, "r", encoding="utf-8") as f:
             my_data = json.load(f)
             log.info("Loading config...")
             # Update values to fix deprecated use_8bit_adam checkbox and set appropriate optimizer if it is set to True
@@ -470,7 +470,7 @@ def train_model(
     save_every_n_steps,
     save_last_n_steps,
     save_last_n_steps_state,
-    use_wandb,
+    log_with,
     wandb_api_key,
     wandb_run_name,
     log_tracker_name,
@@ -783,7 +783,7 @@ def train_model(
         ),
         "train_batch_size": train_batch_size,
         "train_data_dir": train_data_dir,
-        "use_wandb": use_wandb,
+        "log_with": log_with,
         "v2": v2,
         "v_parameterization": v_parameterization,
         "v_pred_like_loss": v_pred_like_loss if v_pred_like_loss != 0 else None,
@@ -810,7 +810,7 @@ def train_model(
 
     tmpfilename = "./outputs/tmpfiledbooth.toml"
     # Save the updated TOML data back to the file
-    with open(tmpfilename, "w") as toml_file:
+    with open(tmpfilename, "w", encoding="utf-8") as toml_file:
         toml.dump(config_toml_data, toml_file)
 
         if not os.path.exists(toml_file.name):
@@ -1056,7 +1056,7 @@ def dreambooth_tab(
             advanced_training.save_every_n_steps,
             advanced_training.save_last_n_steps,
             advanced_training.save_last_n_steps_state,
-            advanced_training.use_wandb,
+            advanced_training.log_with,
             advanced_training.wandb_api_key,
             advanced_training.wandb_run_name,
             advanced_training.log_tracker_name,

kohya_gui/finetune_gui.py

@@ -169,7 +169,7 @@ def save_configuration(
     save_every_n_steps,
     save_last_n_steps,
     save_last_n_steps_state,
-    use_wandb,
+    log_with,
     wandb_api_key,
     wandb_run_name,
     log_tracker_name,
@@ -335,7 +335,7 @@ def open_configuration(
     save_every_n_steps,
     save_last_n_steps,
     save_last_n_steps_state,
-    use_wandb,
+    log_with,
     wandb_api_key,
     wandb_run_name,
     log_tracker_name,
@@ -383,7 +383,7 @@ def open_configuration(
 
     if not file_path == "" and not file_path == None:
         # load variables from JSON file
-        with open(file_path, "r") as f:
+        with open(file_path, "r", encoding="utf-8") as f:
             my_data = json.load(f)
             log.info("Loading config...")
             # Update values to fix deprecated use_8bit_adam checkbox and set appropriate optimizer if it is set to True
@@ -507,7 +507,7 @@ def train_model(
     save_every_n_steps,
     save_last_n_steps,
     save_last_n_steps_state,
-    use_wandb,
+    log_with,
     wandb_api_key,
     wandb_run_name,
     log_tracker_name,
@@ -856,7 +856,7 @@ def train_model(
         "train_batch_size": train_batch_size,
         "train_data_dir": image_folder,
         "train_text_encoder": train_text_encoder,
-        "use_wandb": use_wandb,
+        "log_with": log_with,
         "v2": v2,
         "v_parameterization": v_parameterization,
         "v_pred_like_loss": v_pred_like_loss if v_pred_like_loss != 0 else None,
@@ -882,7 +882,7 @@ def train_model(
 
     tmpfilename = "./outputs/tmpfilefinetune.toml"
     # Save the updated TOML data back to the file
-    with open(tmpfilename, "w") as toml_file:
+    with open(tmpfilename, "w", encoding="utf-8") as toml_file:
         toml.dump(config_toml_data, toml_file)
 
         if not os.path.exists(toml_file.name):
@@ -1209,7 +1209,7 @@ def list_presets(path):
             advanced_training.save_every_n_steps,
             advanced_training.save_last_n_steps,
             advanced_training.save_last_n_steps_state,
-            advanced_training.use_wandb,
+            advanced_training.log_with,
             advanced_training.wandb_api_key,
             advanced_training.wandb_run_name,
             advanced_training.log_tracker_name,
@@ -1322,6 +1322,6 @@ def list_presets(path):
         gr.Markdown("This section provide Various Finetuning guides and information...")
         top_level_path = rf'"{scriptdir}/docs/Finetuning/top_level.md"'
         if os.path.exists(top_level_path):
-            with open(os.path.join(top_level_path), "r", encoding="utf8") as file:
+            with open(os.path.join(top_level_path), "r", encoding="utf-8") as file:
                 guides_top_level = file.read() + "\n"
             gr.Markdown(guides_top_level)

kohya_gui/localization.py

@@ -20,7 +20,7 @@ def load_language_js(language_name: str) -> str:
     data = {}
     if fn is not None:
         try:
-            with open(fn, "r", encoding="utf8") as file:
+            with open(fn, "r", encoding="utf-8") as file:
                 data = json.load(file)
         except Exception:
             logging.ERROR(f"Error loading localization from {fn}")