Refactor project structure and enhance functionality

- Updated .gitignore to simplify ignored paths.
- Updated LTX submodule to the latest commit for improved features.
- Added 'imageio' to requirements.txt for enhanced image processing capabilities.
- Introduced a new client.py file for making predictions via the API.
- Enhanced logging in ltx_api.log to capture server start attempts and errors.
- Improved LTXVideoSettings in ltx_settings.py by adding model_revision for stable model downloads.

These changes aim to streamline the project setup, improve API functionality, and enhance logging for better troubleshooting.

.gitignore

@@ -1 +1 @@
-api/checkpoints
+checkpoints

api/client.py

@@ -0,0 +1,18 @@
+
+# Copyright The Lightning AI team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import requests
+
+response = requests.post("http://127.0.0.1:8000/predict", json={"input": 4.0})
+print(f"Status: {response.status_code}\nResponse:\n {response.text}")

api/logs/ltx_api.log

@@ -23,3 +23,10 @@
 2024-11-28 19:09:10.161 | INFO     | __main__:main:348 - Starting LTX video generation server on port 8000
 2024-11-28 19:11:22.845 | INFO     | __main__:main:356 - Starting LTX video generation server on port 8000
 2024-11-28 19:13:03.394 | INFO     | __main__:main:356 - Starting LTX video generation server on port 8000
+2024-12-16 12:54:28.451 | INFO     | __main__:main:356 - Starting LTX video generation server on port 8000
+2024-12-16 12:58:14.459 | ERROR    | __main__:main:360 - Server failed to start: One or more workers failed to start. Shutting down LitServe
+2024-12-16 12:59:43.351 | INFO     | __main__:main:356 - Starting LTX video generation server on port 8000
+2024-12-16 12:59:46.614 | ERROR    | __main__:main:360 - Server failed to start: One or more workers failed to start. Shutting down LitServe
+2024-12-16 13:00:02.774 | INFO     | __main__:main:356 - Starting LTX video generation server on port 8000
+2024-12-16 13:03:49.870 | ERROR    | __main__:main:360 - Server failed to start: One or more workers failed to start. Shutting down LitServe
+2024-12-16 13:15:44.243 | INFO     | __main__:main:356 - Starting LTX video generation server on port 8000

configs/ltx_settings.py

@@ -64,6 +64,7 @@ class LTXVideoSettings(BaseSettings):
         device (str): Device for inference ('cuda' or 'cpu')
         prompt (Optional[str]): Generation prompt text
         negative_prompt (str): Negative prompt for undesired features
+        model_revision (str): Specific model revision to use for stability
 
     Example:
         >>> settings = LTXVideoSettings(
@@ -81,7 +82,7 @@ class LTXVideoSettings(BaseSettings):
         description="HuggingFace model ID"
     )
     ckpt_dir: Path = Field(
-        default_factory=lambda: Path(os.getenv('LTX_CKPT_DIR', 'checkpoints')),
+        default_factory=lambda: Path(os.getenv('LTX_CKPT_DIR', '../checkpoints')),
         description="Directory containing model checkpoints"
     )
     use_auth_token: Optional[str] = Field(
@@ -137,6 +138,12 @@ class LTXVideoSettings(BaseSettings):
     MAX_WIDTH: int = 1280
     MAX_NUM_FRAMES: int = 257
 
+    # Add model revision to ensure stable downloads
+    model_revision: str = Field(
+        default="f1994f6731091f828ecc135923c978155928c031",
+        description="Specific model revision to use for stability"
+    )
+
     def download_model(self) -> Path:
         """
         Download model from HuggingFace Hub if not already present.
@@ -168,10 +175,11 @@ def download_model(self) -> Path:
             # Create checkpoint directory if it doesn't exist
             self.ckpt_dir.mkdir(parents=True, exist_ok=True)
 
-            # Download model from HuggingFace
-            logger.info(f"Downloading model {self.model_id} to {self.ckpt_dir}")
+            # Download model from HuggingFace with specific revision
+            logger.info(f"Downloading model {self.model_id} (revision: {self.model_revision}) to {self.ckpt_dir}")
             snapshot_download(
                 repo_id=self.model_id,
+                revision=self.model_revision,  # Use specific working revision
                 local_dir=self.ckpt_dir,
                 local_dir_use_symlinks=False,
                 repo_type='model',

requirements.txt

@@ -10,3 +10,4 @@ boto3
 prometheus-client
 loguru
 bitsandbytes
+imageio