arbor-ai 0.2.2__py3-none-any.whl → 0.2.4__py3-none-any.whl

arbor/cli.py

@@ -4,6 +4,7 @@ from datetime import datetime
 import click
 import uvicorn
 
+from arbor.client.arbor_client import create_app
 from arbor.server.core.config import Config
 from arbor.server.core.config_manager import ConfigManager
 from arbor.server.main import app
@@ -13,13 +14,92 @@ from arbor.server.services.health_manager import HealthManager
 from arbor.server.services.inference_manager import InferenceManager
 from arbor.server.services.job_manager import JobManager
 from arbor.server.services.training_manager import TrainingManager
-from arbor.client.arbor_client import create_app
+from arbor.server.utils.logging import (
+    get_logger,
+    log_configuration,
+    log_system_info,
+    setup_logging,
+)
+
+
+def make_log_dir(storage_path: str):
+    # Create a timestamped log directory under the storage path
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    log_dir = os.path.join(storage_path, "logs", timestamp)
+    os.makedirs(log_dir, exist_ok=True)
+    return log_dir
 
 
 @click.group()
 def cli():
     pass
 
+
+def create_app(arbor_config_path: str):
+    """Create and configure the Arbor API application
+
+    Args:
+        arbor_config_path (str): Path to config file
+
+    Returns:
+        FastAPI: Configured FastAPI application
+    """
+    # Create new settings instance with overrides
+    config = Config.load_config_from_yaml(arbor_config_path)
+    log_dir = make_log_dir(config.STORAGE_PATH)
+    app.state.log_dir = log_dir
+
+    # Setup logging
+    logging_config = setup_logging(
+        log_level="INFO",
+        log_dir=log_dir,
+        enable_file_logging=True,
+        enable_console_logging=True,
+    )
+
+    # Log configuration and system info
+    log_configuration(logging_config)
+
+    # Get logger for this module
+    logger = get_logger(__name__)
+    logger.info("Initializing Arbor application...")
+
+    # Log system information via health manager
+    health_manager = HealthManager(config=config)
+    try:
+        versions = config.get_system_versions()
+        logger.info("System versions:")
+        for category, version_info in versions.items():
+            if isinstance(version_info, dict):
+                logger.info(f"  {category}:")
+                for lib, version in version_info.items():
+                    logger.info(f"    {lib}: {version}")
+            else:
+                logger.info(f"  {category}: {version_info}")
+    except Exception as e:
+        logger.warning(f"Could not log system versions: {e}")
+
+    # Initialize services with settings
+    logger.info("Initializing services...")
+    file_manager = FileManager(config=config)
+    job_manager = JobManager(config=config)
+    training_manager = TrainingManager(config=config)
+    inference_manager = InferenceManager(config=config)
+    grpo_manager = GRPOManager(config=config)
+
+    # Inject settings into app state
+    app.state.config = config
+    app.state.file_manager = file_manager
+    app.state.job_manager = job_manager
+    app.state.training_manager = training_manager
+    app.state.inference_manager = inference_manager
+    app.state.grpo_manager = grpo_manager
+    app.state.health_manager = health_manager
+
+    logger.info("Arbor application initialized successfully")
+    return app
+
+
 def start_server(host="0.0.0.0", port=7453, storage_path="./storage", timeout=10):
     """Start the Arbor API server with a single function call"""
     import socket
@@ -93,6 +173,7 @@ def serve(host, port, arbor_config):
         # configure_uvicorn_logging()
         uvicorn.run(app, host=host, port=port)
     except Exception as e:
+
         click.echo(f"Failed to start server: {e}", err=True)
         raise click.Abort()
 

arbor/server/services/inference/vllm_serve.py

@@ -352,7 +352,7 @@ class ScriptArguments:
         enable_prefix_caching (`bool` or `None`, *optional*, defaults to `None`):
             Whether to enable prefix caching in vLLM. If set to `True`, ensure that the model and the hardware support
             this feature.
-        enforce_eager (`bool` or `None`, *optional*, defaults to `None`):
+        enforce_eager (`bool` or `None`, *optional*, defaults to `False`):
             Whether to enforce eager execution. If set to `True`, we will disable CUDA graph and always execute the
             model in eager mode. If `False` (default behavior), we will use CUDA graph and eager execution in hybrid.
         kv_cache_dtype (`str`, *optional*, defaults to `"auto"`):
@@ -423,7 +423,7 @@ class ScriptArguments:
         },
     )
     enforce_eager: Optional[bool] = field(
-        default=None,
+        default=False,
         metadata={
             "help": "Whether to enforce eager execution. If set to `True`, we will disable CUDA graph and always "
             "execute the model in eager mode. If `False` (default behavior), we will use CUDA graph and eager "

{arbor_ai-0.2.2.dist-info → arbor_ai-0.2.4.dist-info}/METADATA

@@ -1,11 +1,11 @@
 Metadata-Version: 2.4
 Name: arbor-ai
-Version: 0.2.2
+Version: 0.2.4
 Summary: A framework for fine-tuning and managing language models
 Author-email: Noah Ziems <nziems2@nd.edu>
 Project-URL: Homepage, https://github.com/Ziems/arbor
 Project-URL: Issues, https://github.com/Ziems/arbor/issues
-Requires-Python: >=3.10
+Requires-Python: >=3.11
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: torch>=2.6.0
@@ -79,6 +79,14 @@ export NCCL_P2P_DISABLE=1
 export NCCL_IB_DISABLE=1
 ```
 
+**NVCC**
+If you run into issues, double check that you have [nvcc](https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/) installed:
+```bash
+nvcc --version
+```
+If you don't have admin permissions, you can often install nvcc using conda.
+
+
 ## 🙏 Acknowledgements
 
 Arbor builds on the shoulders of great work. We extend our thanks to:

{arbor_ai-0.2.2.dist-info → arbor_ai-0.2.4.dist-info}/RECORD

@@ -1,5 +1,5 @@
 arbor/__init__.py,sha256=DKqMupRY_1oOt7WfwU_6g4FVbTBJjU83p7yEGC-JAhs,458
-arbor/cli.py,sha256=YK-MIFxWb4ZfNb-UtK6t-EgTk1XQRq0dVOzdGf7_AY8,4867
+arbor/cli.py,sha256=wTyAK1uN7X9r4rinQVBhNJ8LaIj_100X9EF_Csw5Yzk,7453
 arbor/client/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 arbor/client/api.py,sha256=86bgHuGM_AvI1Uhic_QaCnpF4VFqXie9ZzxmbTXUPpQ,19
 arbor/client/arbor_client.py,sha256=laGmeTMyGzwKOBGjpOWEZdDXTwsp_MUhp6umiLufncs,8320
@@ -28,7 +28,7 @@ arbor/server/services/comms/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMp
 arbor/server/services/comms/comms.py,sha256=zTV1Wo4FGnOcl5tIjBOfcuvDPMu-VTSKBlDBXV4h6LY,7833
 arbor/server/services/inference/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 arbor/server/services/inference/vllm_client.py,sha256=EwVuv4bUSK2rOGEL6YdGSfaYtA4NsZfTwWw_KKfMAiQ,18367
-arbor/server/services/inference/vllm_serve.py,sha256=UZAGo7CyshR3-9fhXCTKhXeidqNqbY6LyU9DDNiX_Sw,109543
+arbor/server/services/inference/vllm_serve.py,sha256=etg7EhihRoPGl4fE-g0I5wYIhOEwLy2jC9mpgB3Agbw,109545
 arbor/server/services/scripts/dpo_training.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 arbor/server/services/scripts/grpo_training.py,sha256=XJZ3PBJdXRP__XfiQBtqEqjRE4YKjjYrJ4ze5W9SRnY,22160
 arbor/server/services/scripts/mmgrpo_training.py,sha256=aDaTGkyfceA-jhAdzfpehAzuSSGIT8gdfD_5z9IoBEk,19373
@@ -43,9 +43,9 @@ arbor/server/services/scripts/utils/mock_server.py,sha256=NX38oBtxEVS9-kreayTCSz
 arbor/server/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 arbor/server/utils/helpers.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 arbor/server/utils/logging.py,sha256=UsG3cYgNk04pXadf5-NcYxvI8qYjB6o91N_P08FuoIg,10368
-arbor_ai-0.2.2.dist-info/licenses/LICENSE,sha256=5vFGrbOFeXXM83JV9o16w7ohH4WLeu3-57GocJSz8ow,1067
-arbor_ai-0.2.2.dist-info/METADATA,sha256=1Jw-RJostZFj8mR9mIz_p8LLLpyh7qKZSPmrRn0Otzw,2776
-arbor_ai-0.2.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-arbor_ai-0.2.2.dist-info/entry_points.txt,sha256=PGBX-MfNwfIl8UPFgsX3gjtXLqSogRhOktKMpZUysD0,40
-arbor_ai-0.2.2.dist-info/top_level.txt,sha256=jzWdp3BRYqvZDMFsPajrcftvvlluzVDErkD8IMRfhYs,6
-arbor_ai-0.2.2.dist-info/RECORD,,
+arbor_ai-0.2.4.dist-info/licenses/LICENSE,sha256=5vFGrbOFeXXM83JV9o16w7ohH4WLeu3-57GocJSz8ow,1067
+arbor_ai-0.2.4.dist-info/METADATA,sha256=Tb9KSDqEDVsueaPIAJk80g1dGlHjeIiF7bb0mQGEFNM,3017
+arbor_ai-0.2.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+arbor_ai-0.2.4.dist-info/entry_points.txt,sha256=PGBX-MfNwfIl8UPFgsX3gjtXLqSogRhOktKMpZUysD0,40
+arbor_ai-0.2.4.dist-info/top_level.txt,sha256=jzWdp3BRYqvZDMFsPajrcftvvlluzVDErkD8IMRfhYs,6
+arbor_ai-0.2.4.dist-info/RECORD,,