isa-model 0.3.7__tar.gz → 0.3.8__tar.gz

{isa_model-0.3.7 → isa_model-0.3.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: isa_model
-Version: 0.3.7
+Version: 0.3.8
 Summary: Unified AI model serving framework
 Author: isA_Model Contributors
 Classifier: Development Status :: 3 - Alpha

{isa_model-0.3.7 → isa_model-0.3.8}/isa_model/client.py

@@ -138,6 +138,7 @@ class ISAModelClient:
         model_hint: Optional[str] = None,
         provider_hint: Optional[str] = None,
         stream: bool = False,
+        tools: Optional[List[Any]] = None,
         **kwargs
     ) -> Union[Dict[str, Any], object]:
         """
@@ -150,6 +151,7 @@ class ISAModelClient:
             model_hint: Optional model preference
             provider_hint: Optional provider preference
             stream: Enable streaming for text services (returns AsyncGenerator)
+            tools: Optional list of tools for function calling (only for text services)
             **kwargs: Additional task-specific parameters
             
         Returns:
@@ -174,6 +176,13 @@ class ISAModelClient:
             async for token in await client.invoke("Hello", "chat", "text", stream=True):
                 print(token, end="", flush=True)
             
+            # Text with tools
+            await client.invoke("What's 5+3?", "chat", "text", tools=[calculator_function])
+            
+            # Streaming with tools
+            async for token in await client.invoke("What's 5+3?", "chat", "text", stream=True, tools=[calculator_function]):
+                print(token, end="")
+            
             # Image generation
             await client.invoke("A beautiful sunset", "generate_image", "image")
             
@@ -193,6 +202,7 @@ class ISAModelClient:
                         service_type=service_type,
                         model_hint=model_hint,
                         provider_hint=provider_hint,
+                        tools=tools,
                         **kwargs
                     )
                 else:
@@ -202,6 +212,7 @@ class ISAModelClient:
                         service_type=service_type,
                         model_hint=model_hint,
                         provider_hint=provider_hint,
+                        tools=tools,
                         **kwargs
                     )
             
@@ -213,6 +224,7 @@ class ISAModelClient:
                     service_type=service_type,
                     model_hint=model_hint,
                     provider_hint=provider_hint,
+                    tools=tools,
                     **kwargs
                 )
             else:
@@ -222,6 +234,7 @@ class ISAModelClient:
                     service_type=service_type,
                     model_hint=model_hint,
                     provider_hint=provider_hint,
+                    tools=tools,
                     **kwargs
                 )
                 
@@ -349,7 +362,8 @@ class ISAModelClient:
         service_type: str,
         model_name: str,
         provider: str,
-        task: str
+        task: str,
+        tools: Optional[List[Any]] = None
     ) -> Any:
         """Get appropriate service instance"""
         
@@ -357,7 +371,11 @@ class ISAModelClient:
         
         # Check cache first
         if cache_key in self._service_cache:
-            return self._service_cache[cache_key]
+            service = self._service_cache[cache_key]
+            # If tools are needed, bind them to the service
+            if tools and service_type == "text":
+                return service.bind_tools(tools)
+            return service
         
         try:
             # Route to appropriate AIFactory method
@@ -387,6 +405,11 @@ class ISAModelClient:
             
             # Cache the service
             self._service_cache[cache_key] = service
+            
+            # If tools are needed, bind them to the service
+            if tools and service_type == "text":
+                return service.bind_tools(tools)
+            
             return service
             
         except Exception as e:
@@ -616,6 +639,7 @@ class ISAModelClient:
         service_type: str,
         model_hint: Optional[str] = None,
         provider_hint: Optional[str] = None,
+        tools: Optional[List[Any]] = None,
         **kwargs
     ) -> Dict[str, Any]:
         """Local invoke using AI Factory (original logic)"""
@@ -634,7 +658,8 @@ class ISAModelClient:
                 service_type=service_type,
                 model_name=selected_model["model_id"],
                 provider=selected_model["provider"],
-                task=task
+                task=task,
+                tools=tools
             )
             
             # Step 3: Execute task with unified interface
@@ -823,6 +848,7 @@ class ISAModelClient:
         service_type: str,
         model_hint: Optional[str] = None,
         provider_hint: Optional[str] = None,
+        tools: Optional[List[Any]] = None,
         **kwargs
     ):
         """Local streaming using AI Factory"""
@@ -840,7 +866,8 @@ class ISAModelClient:
             service_type=service_type,
             model_name=selected_model["model_id"],
             provider=selected_model["provider"],
-            task=task
+            task=task,
+            tools=tools
         )
         
         # Step 3: Yield tokens from the stream

{isa_model-0.3.7 → isa_model-0.3.8}/isa_model.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: isa_model
-Version: 0.3.7
+Version: 0.3.8
 Summary: Unified AI model serving framework
 Author: isA_Model Contributors
 Classifier: Development Status :: 3 - Alpha

{isa_model-0.3.7 → isa_model-0.3.8}/isa_model.egg-info/SOURCES.txt

@@ -59,16 +59,6 @@ isa_model/inference/__init__.py
 isa_model/inference/ai_factory.py
 isa_model/inference/base.py
 isa_model/inference/adapter/unified_api.py
-isa_model/inference/providers/__init__.py
-isa_model/inference/providers/base_provider.py
-isa_model/inference/providers/ml_provider.py
-isa_model/inference/providers/modal_provider.py
-isa_model/inference/providers/model_cache_manager.py
-isa_model/inference/providers/ollama_provider.py
-isa_model/inference/providers/openai_provider.py
-isa_model/inference/providers/replicate_provider.py
-isa_model/inference/providers/triton_provider.py
-isa_model/inference/providers/yyds_provider.py
 isa_model/inference/services/__init__.py
 isa_model/inference/services/base_service.py
 isa_model/inference/services/audio/base_stt_service.py

{isa_model-0.3.7 → isa_model-0.3.8}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "isa_model"
-version = "0.3.7"
+version = "0.3.8"
 description = "Unified AI model serving framework"
 authors = [{name = "isA_Model Contributors"}]
 readme = "README.md"

isa_model-0.3.7/isa_model/inference/providers/__init__.py

@@ -1,19 +0,0 @@
-"""
-Providers - Components for integrating with different model providers
-
-File: isa_model/inference/providers/__init__.py
-This module contains provider implementations for different AI model backends.
-"""
-
-from .base_provider import BaseProvider
-
-__all__ = [
-    "BaseProvider",
-]
-
-# Provider implementations can be imported individually as needed
-# from .triton_provider import TritonProvider
-# from .ollama_provider import OllamaProvider
-# from .yyds_provider import YYDSProvider
-# from .openai_provider import OpenAIProvider
-# from .replicate_provider import ReplicateProvider 

isa_model-0.3.7/isa_model/inference/providers/base_provider.py

@@ -1,77 +0,0 @@
-from abc import ABC, abstractmethod
-from typing import Dict, List, Any, Optional
-import os
-import logging
-from pathlib import Path
-import dotenv
-
-from isa_model.inference.base import ModelType, Capability
-
-logger = logging.getLogger(__name__)
-
-class BaseProvider(ABC):
-    """Base class for all AI providers - handles API key management"""
-    
-    def __init__(self, config: Optional[Dict[str, Any]] = None):
-        self.config = config or {}
-        self._load_environment_config()
-        self._validate_config()
-    
-    def _load_environment_config(self):
-        """Load configuration from environment variables"""
-        # Load .env file if it exists
-        project_root = Path(__file__).parent.parent.parent.parent
-        env_path = project_root / ".env"
-        
-        if env_path.exists():
-            dotenv.load_dotenv(env_path)
-        
-        # Subclasses should override this to load provider-specific env vars
-        self._load_provider_env_vars()
-    
-    @abstractmethod
-    def _load_provider_env_vars(self):
-        """Load provider-specific environment variables"""
-        pass
-    
-    def _validate_config(self):
-        """Validate that required configuration is present"""
-        # Subclasses should override this to validate provider-specific config
-        pass
-    
-    def get_api_key(self) -> Optional[str]:
-        """Get the API key for this provider"""
-        return self.config.get("api_key")
-    
-    def has_valid_credentials(self) -> bool:
-        """Check if provider has valid credentials"""
-        return bool(self.get_api_key())
-    
-    @abstractmethod
-    def get_capabilities(self) -> Dict[ModelType, List[Capability]]:
-        """Get provider capabilities by model type"""
-        pass
-    
-    @abstractmethod
-    def get_models(self, model_type: ModelType) -> List[str]:
-        """Get available models for given type"""
-        pass
-    
-    def get_config(self) -> Dict[str, Any]:
-        """Get provider configuration (without sensitive data)"""
-        # Return a copy without sensitive information
-        config_copy = self.config.copy()
-        if "api_key" in config_copy:
-            config_copy["api_key"] = "***" if config_copy["api_key"] else ""
-        if "api_token" in config_copy:
-            config_copy["api_token"] = "***" if config_copy["api_token"] else ""
-        return config_copy
-    
-    def get_full_config(self) -> Dict[str, Any]:
-        """Get full provider configuration (including sensitive data) - for internal use only"""
-        return self.config.copy()
-    
-    @abstractmethod
-    def is_reasoning_model(self, model_name: str) -> bool:
-        """Check if the model is optimized for reasoning tasks"""
-        pass

isa_model-0.3.7/isa_model/inference/providers/ml_provider.py

@@ -1,50 +0,0 @@
-from isa_model.inference.providers.base_provider import BaseProvider
-from isa_model.inference.base import ModelType, Capability
-from typing import Dict, List, Any
-import logging
-
-logger = logging.getLogger(__name__)
-
-class MLProvider(BaseProvider):
-    """Provider for traditional ML models"""
-    
-    def __init__(self, config=None):
-        default_config = {
-            "model_directory": "./models/ml",
-            "cache_models": True,
-            "max_cache_size": 5
-        }
-        
-        merged_config = {**default_config, **(config or {})}
-        super().__init__(config=merged_config)
-        self.name = "ml"
-        
-        logger.info(f"Initialized MLProvider with model directory: {self.config['model_directory']}")
-    
-    def get_capabilities(self) -> Dict[ModelType, List[Capability]]:
-        """Get provider capabilities"""
-        return {
-            ModelType.LLM: [],  # ML models are not LLMs
-            ModelType.EMBEDDING: [],
-            ModelType.VISION: [],
-            "ML": [  # Custom model type for traditional ML
-                "CLASSIFICATION",
-                "REGRESSION", 
-                "CLUSTERING",
-                "FEATURE_EXTRACTION"
-            ]
-        }
-    
-    def get_models(self, model_type: str = "ML") -> List[str]:
-        """Get available ML models"""
-        # In practice, this would scan the model directory
-        return [
-            "fraud_detection_rf",
-            "customer_churn_xgb", 
-            "price_prediction_lr",
-            "recommendation_kmeans"
-        ]
-    
-    def get_config(self) -> Dict[str, Any]:
-        """Get provider configuration"""
-        return self.config

isa_model-0.3.7/isa_model/inference/providers/modal_provider.py

@@ -1,109 +0,0 @@
-"""
-Modal Provider
-
-Provider for ISA self-hosted Modal services
-No API keys needed since we deploy our own services
-"""
-
-import os
-import logging
-from typing import Dict, Any, Optional, List
-from .base_provider import BaseProvider
-from isa_model.inference.base import ModelType, Capability
-
-logger = logging.getLogger(__name__)
-
-class ModalProvider(BaseProvider):
-    """Provider for ISA Modal services"""
-    
-    def __init__(self, config: Optional[Dict[str, Any]] = None):
-        super().__init__(config)
-        self.name = "modal"
-        self.base_url = "https://modal.com"  # Not used directly
-    
-    def _load_provider_env_vars(self):
-        """Load Modal-specific environment variables"""
-        # Modal doesn't need API keys for deployed services
-        # But we can load Modal token if available
-        modal_token = os.getenv("MODAL_TOKEN_ID") or os.getenv("MODAL_TOKEN_SECRET")
-        if modal_token:
-            self.config["modal_token"] = modal_token
-        
-        # Set default config
-        if "timeout" not in self.config:
-            self.config["timeout"] = 300
-        if "deployment_region" not in self.config:
-            self.config["deployment_region"] = "us-east-1"
-        if "gpu_type" not in self.config:
-            self.config["gpu_type"] = "T4"
-        
-    def get_api_key(self) -> str:
-        """Modal services don't need API keys for deployed apps"""
-        return "modal-deployed-service"  # Placeholder
-    
-    def get_base_url(self) -> str:
-        """Get base URL for Modal services"""
-        return self.base_url
-    
-    def validate_credentials(self) -> bool:
-        """
-        Validate Modal credentials
-        For deployed services, we assume they're accessible
-        """
-        try:
-            # Check if Modal is available
-            import modal
-            return True
-        except ImportError:
-            logger.warning("Modal package not available")
-            return False
-    
-    def get_capabilities(self) -> Dict[ModelType, List[Capability]]:
-        """Get Modal provider capabilities"""
-        return {
-            ModelType.VISION: [
-                Capability.OBJECT_DETECTION,
-                Capability.IMAGE_ANALYSIS,
-                Capability.UI_DETECTION,
-                Capability.OCR,
-                Capability.DOCUMENT_ANALYSIS
-            ]
-        }
-    
-    def get_models(self, model_type: ModelType) -> List[str]:
-        """Get available models for given type"""
-        if model_type == ModelType.VISION:
-            return [
-                "omniparser-v2.0",
-                "table-transformer-detection",
-                "table-transformer-structure-v1.1",
-                "paddleocr-3.0",
-                "yolov8"
-            ]
-        return []
-    
-    def is_reasoning_model(self, model_name: str) -> bool:
-        """Check if the model is optimized for reasoning tasks"""
-        # Vision models are not reasoning models
-        return False
-    
-    def get_default_config(self) -> Dict[str, Any]:
-        """Get default configuration for Modal services"""
-        return {
-            "timeout": 300,  # 5 minutes
-            "max_retries": 3,
-            "deployment_region": "us-east-1",
-            "gpu_type": "T4"
-        }
-    
-    def get_billing_info(self) -> Dict[str, Any]:
-        """Get billing information for Modal services"""
-        return {
-            "provider": "modal",
-            "billing_model": "compute_usage",
-            "cost_per_hour": {
-                "T4": 0.60,
-                "A100": 4.00
-            },
-            "note": "Costs depend on actual usage time, scales to zero when not in use"
-        }