refactor: simplify training and inference interfaces

- avoid constructing model_fn more than once when reading config
- keep inference package exports focused on public entry points
- rename extra strategy arguments to strategy_kwargs

astrai/inference/__init__.py

@@ -12,45 +12,10 @@ Modules:
   - engine.py:     Facade (InferenceEngine)
 """
 
-from astrai.inference.cache import (
-    Allocator,
-    KVCache,
-    KVStorage,
-    PagePool,
-    RadixCache,
-    ReqToTokenPool,
-    TaskCacheManager,
-    page_hash,
-)
 from astrai.inference.engine import InferenceEngine
-from astrai.inference.network import (
-    AnthropicMessage,
-    BaseToolParser,
-    ChatCompletionRequest,
-    ChatMessage,
-    FunctionDef,
-    GenContext,
-    MessagesRequest,
-    ProtocolHandler,
-    SimpleJsonToolParser,
-    StopChecker,
-    ToolDef,
-    ToolParserFactory,
-    get_app,
-    run_server,
-)
-from astrai.inference.network.anthropic import AnthropicResponseBuilder
-from astrai.inference.network.openai import OpenAIResponseBuilder
+from astrai.inference.network import get_app, run_server
 from astrai.inference.runtime.executor import Executor
-from astrai.inference.runtime.sample import (
-    BaseSamplingStrategy,
-    FrequencyPenaltyStrategy,
-    SamplingPipeline,
-    TemperatureStrategy,
-    TopKStrategy,
-    TopPStrategy,
-    sample,
-)
+from astrai.inference.runtime.sample import sample
 from astrai.inference.scheduler import InferenceScheduler
 from astrai.inference.task import STOP, Task, TaskManager, TaskStatus
 
@@ -62,35 +27,7 @@ __all__ = [
     "Task",
     "TaskManager",
     "TaskStatus",
-    "Allocator",
-    "KVCache",
-    "KVStorage",
-    "PagePool",
-    "RadixCache",
-    "ReqToTokenPool",
-    "TaskCacheManager",
-    "page_hash",
     "sample",
-    "BaseSamplingStrategy",
-    "TemperatureStrategy",
-    "TopKStrategy",
-    "TopPStrategy",
-    "FrequencyPenaltyStrategy",
-    "SamplingPipeline",
-    "ProtocolHandler",
-    "StopChecker",
-    "GenContext",
-    "BaseToolParser",
-    "SimpleJsonToolParser",
-    "ToolParserFactory",
-    "OpenAIResponseBuilder",
-    "AnthropicResponseBuilder",
-    "ChatMessage",
-    "ChatCompletionRequest",
-    "FunctionDef",
-    "ToolDef",
-    "AnthropicMessage",
-    "MessagesRequest",
     "get_app",
     "run_server",
 ]