refactor : merge max_prompt_len into max_seq_len, replace assert with raise

- Engine/Scheduler/TaskManager: merge max_prompt_len into max_seq_len
- train.py: replace bare assert with ValueError/FileNotFoundError
- server.py: add --max_seq_len CLI option
- engine.py: remove dead page_size param

astrai/inference/api/server.py

@@ -110,6 +110,7 @@ def _create_engine(
     device: str = "cuda",
     dtype: torch.dtype = torch.bfloat16,
     max_batch_size: int = 16,
+    max_seq_len: Optional[int] = None,
 ) -> InferenceEngine:
     if not param_path.exists():
         raise FileNotFoundError(f"Parameter directory not found: {param_path}")
@@ -123,6 +124,7 @@ def _create_engine(
         model=model,
         tokenizer=tokenizer,
         max_batch_size=max_batch_size,
+        max_seq_len=max_seq_len,
     )
     logger.info(f"Inference engine initialized with max_batch_size={max_batch_size}")
     return engine
@@ -186,6 +188,7 @@ def run_server(
     device: str = "cuda",
     dtype: torch.dtype = torch.bfloat16,
     max_batch_size: int = 16,
+    max_seq_len: Optional[int] = None,
 ):
     app = get_app()
     app.state.server_config = {
@@ -193,6 +196,7 @@ def run_server(
         "dtype": dtype,
         "param_path": param_path,
         "max_batch_size": max_batch_size,
+        "max_seq_len": max_seq_len,
     }
     uvicorn.run(
         app,

astrai/inference/core/scheduler.py

@@ -23,7 +23,6 @@ class InferenceScheduler:
         tokenizer: AutoTokenizer,
         max_batch_size: int = 16,
         max_seq_len: Optional[int] = None,
-        max_prompt_len: int = 2048,
         device: Optional[str] = None,
         dtype: Optional[torch.dtype] = None,
         cache: Optional[KVCache] = None,
@@ -61,7 +60,6 @@ class InferenceScheduler:
             tokenizer=tokenizer,
             max_batch_size=max_batch_size,
             max_seq_len=self.max_seq_len,
-            max_prompt_len=max_prompt_len,
         )
 
         self._executor = Executor(

astrai/inference/core/task.py

@@ -135,12 +135,10 @@ class TaskManager:
         tokenizer: AutoTokenizer,
         max_batch_size: int = 16,
         max_seq_len: int = 8192,
-        max_prompt_len: int = 512,
     ):
         self.tokenizer = tokenizer
         self.max_batch_size = max_batch_size
         self.max_seq_len = max_seq_len
-        self.max_prompt_len = max_prompt_len
 
         self.waiting_queue: Deque[Task] = deque()
         self.active_tasks: List[Task] = []
@@ -165,10 +163,10 @@ class TaskManager:
     ) -> str:
         task_id = f"task_{int(time.time())}_{uuid.uuid4().hex[:8]}"
         prompt_ids = self.tokenizer.encode(prompt)
-        if len(prompt_ids) > self.max_prompt_len:
-            prompt_ids = prompt_ids[-self.max_prompt_len :]
+        if len(prompt_ids) > self.max_seq_len:
+            prompt_ids = prompt_ids[-self.max_seq_len :]
 
-        if len(prompt_ids) >= self.max_seq_len:
+        if len(prompt_ids) > self.max_seq_len:
             if stream_callback:
                 stream_callback(STOP)
             return task_id

astrai/inference/engine.py

@@ -111,8 +111,6 @@ class InferenceEngine:
         tokenizer: AutoTokenizer,
         max_batch_size: int = 1,
         max_seq_len: Optional[int] = None,
-        max_prompt_len: int = 2048,
-        page_size: int = 128,
         cache: Optional[KVCache] = None,
     ):
         self.model = model
@@ -122,7 +120,6 @@ class InferenceEngine:
             tokenizer=self.tokenizer,
             max_batch_size=max_batch_size,
             max_seq_len=max_seq_len,
-            max_prompt_len=max_prompt_len,
             cache=cache,
         )