refactor: 更新inference 部分的实现

scripts/demo/generate_ar.py

@@ -3,7 +3,7 @@ from pathlib import Path
 import torch
 
 from astrai.config.param_config import ModelParameter
-from astrai.inference.generator import GenerationRequest, GeneratorFactory
+from astrai.inference import InferenceEngine
 
 PROJECT_ROOT = Path(__file__).resolve().parents[2]
 PARAMETER_ROOT = Path(PROJECT_ROOT, "params")
@@ -15,17 +15,15 @@ def generate_text():
 
     query = input(">> ")
 
-    request = GenerationRequest(
-        query=query,
+    engine = InferenceEngine(param)
+    response = engine.generate(
+        prompt=query,
+        stream=False,
+        max_tokens=param.config.max_len,
         temperature=0.8,
         top_p=0.95,
         top_k=50,
-        max_len=param.config.max_len,
-        history=None,
-        system_prompt=None,
     )
-    generator = GeneratorFactory.create(param, request)
-    response = generator.generate(request)
 
     print(response)