Update app.py

app.py

@@ -4,7 +4,6 @@ from transformers import (
     pipeline,
     AutoTokenizer,
     AutoModelForCausalLM,
-    GenerationConfig,
 )
 from langchain_huggingface import HuggingFacePipeline
 from langchain_core.prompts import PromptTemplate
@@ -15,19 +14,16 @@ MODEL_ID = "bigcode/starcoder2-3b"
 tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
 model = AutoModelForCausalLM.from_pretrained(MODEL_ID, trust_remote_code=True)
 
-# — Generation config & pipeline (new API) —
-gen_config = GenerationConfig(
-    max_new_tokens=64,
-    temperature=0.2,
-    top_p=0.95,
-    do_sample=False,
-)
+# — Pipeline setup (pass generation parameters directly) —
 pipe = pipeline(
     "text-generation",
     model=model,
     tokenizer=tokenizer,
     device_map="auto",
-    generation_config=gen_config,
+    max_new_tokens=64,
+    temperature=0.2,
+    top_p=0.95,
+    do_sample=False,
 )
 llm = HuggingFacePipeline(pipeline=pipe)