default to qwen3 0.6b

summarize_transcript.py

@@ -56,10 +56,11 @@ def stream_summarize_transcript(llm, transcript):
 
     stream = llm.create_chat_completion(
         messages=messages,
-        max_tokens=512,
-        temperature=0.3,
-        top_p=0.9,
-        repeat_penalty=1.5,
+        max_tokens=1024,
+        temperature=0.6,
+        min_p=0.0,
+        top_p=0.95,
+        top_k=20,
         stop=["<|end_of_text|>", "<|eot_id|>", "<|eom_id|>"],
         stream=True
     )
@@ -86,8 +87,8 @@ def main():
                         default="./transcripts/short.txt", 
                         help="Path to the input transcript file (default: ./transcripts/short.txt)")
     parser.add_argument("-m", "--model", type=str, 
-                        default="bartowski/baidu_ERNIE-4.5-0.3B-PT-GGUF:Q6_K", 
-                        help="HuggingFace model in format repo_id:quant (e.g., Luigi/Falcon-H1-Tiny-Multilingual-100M-Instruct-GGUF:IQ4_NL)")
+                        default="unsloth/Qwen3-0.6B-GGUF:Q4_0", 
+                        help="HuggingFace model in format repo_id:quant (e.g., bartowski/baidu_ERNIE-4.5-0.3B-PT-GGUF:Q4_0)")
     parser.add_argument("-c", "--cpu", action="store_true", help="Force CPU only inference")
     args = parser.parse_args()