minor

app.py

@@ -11,7 +11,7 @@ import uuid
 import json
 
 # Constants
-SYSTEM_PROMPT = """You are a helpful assistant."""
+SYSTEM_PROMPT = """You are SmallThinker-3B, a helpful AI assistant. You try to follow instructions as much as possible while being accurate and brief."""
 device = "cuda" if torch.cuda.is_available() else "cpu"
 TITLE = "<h1><center>SmallThinker-3B Chat</center></h1>"
 MODEL_PATH = "PowerInfer/SmallThinker-3B-Preview"
@@ -64,8 +64,8 @@ logs_file = Path("logs/") / f"data_{uuid.uuid4()}.json"
 logs_folder = logs_file.parent
 
 # scheduler = CommitScheduler(
-#     repo_id="PowerInfer/SmallThinker-3B-Preview",
-#     repo_type="model",
+#     repo_id=logs_id,
+#     repo_type="dataset",
 #     folder_path=logs_folder,
 #     path_in_repo="data",
 #     every=5,
@@ -81,7 +81,7 @@ def stream_chat(
     max_new_tokens: int = 1024,
     top_p: float = 1.0,
     top_k: int = 20,
-    repetition_penalty: float = 1.2,
+    repetition_penalty: float = 1.1,
 ):
     # Create new history list with current message
     new_history = history + [[message, ""]]
@@ -156,8 +156,8 @@ def main():
                 )
                 
                 with gr.Accordion(label="⚙️ Parameters", open=False):
-                    temperature = gr.Slider(minimum=0, maximum=1, step=0.1, value=0.7, label="Temperature")
-                    max_new_tokens = gr.Slider(minimum=128, maximum=32768, step=128, value=16384, label="Max new tokens")
+                    temperature = gr.Slider(minimum=0, maximum=1, step=0.1, value=0.3, label="Temperature")
+                    max_new_tokens = gr.Slider(minimum=128, maximum=32768, step=128, value=1024, label="Max new tokens")
                     top_p = gr.Slider(minimum=0.0, maximum=1.0, step=0.1, value=0.8, label="Top-p")
                     top_k = gr.Slider(minimum=1, maximum=100, step=1, value=20, label="Top-k")
                     repetition_penalty = gr.Slider(minimum=1.0, maximum=2.0, step=0.1, value=1.1, label="Repetition penalty")