Upload serve_ministral.py with huggingface_hub

serve_ministral.py

@@ -15,6 +15,20 @@ import traceback
 
 app = FastAPI()
 
+def fix_bpe_tokens(text):
+    """Fix BPE tokenization artifacts"""
+    text = text.replace("Ġ", " ")
+    text = text.replace("Ċ", "\n")
+    text = text.replace("ĉ", "\t")
+    text = text.replace("âĢĻ", "'")
+    text = text.replace("âĢľ", '"')
+    text = text.replace("âĢĿ", '"')
+    text = text.replace("âĢĶ", "—")
+    text = text.replace("âĢĵ", "–")
+    text = text.replace("âĢ¦", "…")
+    text = text.replace("âĢĺ", "'")
+    return text
+
 # Global model and tokenizer
 model = None
 processor = None
@@ -105,6 +119,7 @@ async def chat_completions(request: ChatRequest):
         # Decode only the new tokens
         new_tokens = outputs[0][input_len:]
         response_text = processor.tokenizer.decode(new_tokens, skip_special_tokens=True)
+        response_text = fix_bpe_tokens(response_text)
         print(f"Generated {len(new_tokens)} tokens")
 
         return ChatResponse(