fix response + bump version (#319)

mlx_lm/utils.py

@@ -133,9 +133,9 @@ def generate(
             print(s[skip:], end="", flush=True)
             skip = len(s)
 
-    tokens = tokenizer.decode(tokens)[skip:]
+    tokens = tokenizer.decode(tokens)
     if verbose:
-        print(tokens, flush=True)
+        print(tokens[skip:], flush=True)
     return tokens
 
 

setup.py

@@ -8,7 +8,7 @@ with open(Path(__file__).parent / "mlx_lm/requirements.txt") as fid:
     requirements = [str(r) for r in pkg_resources.parse_requirements(fid)]
 setup(
     name="mlx-lm",
-    version="0.0.2",
+    version="0.0.3",
     description="LLMs on Apple silicon with MLX and the Hugging Face Hub",
     long_description=open("README.md", encoding="utf-8").read(),
     long_description_content_type="text/markdown",