chore(mlx-lm): add adapter support in generate.py (#494)

* chore(mlx-lm): add adapter support in generate.py

* chore: remove generate from lora.py and raise error to let user use mlx_lm.generate instead

llms/mlx_lm/LORA.md

@@ -72,6 +72,17 @@ python -m mlx_lm.lora \
     --test
 ```
 
+### Generate
+
+For generation use mlx_lm.generate:
+
+```shell
+python -m mlx_lm.generate \
+    --model <path_to_model> \
+    --adapter-file <path_to_adapters.npz> \
+    --prompt "<your_model_prompt>"
+```
+
 ## Fuse and Upload
 
 You can generate a model fused with the low-rank adapters using the

llms/mlx_lm/generate.py

@@ -23,6 +23,11 @@ def setup_arg_parser():
         default="mlx_model",
         help="The path to the local model directory or Hugging Face repo.",
     )
+    parser.add_argument(
+        "--adapter-file",
+        type=str,
+        help="Optional path for the trained adapter weights.",
+    )
     parser.add_argument(
         "--trust-remote-code",
         action="store_true",
@@ -99,7 +104,9 @@ def main(args):
     if args.eos_token is not None:
         tokenizer_config["eos_token"] = args.eos_token
 
-    model, tokenizer = load(args.model, tokenizer_config=tokenizer_config)
+    model, tokenizer = load(
+        args.model, adapter_file=args.adapter_file, tokenizer_config=tokenizer_config
+    )
 
     if not args.ignore_chat_template and (
         hasattr(tokenizer, "apply_chat_template")

llms/mlx_lm/lora.py

@@ -9,7 +9,7 @@ from mlx.utils import tree_flatten
 
 from .tuner.trainer import TrainingArgs, TrainingCallback, evaluate, train
 from .tuner.utils import linear_to_lora_layers
-from .utils import generate, load
+from .utils import load
 
 
 def build_parser():
@@ -234,15 +234,8 @@ def run(args, training_callback: TrainingCallback = None):
         print(f"Test loss {test_loss:.3f}, Test ppl {test_ppl:.3f}.")
 
     if args.prompt is not None:
-        print("Generating")
+        raise NotImplementedError(
-        model.eval()
+            "Please use mlx_lm.generate with trained adapter for generation."
-        generate(
-            model=model,
-            tokenizer=tokenizer,
-            temp=args.temp,
-            max_tokens=args.max_tokens,
-            prompt=args.prompt,
-            verbose=True,
         )