From a7da4ad7a34e487b6c2c73de7e195feffb8654d0 Mon Sep 17 00:00:00 2001 From: Anchen Date: Fri, 29 Dec 2023 23:01:20 +1100 Subject: [PATCH] chore: fix the load quantization model --- llms/deepseek-coder/deepseek_coder.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llms/deepseek-coder/deepseek_coder.py b/llms/deepseek-coder/deepseek_coder.py index 3f3928e7..5108c8fe 100644 --- a/llms/deepseek-coder/deepseek_coder.py +++ b/llms/deepseek-coder/deepseek_coder.py @@ -244,7 +244,7 @@ def load_model(model_path: str): model = DeepseekCoder(model_args) weights = mx.load(str(model_path / "weights.npz")) - if quantization := config.get("quantization", False): + if quantization: nn.QuantizedLinear.quantize_module(model, **quantization) model.update(tree_unflatten(list(weights.items())))