Fix lm-eval neuralspeed loading model (#1359)

intel · Mar 11, 2024 · cd6e488 · cd6e488
1 parent c65cec7
commit cd6e488
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/intel_extension_for_transformers/llm/evaluation/lm_eval/models/huggingface.py b/intel_extension_for_transformers/llm/evaluation/lm_eval/models/huggingface.py
@@ -623,7 +623,8 @@ def __init__(self, *args, pretrained, model_format, **kwargs):
         if self.model_format == "runtime":
             from transformers import AutoTokenizer, TextStreamer
             from intel_extension_for_transformers.transformers import AutoModelForCausalLM
-            self.runtime_model = AutoModelForCausalLM.from_pretrained(pretrained, quantization_config=self.woq_config)
+            self.runtime_model = AutoModelForCausalLM.from_pretrained(pretrained, quantization_config=self.woq_config,
+                                                        trust_remote_code=kwargs.get("trust_remote_code", False))
 
         if self.model_format == "onnx":
             if not os.path.exists(os.path.join(pretrained, "decoder_model.onnx")) and \