[NeuralChat] Enable finetune for Qwen-7b-chat on CPU (WIP) (#462)

intel · Nov 30, 2023 · 6bc9382 · 6bc9382
1 parent 0487e00
commit 6bc9382
Showing 1 changed file with 3 additions and 1 deletion.
diff --git a/intel_extension_for_transformers/llm/finetuning/finetuning.py b/intel_extension_for_transformers/llm/finetuning/finetuning.py
@@ -231,6 +231,7 @@ def load_tokenizer(self, model_args):
             "use_fast": model_args.use_fast_tokenizer,
             "revision": model_args.model_revision,
             "use_auth_token": True if model_args.use_auth_token else None,
+            "trust_remote_code": model_args.trust_remote_code,
         }
         if model_args.tokenizer_name:
             tokenizer = AutoTokenizer.from_pretrained(
@@ -285,7 +286,8 @@ def finetune(self):
                 )
 
         config = self.load_model_config(self.model_args)
-        if config.architectures[0].endswith("ForCausalLM"):
+        if config.architectures[0].endswith("ForCausalLM") \
+            or config.architectures[0].endswith("QWenLMHeadModel"):
             self.finetune_clm(model_args, data_args, training_args, finetune_args, config)
         elif config.architectures[0].endswith("ForConditionalGeneration"):
             self.finetune_seq2seq(model_args, data_args, training_args, finetune_args, config)