Upload modeling_qwen2_5_vl.py

modeling_qwen2_5_vl.py

@@ -0,0 +1,9 @@
+from transformers import AutoModelForCausalLM
+import torch
+
+class QWenVLChatModel(AutoModelForCausalLM):
+    def chat(self, tokenizer, query: str, image=None, history=None, **kwargs):
+        inputs = tokenizer(query, return_tensors="pt").to(self.device)
+        with torch.no_grad():
+            outputs = self.generate(**inputs, max_new_tokens=512)
+        return tokenizer.decode(outputs[0], skip_special_tokens=True), history