srimanth-d
/

GOT_CPU

Image-Text-to-Text

feature-extraction

vision-language

Model card Files Files and versions

srimanth-d commited on Dec 12, 2024

Commit

fb190d7

·

verified ·

1 Parent(s): f40c6ec

Update modeling_got.py

Files changed (1) hide show

modeling_got.py +2 -2

modeling_got.py CHANGED Viewed

@@ -558,7 +558,7 @@ class GOTQwenForCausalLM(Qwen2ForCausalLM):
         inputs = tokenizer([prompt])
-        image_tensor_1 = image_processor_high(image)
         input_ids = torch.as_tensor(inputs.input_ids).to(device)
@@ -842,7 +842,7 @@ class GOTQwenForCausalLM(Qwen2ForCausalLM):
         inputs = tokenizer([prompt])
-        input_ids = torch.as_tensor(inputs.input_ids).cpu()
         stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
         keywords = [stop_str]

         inputs = tokenizer([prompt])
+        image_tensor_1 = image_processor_high(image).to(device)
         input_ids = torch.as_tensor(inputs.input_ids).to(device)
         inputs = tokenizer([prompt])
+        input_ids = torch.as_tensor(inputs.input_ids).to(device)
         stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
         keywords = [stop_str]