Salvando alterações

app.py

@@ -1,28 +1,31 @@
-from transformers import (
-    PaliGemmaProcessor,
-    PaliGemmaForConditionalGeneration,
-)
-from transformers.image_utils import load_image
+from transformers import pipeline
 import torch
 
-model_id = "google/paligemma2-3b-mix-224"
-
-url = "cnh-michele-digital.jpeg"
-image = load_image(url)
-
-model = PaliGemmaForConditionalGeneration.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto").eval()
-processor = PaliGemmaProcessor.from_pretrained(model_id)
-
-prompt = "Extraia o nome, cpf, data de nascimento e número de registro. Retorno apenas um JSON com esses dados."
-prompt = "describe en"
-model_inputs = processor(text=prompt, images=image, return_tensors="pt").to(torch.bfloat16).to(model.device)
-input_len = model_inputs["input_ids"].shape[-1]
+pipe = pipeline(
+    "image-text-to-text",
+    model="google/gemma-3-4b-it",
+    device="cuda",
+    torch_dtype=torch.bfloat16
+)
+messages = [
+    {
+        "role": "system",
+        "content": [{"type": "text", "text": "You are a helpful assistant."}]
+    },
+    {
+        "role": "user",
+        "content": [
+            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
+            {"type": "text", "text": "What animal is on the candy?"}
+        ]
+    }
+]
 
-with torch.inference_mode():
-    generation = model.generate(**model_inputs, max_new_tokens=100, do_sample=False)
-    generation = generation[0][input_len:]
-    decoded = processor.decode(generation, skip_special_tokens=True)
-    print("="*20)
-    print(decoded)
-    print("=" * 20)
-    print("FIM PROCESSO")
+output = pipe(text=messages, max_new_tokens=200)
+print("=" * 20)
+print(output[0]["generated_text"][-1]["content"])
+print("=" * 20)
+print("FIM PROCESSO")
+# Okay, let's take a look!
+# Based on the image, the animal on the candy is a **turtle**.
+# You can see the shell shape and the head and legs.