import torchfrom transformers import AutoProcessor, AutoModelForMultimodalLM, AutoModelForCausalLM MODEL_ID_HUB = "tepirale/gemma-4-12B-merge-coder40-agentic40-it20" MA= "tepirale/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-assistant-safetensors-yuxinlu1" model = AutoModelForMultimodalLM.from_pretrained( MODEL_ID_HUB, dtype="auto", device_map="auto", # local_files_only=True) assistant_model = AutoModelForCausalLM.from_pretrained(MA, dtype=torch.bfloat16, device_map="auto" ) processor = AutoProcessor.from_pretrained(MODEL_ID_HUB) # Prompt - add image before textmessages = [ { "role": "user", "content": [ {"type": "image", "url": "https://raw.githubusercontent.com/google-gemma/cookbook/refs/heads/main/apps/sample-data/GoldenGate.png"}, {"type": "text", "text": "What is shown in this image?"} ] }] # Process inputinputs = processor.apply_chat_template( messages, tokenize=True, return_dict=True, return_tensors="pt", add_generation_prompt=True, enable_thinking=True).to(model.device)input_len = inputs["input_ids"].shape[-1] # Generate outputoutputs = model.generate(**inputs, max_new_tokens=3512, assistant_model=assistant_model)response = processor.decode(outputs[0][input_len:], skip_special_tokens=False) # Parse outputprocessor.parse_response(response)