Update README.md

README.md

@@ -110,18 +110,48 @@ print(processor.decode(out[0].split("<answer>")[1].split("</answer>")[0], skip_s
 ## VLLM:
 ```python
 from PIL import Image
-from vllm import LLM, SamplingParameters
+from vllm import LLM, SamplingParams
 from transformers import AutoProcessor
 
-model_id = "NM-dev/NuMarkdown-Qwen2.5-VL"
-llm  = LLM(model=model_id, trust_remote_code=True, dtype="bfloat16")
-proc = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)
+model_id = "NM-dev/Qwen7B-m-5"
+
+llm = LLM(
+        model=model_id,
+        tokenizer=model_id,
+        dtype="bfloat16",
+        gpu_memory_utilization=0.85,
+        max_num_seqs=256,
+        enforce_eager=True,
+        trust_remote_code=True
+    )
+
+sampling_params = SamplingParams(
+                temperature=0.8,
+                max_tokens=5000,
+)
+processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)
+
+inputs = []
+messages = [{
+        "role": "user",
+        "content": [
+            {"type": "image"},
+            # {"type": "text", "text": guideline},
+        ]
+}]
+
+prompt = proc.apply_chat_template(
+        messages,
+        tokenize=False,
+        add_generation_prompt=True,
+    )
+image = Image.open("invoice.png").convert("RGB")
 
-img = Image.open("invoice_scan.png")
-prompt = proc(text="Convert this to Markdown with reasoning.", image=img,
-              return_tensors="np")  # numpy arrays for vLLM
+inputs.append({
+        "prompt": prompt,
+        "multi_modal_data": {"image": image}
+})
 
-params = SamplingParameters(max_tokens=1024, temperature=0.8, top_p=0.95)
-result = llm.generate([{"prompt": prompt}], params)[0].outputs[0].text.split("<answer>")[1].split("</answer>")[0]
-print(result)
+outs = llm.generate(inputs, sampling_params)
+preds = [o.outputs[0].text.strip().split("<answer>")[1].split("</answer>")[0] for o in outs]
 ```