Integrate Gradio client for LLM predictions and update requirements

app.py

@@ -15,8 +15,10 @@ from trellis.pipelines import TrellisImageTo3DPipeline
 from trellis.representations import Gaussian, MeshExtractResult
 from trellis.utils import render_utils, postprocessing_utils
 from huggingface_hub import InferenceClient
+from gradio_client import Client
 
 client = InferenceClient(api_key=os.environ["HF_API_KEY"])
+llm_client = Client("Qwen/Qwen2.5-72B-Instruct")
 
 def generate_t2i_prompt(item_name):
     llm_prompt_template = """You are tasked with creating a concise yet highly detailed description of an item to be used for generating an image in a game development pipeline. The image should show the **entire item** with no parts cropped or hidden. The background should always be plain and monocolor, with no focus on it.  
@@ -37,19 +39,12 @@ Now generate a concise description for the item: "{item_name}"
 Focus on the item itself, ensuring it is fully described, and specify a plain, white background and the output is no longer than 77 tokens.
 """
 
-    messages = [
-        {
-            "role": "user",
-            "content": llm_prompt_template.format(item_name=item_name)
-        }
-    ]
-
-    completion = client.chat.completions.create(
-        model="Qwen/Qwen2.5-72B-Instruct", 
-        messages=messages, 
-        max_tokens=500
-    )
-    object_t2i_prompt = completion.choices[0].message.content
+    object_t2i_prompt = llm_client.predict(
+        query=llm_prompt_template.format(item_name=item_name),
+        history=[],
+        system="You are Qwen, created by Alibaba Cloud. You are a helpful assistant.",
+        api_name="/model_chat",
+    )[1][0][-1]
     print(object_t2i_prompt)
 
     return object_t2i_prompt

requirements.txt

@@ -1,6 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/cu121
 --find-links https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.4.0_cu121.html
 
+gradio_client
 huggingface-hub==0.26.5
 torch==2.4.0
 torchvision==0.19.0