Unexpected Output While Generating a 3D Model Using Zhengyi/LLaMA-Mesh
#5
by
orha
- opened
Hi everyone,
I recently used the Zhengyi/LLaMA-Mesh model to generate a 3D model of a fire extinguisher, but the results were unexpected. Instead of a coherent mesh, I got an odd mesh like this:
Generated OBJ:
The format resembles a 3D mesh structure (vertices and faces), but rendering it produced no meaningful object. Adjusting the prompt or cleaning the data didn’t help much.
Code:
from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "Zhengyi/LLaMA-Mesh"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto")
prompt = "Generate a 3D model of a fire extinguisher."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_length=3000)
mesh_data = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(mesh_data)