taipei-1-mllama-project-2024/multi-modal-llama-tp1

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->

Model Details

Model Description

<!-- Provide a longer summary of what this model is. -->

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

• —Developed by: ASUS, NTHU, NTU
• —Model type: Based on Llama-3.2-11B-Vision-Instruct, with added support for voice input.
• —Language(s) (NLP): Supports multiple languages, but optimized for Traditional Chinese.
• —License: MIT
• —Finetuned from model [optional]: meta-llama/Llama-3.2-11B-Vision-Instruct

Uses

The purpose of this multimodal model is to enrich knowledge about tourist attractions in Taiwan and engage travelers through interactive voice responses. You can provide a picture of a Taiwan's landscape to initiate a conversation.

How to Get Started with the Model

Use the code below to get started with the model.

import torch
from transformers import pipeline
import librosa
from PIL import Image

model_path = "taipei-1-mllama-project-2024/multi-modal-llama-tp1"
pipe = pipeline(model=model_path, trust_remote_code=True, device_map='auto')
audio, sr = librosa.load("/path/to/請問圖片中的景點是哪裡.wav", sr=16000)
image = Image.open("/path/to/台南孔廟.jpg")
turns = [
  dict(
    role='system',
    content = "You are a travel expert who can accurately analyze the attractions in the pictures. All conversations should be conducted in Traditional Chinese.",
  ),
  dict(
    role='user',
    content='<|image|><|begin_of_audio|><|audio|><|end_of_audio|>'
  )
]
y_pred = pipe({'audio': [audio], 'images': [image], 'turns': turns, 'sampling_rate': sr}, max_new_tokens=300)
print(y_pred) # 這張照片中的景點是台灣的「台南孔廟」。...

Training Details

Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Training Hyperparameters

• —Training regime: [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->

Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

Testing Data, Factors & Metrics

Testing Data

<!-- This should link to a Dataset Card if possible. -->

[More Information Needed]

Factors

<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->

[More Information Needed]

Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

[More Information Needed]

Results

[More Information Needed]

Summary

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

Taiwan-1 Super Computer)

Hardware

H100 x 8 GPUs per node x 16 nodes