---
datasets:
- homebrewltd/instruction-speech-whispervq-v2
language:
- en
license: apache-2.0
tags:
- sound language model
---
## Model Details
We have developed and released the family [Ichigo-llama3s](https://huggingface.co/collections/homebrew-research/llama3-s-669df2139f0576abc6eb7405). This family is natively understanding audio and text input.
We expand the Semantic tokens experiment with WhisperVQ as a tokenizer for audio files from [homebrewltd/mini-Ichigo-llama3.2-3B-s-base](https://huggingface.co/homebrewltd/mini-Ichigo-llama3.2-3B-s-base) with nearly 1B tokens from [Instruction Speech WhisperVQ v3](homebrewltd/mixed-instruction-speech-whispervq-v3-full) dataset.
**Model developers** Homebrew Research.
**Input** Text and sound.
**Output** Text.
**Model Architecture** Llama-3.
**Language(s):** English.
## Intended Use
**Intended Use Cases** This family is primarily intended for research applications. This version aims to further improve the LLM on sound understanding capabilities.
**Out-of-scope** The use of llama3-s in any manner that violates applicable laws or regulations is strictly prohibited.
## How to Get Started with the Model
Try this model using [Google Colab Notebook](https://colab.research.google.com/drive/18IiwN0AzBZaox5o0iidXqWD1xKq11XbZ?usp=sharing).
First, we need to convert the audio file to sound tokens
```python
device = "cuda" if torch.cuda.is_available() else "cpu"
if not os.path.exists("whisper-vq-stoks-medium-en+pl-fixed.model"):
hf_hub_download(
repo_id="jan-hq/WhisperVQ",
filename="whisper-vq-stoks-medium-en+pl-fixed.model",
local_dir=".",
)
vq_model = RQBottleneckTransformer.load_model(
"whisper-vq-stoks-medium-en+pl-fixed.model"
).to(device)
vq_model.ensure_whisper(device)
def audio_to_sound_tokens(audio_path, target_bandwidth=1.5, device=device):
wav, sr = torchaudio.load(audio_path)
if sr != 16000:
wav = torchaudio.functional.resample(wav, sr, 16000)
with torch.no_grad():
codes = vq_model.encode_audio(wav.to(device))
codes = codes[0].cpu().tolist()
result = ''.join(f'<|sound_{num:04d}|>' for num in codes)
return f'<|sound_start|>{result}<|sound_end|>'
```
Then, we can inference the model the same as any other LLM.
```python
def setup_pipeline(model_path, use_4bit=False, use_8bit=False):
tokenizer = AutoTokenizer.from_pretrained(model_path)
model_kwargs = {"device_map": "auto"}
if use_4bit:
model_kwargs["quantization_config"] = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_compute_dtype=torch.bfloat16,
bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type="nf4",
)
elif use_8bit:
model_kwargs["quantization_config"] = BitsAndBytesConfig(
load_in_8bit=True,
bnb_8bit_compute_dtype=torch.bfloat16,
bnb_8bit_use_double_quant=True,
)
else:
model_kwargs["torch_dtype"] = torch.bfloat16
model = AutoModelForCausalLM.from_pretrained(model_path, **model_kwargs)
return pipeline("text-generation", model=model, tokenizer=tokenizer)
def generate_text(pipe, messages, max_new_tokens=64, temperature=0.0, do_sample=False):
generation_args = {
"max_new_tokens": max_new_tokens,
"return_full_text": False,
"temperature": temperature,
"do_sample": do_sample,
}
output = pipe(messages, **generation_args)
return output[0]['generated_text']
# Usage
llm_path = "homebrewltd/llama3.1-s-instruct-v0.2"
pipe = setup_pipeline(llm_path, use_8bit=True)
```
## Training process
**Training Metrics Image**: Below is a snapshot of the training loss curve visualized.
**[MMLU](https://huggingface.co/datasets/cais/mmlu)**:
| Model | MMLU Score |
| --- | --- |
| llama3.5-instruct-8b | 69.40 |
| ichigo-llama3.1-s-v0.3: phase 3 | 63.79 |
| ichigo-llama3.1-s-v0.3: phase 2 | 63.08 |
| ichigo-llama3.1-s-base-v0.3 | 42.11 |
| mini-ichigo-llama3.2-3B-s-instruct | **59.61** |
| mini-ichigo-llama3.2-3B-s-base | 58.68 |
| llama3.5-instruct-v0.2 | 50.27 |
**[AudioBench](https://arxiv.org/abs/2406.16020) Eval**:
| Model Bench | [Open-hermes Instruction Audio](https://huggingface.co/datasets/AudioLLMs/openhermes_instruction_test) (GPT-4-O judge 0:5) | [Alpaca Instruction Audio](https://huggingface.co/datasets/AudioLLMs/alpaca_audio_test) (GPT-4-O judge 0:5) |
| --- | --- | --- |
| [Llama3.1-s-v2](https://huggingface.co/homebrewltd/llama3-s-instruct-v0.2) | 3.45 | 3.53 |
| [Ichigo-llama3.1-s v0.3-phase2 -cp7000](https://huggingface.co/homebrewltd/Ichigo-llama3.1-s-instruct-v0.3-phase-2) | 3.42 | 3.62 |
| [Ichigo-llama3.1-s v0.3-phase2-cplast](https://huggingface.co/jan-hq/llama3-s-instruct-v0.3-checkpoint-last) | 3.31 | 3.6 |
| [Ichigo-llama3.1-s v0.3-phase3](https://huggingface.co/homebrewltd/Ichigo-llama3.1-s-instruct-v0.3-phase-3) | 3.64 | 3.68 |
| [mini-Ichigo-llama3.2-3B-s-instruct](https://huggingface.co/homebrewltd/mini-Ichigo-llama3.2-3B-s-instruct) | 2.58 | 2.07 |
| [Qwen2-audio-7B](https://huggingface.co/Qwen/Qwen2-Audio-7B) | 2.63 | 2.24 |
### Hardware
**GPU Configuration**: Cluster of 10x NVIDIA A6000-48GB.
**GPU Usage**:
- **Fine-tuning**: 12 hours.
### Training Arguments
We utilize [torchtune](https://github.com/pytorch/torchtune) library for the latest FSDP2 training code implementation.
| Parameter | Instruction Fine-tuning |
|----------------------------|-------------------------|
| **Epoch** | 1 |
| **Global batch size** | 360 |
| **Learning Rate** | 7e-5 |
| **Learning Scheduler** | LambdaLR with warmup |
| **Optimizer** | Adam torch fused |
| **Warmup Ratio** | 0.01 |
| **Weight Decay** | 0.005 |
| **Max Sequence Length** | 4096 |
## Examples
1. Good example:
Click to toggle Example 1
```
```
Click to toggle Example 2
```
```
2. Misunderstanding example:
Click to toggle Example 3
```
```
3. Off-tracked example:
Click to toggle Example 4
```
```
## Citation Information
**BibTeX:**
```
@article{Llama3-S: Sound Instruction Language Model 2024,
title={Llama3-S},
author={Homebrew Research},
year=2024,
month=August},
url={https://huggingface.co/homebrewltd/llama3.1-s-2024-08-20}
```
## Acknowledgement
- **[WhisperSpeech](https://github.com/collabora/WhisperSpeech)**
- **[Meta-Llama-3.1-8B-Instruct ](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct)**