---
datasets:
- homebrewltd/instruction-speech-whispervq-v2
language:
- en
license: apache-2.0
tags:
- sound language model
pipeline_tag: audio-text-to-text
---

## Model Details

We have developed and released the family [llama3s](https://huggingface.co/collections/homebrew-research/llama3-s-669df2139f0576abc6eb7405). This family is natively understanding audio and text input.

We continual pretrain on the expanded vocabulary [homebrewltd/llama3.1-s-whispervq-init](https://huggingface.co/homebrewltd/llama3.1-s-whispervq-init) with 900M tokens from [homebrewltd/raw-speech-whispervq-v1](https://huggingface.co/datasets/homebrewltd/raw-speech-whispervq-v1) 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.

## Training process
**Training Metrics Image**: Below is a snapshot of the training loss curve visualized.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/65713d70f56f9538679e5a56/gtpDSs750SkMPJO0-UtFq.png)

**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** |
| llama3.5-instruct-v0.2 | 50.27 |

### Hardware

**GPU Configuration**: Cluster of 10x NVIDIA A6000-48GB.

**GPU Usage**:
  - **Continual Training**: 30 hours.

### Training Arguments

We utilize [torchtune](https://github.com/pytorch/torchtune) library for the latest FSDP2 training code implementation. 

| Parameter                  | Continual Training      | 
|----------------------------|-------------------------|
| **Epoch**                  | 1                       | 
| **Global batch size**      | 480                     | 
| **Learning Rate**          | 2e-4                    | 
| **Learning Scheduler**     | Cosine with warmup      | 
| **Optimizer**              | AdamW fused             | 
| **Warmup Steps**           | 50                      | 
| **Weight Decay**           | 0.01                    |
| **Max Sequence Length**    | 512                     |


## 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-2024-08-15}
```

## Acknowledgement

- **[WhisperSpeech](https://github.com/collabora/WhisperSpeech)**

- **[Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct)**