CoolFace
Modelpublic

Menlo/mini-Ichigo-llama3.2-3B-s-base

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
1likes33downloads
Model Card

Model Details

We have developed and released the family llama3s. This family is natively understanding audio and text input.

We continual pretrain on the expanded vocabulary homebrewltd/llama3.2-3B-s-whispervq-init with 900M tokens from 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

MMLU:

ModelMMLU Score
llama3.5-instruct-8b69.40
ichigo-llama3.1-s-v0.3: phase 363.79
ichigo-llama3.1-s-v0.3: phase 263.08
ichigo-llama3.1-s-base-v0.342.11
mini-ichigo-llama3.2-3B-s-instruct59.61
mini-ichigo-llama3.2-3B-s-base58.68
llama3.5-instruct-v0.250.27

Hardware

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

GPU Usage:

  • —Continual Training: 30 hours.

Training Arguments

We utilize torchtune library for the latest FSDP2 training code implementation.

ParameterContinual Training
Epoch1
Global batch size480
Learning Rate2e-4
Learning SchedulerLambdaLR with warmup
OptimizerAdamW fused
Warmup Steps80
Weight Decay0.01
Max Sequence Length512

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)