|
--- |
|
language: |
|
- en |
|
tags: |
|
- llama |
|
license: other |
|
metrics: |
|
- MMLU |
|
- ARC |
|
- HellaSwag |
|
- TruthfulQA |
|
--- |
|
|
|
# 🥳 Platypus-30B has arrived! |
|
|
|
Platypus-30B is an instruction fine-tuned model based on the LLaMA-30b transformer architecture. |
|
|
|
| Metric | Value | |
|
|-----------------------|-------| |
|
| MMLU (5-shot) | 65.4 | |
|
| ARC (25-shot) | 64.6 | |
|
| HellaSwag (10-shot) | 84.3 | |
|
| TruthfulQA (0-shot) | 45.8 | |
|
|-----------------------|-------| |
|
| Avg. | 65 | 💥 |
|
|
|
## Usage |
|
|
|
```sh |
|
ADD |
|
``` |
|
|
|
## Model Details |
|
|
|
* **Trained by**: [Ariel Lee & Cole Hunter, LINK TO WEBSITES] |
|
* **Model type:** **Platypus-30B** is an auto-regressive language model based on the LLaMA transformer architecture. |
|
* **Language(s)**: English |
|
* **License for base weights**: License for the base LLaMA model's weights is Meta's [non-commercial bespoke license](https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md). |
|
|
|
| Hyperparameter | Value | |
|
|---------------------------|-------| |
|
| \\(n_\text{parameters}\\) | 33B | |
|
| \\(d_\text{model}\\) | 6656 | |
|
| \\(n_\text{layers}\\) | 60 | |
|
| \\(n_\text{heads}\\) | 52 | |
|
|
|
## Training Dataset |
|
|
|
Dataset of highly filtered and curated question and answer pairs. Release TBD. |
|
|
|
## Training Procedure |
|
|
|
`lilloukas/Platypus-30b` was instruction fine-tuned using lora [CITE REPO] on 4 A100 80GB with the following configuration: |
|
|
|
| Hyperparameter | Value | |
|
|---------------------|-------| |
|
| learning_rate | --- | |
|
| batch_size | --- | |
|
| microbatch_size | --- | |
|
| warmup_steps | --- | |
|
| epochs | --- | |
|
| weight_decay | --- | |
|
| optimizer | --- | |
|
| weight_decay | --- | |
|
| cutoff_len | --- | |
|
| lora_target_modules | --- | |
|
|
|
|
|
## Limitations and bias |
|
|
|
The base LLaMA model is trained on various data, some of which may contain offensive, harmful, and biased content that can lead to toxic behavior. See Section 5.1 of the LLaMA [paper](https://arxiv.org/abs/2302.13971). We have not performed any studies to determine how fine-tuning on the aforementioned datasets affect the model's behavior and toxicity. Do not treat chat responses from this model as a substitute for human judgment or as a source of truth. Please use responsibly. |
|
|
|
## Citations |
|
|
|
```bibtex |
|
@article{touvron2023llama, |
|
title={LLaMA: Open and Efficient Foundation Language Models}, |
|
author={Touvron, Hugo and Lavril, Thibaut and Izacard, Gautier and Martinet, Xavier and Lachaux, Marie-Anne and Lacroix, Timoth{\'e}e and Rozi{\`e}re, Baptiste and Goyal, Naman and Hambro, Eric and Azhar, Faisal and Rodriguez, Aurelien and Joulin, Armand and Grave, Edouard and Lample, Guillaume}, |
|
journal={arXiv preprint arXiv:2302.13971}, |
|
year={2023} |
|
} |
|
``` |
