|
|
--- |
|
|
dataset_info: |
|
|
features: |
|
|
- name: questions |
|
|
dtype: string |
|
|
- name: reasoning |
|
|
dtype: string |
|
|
- name: answers |
|
|
dtype: string |
|
|
splits: |
|
|
- name: test |
|
|
num_bytes: 79373605 |
|
|
num_examples: 388856 |
|
|
download_size: 30411606 |
|
|
dataset_size: 79373605 |
|
|
configs: |
|
|
- config_name: default |
|
|
data_files: |
|
|
- split: test |
|
|
path: data/test-* |
|
|
--- |
|
|
|
|
|
# MNLP Quantization Dataset |
|
|
|
|
|
This dataset is a copy of the MCQA dataset used during our project and reused here to calibrate our quantized model using post-training quantization (PTQ). |
|
|
|
|
|
## Origin |
|
|
Originally hosted at: [`hssawhney/MNLP-Dataset`](https://huggingface.co/datasets/hssawhney/MNLP-Dataset) |
|
|
|
|
|
## Use |
|
|
Used to calibrate our quantized W8A8 MCQA model. The quantization setup used 512 examples from the test set. |
|
|
|
