File size: 774 Bytes
59d2e5c 2c14e6e afcfe6e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
---
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.
|