Datasets:

zay25
/

MNLP_M2_quantized_dataset

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

Use

Used to calibrate our quantized W8A8 MCQA model. The quantization setup used 512 examples from the test set.