distilroberta-squad / README.md
Haritz Puerto
Update README.md
583aaae
|
raw
history blame
1.76 kB
---
language:
- en
tags:
- QA
- Question Answering
- SQuAD
license: "mit"
datasets:
- squad
metrics:
- squad
model-index:
- name: distilroberta-base
results:
- task:
type: question-answering # Required. Example: automatic-speech-recognition
name: Question Answering # Optional. Example: Speech Recognition
dataset:
type: squad # Required. Example: common_voice. Use dataset id from https://hf.co/datasets
name: SQuAD # Required. A pretty name for the dataset. Example: Common Voice (French)
split: validation # Optional. Example: test
metrics:
- type: squad # Required. Example: wer. Use metric id from https://hf.co/metrics
value: 76.37653736991486 # Required. Example: 20.90
name: SQuAD EM # Optional. Example: Test WER
config: exact_match # Optional. The name of the metric configuration used in `load_metric()`. Example: bleurt-large-512 in `load_metric("bleurt", "bleurt-large-512")`. See the `datasets` docs for more info: https://huggingface.co/docs/datasets/v2.1.0/en/loading#load-configurations
- type: squad # Required. Example: wer. Use metric id from https://hf.co/metrics
value: 84.5528918750732 # Required. Example: 20.90
name: SQuAD F1 # Optional. Example: Test WER
config: F1
---
distilroberta-base fined-tuned on SQuAD (https://huggingface.co/datasets/squad)
Hyperparameters:
- epochs: 1
- lr: 1e-5
- train batch sie: 16
- optimizer: adamW
- lr_scheduler: linear
- num warming steps: 0
- max_length: 512
Results on the dev set:
- 'exact_match': 76.37653736991486
- 'f1': 84.5528918750732
It took 1h 20 min to train on Colab.