jedidiahkkh/mnli_stats
收藏Hugging Face2024-05-20 更新2024-06-12 收录
下载链接：
https://hf-mirror.com/datasets/jedidiahkkh/mnli_stats
下载链接
链接失效反馈官方服务：
资源简介：
---
dataset_info:
- config_name: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525
features:
- name: epoch
dtype: int32
- name: uid
dtype: int64
- name: logits
sequence: float64
- name: loss
dtype: float64
- name: grad_1norm
dtype: float64
- name: grad_2norm
dtype: float64
- name: grad_infnorm
dtype: float64
- name: label
dtype: int32
splits:
- name: epoch0
num_bytes: 29712000
num_examples: 384000
- name: epoch3
num_bytes: 29712000
num_examples: 384000
- name: epoch6
num_bytes: 29712000
num_examples: 384000
- name: epoch5
num_bytes: 29712000
num_examples: 384000
- name: epoch9
num_bytes: 29712000
num_examples: 384000
- name: epoch4
num_bytes: 29712000
num_examples: 384000
- name: epoch8
num_bytes: 29712000
num_examples: 384000
- name: epoch7
num_bytes: 29712000
num_examples: 384000
- name: epoch2
num_bytes: 29712000
num_examples: 384000
- name: epoch10
num_bytes: 29712000
num_examples: 384000
- name: epoch1
num_bytes: 29712000
num_examples: 384000
download_size: 158595682
dataset_size: 326832000
- config_name: jedidiahkkh__roberta-base_mnli_9c64463f93
features:
- name: epoch
dtype: int32
- name: uid
dtype: int64
- name: logits
sequence: float64
- name: loss
dtype: float64
- name: grad_1norm
dtype: float64
- name: grad_2norm
dtype: float64
- name: grad_infnorm
dtype: float64
- name: label
dtype: int32
splits:
- name: epoch9
num_bytes: 29712000
num_examples: 384000
- name: epoch5
num_bytes: 29712000
num_examples: 384000
- name: epoch2
num_bytes: 29712000
num_examples: 384000
- name: epoch10
num_bytes: 29712000
num_examples: 384000
- name: epoch4
num_bytes: 29712000
num_examples: 384000
- name: epoch0
num_bytes: 29712000
num_examples: 384000
- name: epoch8
num_bytes: 29712000
num_examples: 384000
- name: epoch7
num_bytes: 29712000
num_examples: 384000
- name: epoch6
num_bytes: 29712000
num_examples: 384000
- name: epoch3
num_bytes: 29712000
num_examples: 384000
- name: epoch1
num_bytes: 29712000
num_examples: 384000
download_size: 159987904
dataset_size: 326832000
configs:
- config_name: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525
data_files:
- split: epoch0
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch0-*
- split: epoch3
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch3-*
- split: epoch6
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch6-*
- split: epoch5
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch5-*
- split: epoch9
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch9-*
- split: epoch4
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch4-*
- split: epoch8
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch8-*
- split: epoch7
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch7-*
- split: epoch2
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch2-*
- split: epoch10
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch10-*
- split: epoch1
path: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525/epoch1-*
- config_name: jedidiahkkh__roberta-base_mnli_9c64463f93
data_files:
- split: epoch9
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch9-*
- split: epoch5
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch5-*
- split: epoch2
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch2-*
- split: epoch10
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch10-*
- split: epoch4
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch4-*
- split: epoch0
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch0-*
- split: epoch8
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch8-*
- split: epoch7
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch7-*
- split: epoch6
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch6-*
- split: epoch3
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch3-*
- split: epoch1
path: jedidiahkkh__roberta-base_mnli_9c64463f93/epoch1-*
---
The dataset includes two configurations, one based on the BERT model and the other on the RoBERTa model, for the MNLI task. Each configuration contains data for multiple epochs, with each epoch including features such as epoch number, unique identifier, logits, loss values, gradient norms, and labels. The data size and number of examples are consistent for each epoch.
提供机构：
jedidiahkkh
原始信息汇总
数据集概述
数据集1: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525
- 配置名称: jedidiahkkh__bert-base-uncased_mnli_2f0d7d9525
- 特征:
- epoch: int32
- uid: int64
- logits: float64 (序列)
- loss: float64
- grad_1norm: float64
- grad_2norm: float64
- grad_infnorm: float64
- label: int32
- 分割:
- epoch0: 29712000 字节, 384000 示例
- epoch3: 29712000 字节, 384000 示例
- epoch6: 29712000 字节, 384000 示例
- epoch5: 29712000 字节, 384000 示例
- epoch9: 29712000 字节, 384000 示例
- epoch4: 29712000 字节, 384000 示例
- epoch8: 29712000 字节, 384000 示例
- epoch7: 29712000 字节, 384000 示例
- epoch2: 29712000 字节, 384000 示例
- epoch10: 29712000 字节, 384000 示例
- epoch1: 29712000 字节, 384000 示例
- 下载大小: 158595682 字节
- 数据集大小: 326832000 字节
数据集2: jedidiahkkh__roberta-base_mnli_9c64463f93
- 配置名称: jedidiahkkh__roberta-base_mnli_9c64463f93
- 特征:
- epoch: int32
- uid: int64
- logits: float64 (序列)
- loss: float64
- grad_1norm: float64
- grad_2norm: float64
- grad_infnorm: float64
- label: int32
- 分割:
- epoch9: 29712000 字节, 384000 示例
- epoch5: 29712000 字节, 384000 示例
- epoch2: 29712000 字节, 384000 示例
- epoch10: 29712000 字节, 384000 示例
- epoch4: 29712000 字节, 384000 示例
- epoch0: 29712000 字节, 384000 示例
- epoch8: 29712000 字节, 384000 示例
- epoch7: 29712000 字节, 384000 示例
- epoch6: 29712000 字节, 384000 示例
- epoch3: 29712000 字节, 384000 示例
- epoch1: 29712000 字节, 384000 示例
- 下载大小: 159987904 字节
- 数据集大小: 326832000 字节
