preference-agents-working/kay-mann-train

---
dataset_info:
features:
- name: message_id
dtype: string
- name: from
dtype: string
- name: to
dtype: string
- name: date
dtype: string
- name: subject
dtype: string
- name: content
dtype: string
- name: email_context
dtype: string
- name: token_count_content
dtype: int32
- name: token_count_context
dtype: int32
- name: intent
dtype: string
- name: baseline
struct:
- name: google/gemma-1.1-2b-it
dtype: string
- name: google/gemma-1.1-7b-it
dtype: string
- name: meta-llama/Meta-Llama-3-70B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
dtype: string
- name: mistralai/Mistral-7B-Instruct-v0.2
dtype: string
- name: automatic_eval
struct:
- name: google/gemma-1.1-2b-it
struct:
- name: BERT Cosine Similarity
dtype: float64
- name: BLEU Score
dtype: float64
- name: ROUGE-L Score
dtype: float64
- name: TF-IDF Cosine Similarity
dtype: float64
- name: google/gemma-1.1-7b-it
struct:
- name: BERT Cosine Similarity
dtype: float64
- name: BLEU Score
dtype: float64
- name: ROUGE-L Score
dtype: float64
- name: TF-IDF Cosine Similarity
dtype: float64
- name: meta-llama/Meta-Llama-3-8B-Instruct
struct:
- name: BERT Cosine Similarity
dtype: float64
- name: BLEU Score
dtype: float64
- name: ROUGE-L Score
dtype: float64
- name: TF-IDF Cosine Similarity
dtype: float64
- name: mistralai/Mistral-7B-Instruct-v0.2
struct:
- name: BERT Cosine Similarity
dtype: float64
- name: BLEU Score
dtype: float64
- name: ROUGE-L Score
dtype: float64
- name: TF-IDF Cosine Similarity
dtype: float64
- name: rules
struct:
- name: meta-llama/Meta-Llama-3-70B-Instruct
struct:
- name: meta-llama/Meta-Llama-3-70B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
struct:
- name: meta-llama/Meta-Llama-3-70B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
dtype: string
- name: processed_rules
struct:
- name: meta-llama/Meta-Llama-3-70B-Instruct
struct:
- name: meta-llama/Meta-Llama-3-70B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
struct:
- name: meta-llama/Meta-Llama-3-70B-Instruct
dtype: string
- name: meta-llama/Meta-Llama-3-8B-Instruct
dtype: string
- name: text
dtype: string
- name: naive_ftnk_generated_email
dtype: string
- name: naive_ft_generated_email
dtype: string
splits:
- name: train
num_bytes: 31243436
num_examples: 1332
- name: test
num_bytes: 7924923
num_examples: 333
download_size: 16610705
dataset_size: 39168359
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
---

原始信息汇总

数据集概述

数据集特征

• message_id: 字符串类型
• from: 字符串类型
• to: 字符串类型
• date: 字符串类型
• subject: 字符串类型
• content: 字符串类型
• email_context: 字符串类型
• token_count_content: 整数类型（int32）
• token_count_context: 整数类型（int32）
• intent: 字符串类型
• baseline: 结构类型，包含多个模型名称，均为字符串类型
• automatic_eval: 结构类型，包含多个模型的评估指标，包括BERT Cosine Similarity、BLEU Score、ROUGE-L Score和TF-IDF Cosine Similarity，均为浮点数类型（float64）
• rules: 结构类型，包含多个规则，均为字符串类型
• processed_rules: 结构类型，包含多个处理后的规则，均为字符串类型
• text: 字符串类型
• naive_ftnk_generated_email: 字符串类型
• naive_ft_generated_email: 字符串类型

数据集分割

• train: 1332个样本，占用31243436字节
• test: 333个样本，占用7924923字节

数据集大小

• 下载大小: 16610705字节
• 数据集总大小: 39168359字节

数据文件配置

• config_name: default
• data_files:
  • train: 路径为data/train-*
  • test: 路径为data/test-*