vikp/code_instructions_filtered
收藏Hugging Face2023-09-04 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/vikp/code_instructions_filtered
下载链接
链接失效反馈官方服务:
资源简介:
---
dataset_info:
features:
- name: output
dtype: string
- name: instruction
dtype: string
- name: kind
dtype: string
splits:
- name: train
num_bytes: 250321474.7560524
num_examples: 136147
download_size: 146821284
dataset_size: 250321474.7560524
---
# Dataset Card for "code_instructions_filtered"
This includes data from [xlcost](https://huggingface.co/datasets/vikp/xlcost_filtered_2k), [evol instruct](https://huggingface.co/datasets/vikp/evol_instruct_code_filtered_39k), [code alpaca](https://huggingface.co/datasets/vikp/evol_codealpaca_filtered_87k), and [code instructions](https://huggingface.co/datasets/vikp/code_instructions_filtered_7k). Data is filtered based on quality and learning value.
When used to fine-tune code llama 7B, achieves a `.62` humaneval score.
提供机构:
vikp
原始信息汇总
数据集卡片 "code_instructions_filtered"
数据集信息
特征
- output: 类型为字符串
- instruction: 类型为字符串
- kind: 类型为字符串
分割
- train:
- 字节数: 250321474.7560524
- 样本数: 136147
大小
- 下载大小: 146821284
- 数据集大小: 250321474.7560524



