Menlo/Ichigo-instruction-tokenized-v0.2

数据集包含以下几部分：
1. VTSNLP-instruct-filtered: 包含文本提示、答案、压缩提示和对话内容等字段。
2. english-multiturn: 包含对话内容和索引等字段。
3. instruction-speech-v1: 包含索引、文本提示、答案、对话内容等字段。
4. instruction-speech-v1-rephrase: 包含索引、原始答案、重新表达的答案、差异级别、压缩提示和对话内容等字段。
5. instruction-speech-v2: 包含索引、文本提示、答案、对话内容等字段。
6. sailor2-instruct: 包含提示ID、消息（包括角色和内容）、语言、类别等字段。
7. transcription-speech-vi-en-550k: 包含文本提示、答案、压缩提示和对话内容等字段。

The dataset consists of the following parts:
1. VTSNLP-instruct-filtered: Includes fields such as text prompt, answer, compressed prompt, and conversation content.
2. english-multiturn: Includes fields such as conversation content and index.
3. instruction-speech-v1: Includes fields such as index, text prompt, answer, and conversation content.
4. instruction-speech-v1-rephrase: Includes fields such as index, original answer, rephrased answer, difference level, compressed prompt, and conversation content.
5. instruction-speech-v2: Includes fields such as index, text prompt, answer, and conversation content.
6. sailor2-instruct: Includes fields such as prompt ID, messages (including role and content), language, category, etc.
7. transcription-speech-vi-en-550k: Includes fields such as text prompt, answer, compressed prompt, and conversation content.