Menlo/Ichigo-instruction-tokenized-v0.1

该数据集包含多个配置，每个配置如下：
1. VTSNLP-instruct-filtered：包含指令和对应的输入、输出，以及类别和最大长度等信息。
2. english-multiturn：包含多轮对话的内容和角色信息。
3. instruction-speech-v1：包含指令文本提示和回答，以及相关长度信息。
4. instruction-speech-v1-rephrase-ichigo-tokens：包含原始回答和重写后的回答，以及差异级别等信息。
5. instruction-speech-v2：与instruction-speech-v1类似，包含指令文本提示和回答。
6. sailor2-instruct：包含指令ID，消息内容及其角色，类别等信息。
7. transcription-speech-v1-vi-en-550k：包含文本提示和回答，以及对话内容。

The dataset consists of multiple configurations, each as follows:
1. VTSNLP-instruct-filtered: Includes instructions and corresponding inputs, outputs, categories, and maximum lengths.
2. english-multiturn: Includes content and role information for multi-turn conversations.
3. instruction-speech-v1: Includes text prompts for instructions and answers, along with related length information.
4. instruction-speech-v1-rephrase-ichigo-tokens: Includes original and rephrased answers, along with difference levels.
5. instruction-speech-v2: Similar to instruction-speech-v1, includes text prompts for instructions and answers.
6. sailor2-instruct: Includes instruction ID, message content and role, category, etc.
7. transcription-speech-v1-vi-en-550k: Includes text prompts and answers, along with conversation content.