ifinspire/ultrachat-200k-bliss-raw

该数据集是使用实验性BlissyLM转换工具对HuggingFaceH4/ultrachat_200k数据集的train_sft分片进行的词汇级Blissymbolic音译。它保留了源对话结构和角色元数据，同时基于BCI授权词汇表查找添加了Blissymbol标记序列。这不是人工翻译，也不是临床增强与替代沟通指导，而是用于探索Blissymbol标记数据集和小型语言模型实验的早期研究项目。This dataset is a lexical Blissymbolic transliteration of `HuggingFaceH4/ultrachat_200k` `train_sft` using experimental BlissyLM conversion tooling. It preserves the source conversation structure and role metadata while adding Blissymbol token sequences based on BCI Authorized Vocabulary gloss lookup. This is not a human translation and is not clinical AAC guidance, and is intended for early research exploring Blissymbol token datasets and small language-model experiments.