dialogue-bart-base-chinese

dialogue-bart-base-chinese

This is a seq2seq model fine-tuned on several Chinese dialogue datasets, from bart-base-chinese.

Spaces

Now you can experience our model on HuggingFace Spaces HIT-TMG/dialogue-bart-large-chinese .

Datasets

We utilize 4 Chinese dialogue datasets from LUGE

Data format

Input: [CLS] 对话历史：<history> 知识：<knowledge> [SEP]

Output: [CLS] <response> [SEP]

Example

from transformers import BertTokenizer, BartForConditionalGeneration
# Note that tokenizer is an object of BertTokenizer, instead of BartTokenizer
tokenizer = BertTokenizer.from_pretrained("HIT-TMG/dialogue-bart-base-chinese")
model = BartForConditionalGeneration.from_pretrained("HIT-TMG/dialogue-bart-base-chinese")
# an example from CPC dev data
history = ["可以 认识 一下 吗 ？", "当然 可以 啦 ， 你好 。", "嘿嘿 你好 ， 请问 你 最近 在 忙 什么 呢 ？", "我 最近 养 了 一只 狗狗 ， 我 在 训练 它 呢 。"]
history_str = "对话历史：" + tokenizer.sep_token.join(history)
input_ids = tokenizer(history_str, return_tensors='pt').input_ids
output_ids = model.generate(input_ids)[0]
print(tokenizer.decode(output_ids, skip_special_tokens=True))

Contact

If you encounter any issue, feel free to contact us via the email: <u>[email protected]</u>