ustcJin

where is MxTorch.....

4

e..may be delete

In the function get_adagrad_updates, classical adadelta use recent exp_sqr_grads and exp_sqr_grads, but in your algorithm, the exp_sqr_grads and exp_sqr_grads are accumulated from the first time, while adagrad express like this。

finetune_for_instruction data.py的理解问题

5

你好，https://github.com/FlagOpen/FlagEmbedding/blob/master/FlagEmbedding/llm_reranker/finetune_for_instruction/data.py，这里面 pos 和 neg 我的理解应该是Label对应 Yes 和 No, 为什么所有的query+pos-passage或者neg-passage 都拼接 Yes，怎么理解啊