Fills the input Tensor with values according to the method described in Transformers without Tears: Improving the Normalization of Self-Attention

Create a list of timesteps to use from an original diffusion process, given the number of timesteps we want to take from equally-sized portio

Split a tensor along its last dimension. Arguments: tensor: input tensor. num_partitions: number of partitions to split the tensor

Make sure we are sharing the input and output embeddings. Export to TorchScript can't handle parameter sharing so we are cloning them ins

Filter a distribution of logits using top-k and/or nucleus (top-p) filtering Args: logits: logits distribution shape (vocabulary

L2 total variation loss, as in Mahendran et al.

Init method based on N(0, sigma).

Init method based on N(0, sigma).

Update the reweighting using losses from a model. Sub-classes should override this method to update the reweighting using lo

Get a numpy array of weights, one per diffusion step. The weights needn't be normalized, but must be positive.

Fills the input Tensor with values according to the method described in Understanding the difficulty of training deep feedforward neural networks

Fills the input Tensor with values according to the method described in Understanding the difficulty of training deep feedforward neural networks

ans_num = 1 适用于 Train 数据只有 1 条 answer 取第一条情况 ans_num > 1 适用于 Dev 数据有多条 answer 情况 Input: input_ids: input_ids (text +

samples: 一个sample长这样{"text": "hello world"}

samples: 一个sample长这样{"text": "hello world"}

samples: 一个sample长这样{"text": "hello world"}

Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the `text` and `kwargs` arguments