Standard names for embedding mode Given the vocab tensor shape[batch_size, sequence_len]. The following keys are defined: * `FLAT`: Normal mode, return tensor shape will be * [batch_size, sequence_len, embedding_size] * `MEAN`: Mean mode, return tensor shape will be *
| 40 | |
| 41 | |
| 42 | class EmbeddingProcessType(Type): |
| 43 | """Standard names for embedding mode |
| 44 | Given the vocab tensor shape[batch_size, sequence_len]. |
| 45 | The following keys are defined: |
| 46 | * `FLAT`: Normal mode, return tensor shape will be |
| 47 | * [batch_size, sequence_len, embedding_size] |
| 48 | * `MEAN`: Mean mode, return tensor shape will be |
| 49 | * [batch_size, embedding_size] |
| 50 | * `SUM`: Sum mode, return tensor shape will be |
| 51 | * [batch_size, embedding_size] |
| 52 | """ |
| 53 | FLAT = 'flat' |
| 54 | MEAN = 'mean' |
| 55 | SUM = 'sum' |
| 56 | |
| 57 | @classmethod |
| 58 | def str(cls): |
| 59 | return ",".join([cls.FLAT, cls.MEAN, cls.SUM]) |
| 60 | |
| 61 | |
| 62 | class Embedding(torch.nn.Module): |
nothing calls this directly
no outgoing calls
no test coverage detected