Download source/part2/utils/embedding.py from walke007/anlp-assignment2-part2: direct link, hf CLI and curl.
- Browser
- Download file 819 Bytes
-
https://huggingface.co/walke007/anlp-assignment2-part2/resolve/main/source/part2/utils/embedding.py
- Command line
-
hf download hf://walke007/anlp-assignment2-part2/source/part2/utils/embedding.py
-
curl -L -o embedding.py https://huggingface.co/walke007/anlp-assignment2-part2/resolve/main/source/part2/utils/embedding.py
819 Bytes
| import torch | |
| from torch import nn | |
| # vocab_size, model_dim -> input for embedding layer | |
| class UnEmbedding(nn.Module): | |
| def __init__(self, vocab_size, dimensions_model): | |
| super().__init__() | |
| self.v = vocab_size | |
| self.d = dimensions_model | |
| self.UnEmbedding_stack = nn.Sequential( | |
| nn.Linear(self.d, self.v) | |
| ) | |
| def forward(self, x): | |
| logits = self.UnEmbedding_stack(x) | |
| return logits | |
| class Embedding(nn.Module): | |
| def __init__(self, vocab_size, dimensions_model): | |
| super().__init__() | |
| self.v = vocab_size | |
| self.d = dimensions_model | |
| self.embedding = nn.Embedding(self.v,self.d) | |
| def forward(self, x): | |
| logits = self.embedding(x) | |
| return logits | |