PoliBa-DeepLearning/NanoSocrates
2
0
Fork 0
Code Issues 3 Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Added support for batches

Browse Source
This commit is contained in:
Christian Risi 2025-10-07 12:15:03 +02:00
parent 14b810c451
commit 9b5bb6d5f8
2 changed files with 56 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

57
Playgrounds/embedder.ipynb
View File

File diff suppressed because one or more lines are too long

19
Project_Model/Libs/Embedder/Classes/NanoSocratesEmbedder.py
View File

@ -1,19 +1,13 @@
import torch
from ..Utils import fixed_positional_encoding
# WIP FOR BATCHING
class NanoSocratesEmbedder(torch.nn.Module):
def __init__(
self,
vocabulary_size: int,
embedding_size: int
) -> None:
def __init__(self, vocabulary_size: int, embedding_size: int) -> None:
super().__init__()
self.__embedder = torch.nn.Embedding(
vocabulary_size,
embedding_size
)
self.__embedder = torch.nn.Embedding(vocabulary_size, embedding_size)
def forward(self, tokenized_sentence: list[list[int]]) -> torch.Tensor:
@ -24,11 +18,8 @@ class NanoSocratesEmbedder(torch.nn.Module):
_, SENTENCE_LENGHT, EMBEDDING_SIZE = computed_embeddings.shape # for batching
POSITIONAL_ENCODINGS = fixed_positional_encoding(
SENTENCE_LENGHT,
EMBEDDING_SIZE
SENTENCE_LENGHT, EMBEDDING_SIZE
)
computed_embeddings = computed_embeddings + POSITIONAL_ENCODINGS.unsqueeze(0) # for batching
computed_embeddings = computed_embeddings + POSITIONAL_ENCODINGS # for batching
return computed_embeddings