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

Fixed a bug over task 4

Browse Source
This commit is contained in:
Christian Risi 2025-10-12 12:22:38 +02:00
parent ab3d68bc13
commit f463f699cf
2 changed files with 4 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Playgrounds/nanosocrates-train-experiment-2.py
View File

@ -203,7 +203,7 @@ while current_epoch < MAX_EPOCHS:
decoder_only_optim.zero_grad() decoder_only_optim.zero_grad()
pred_logits = DECODER_ONLY((enc_x, enc_x_pad)) pred_logits = DECODER_ONLY((dec_x, dec_x_pad))
pred_logits = pred_logits.permute(0, 2, 1) pred_logits = pred_logits.permute(0, 2, 1)
loss: torch.Tensor = decoder_ce(pred_logits, tgt) loss: torch.Tensor = decoder_ce(pred_logits, tgt)
@ -291,7 +291,7 @@ while current_epoch < MAX_EPOCHS:
pred_logits = DECODER_ONLY((enc_x, enc_x_pad)) pred_logits = DECODER_ONLY((dec_x, dec_x_pad))
pred_logits = pred_logits.permute(0, 2, 1) pred_logits = pred_logits.permute(0, 2, 1)

5
Project_Model/Libs/Transformer/Utils/truncate_rdf_list.py
View File

@ -27,7 +27,6 @@ def truncate_rdf_list(
END_OF_TRIPLES.append(i + 1) END_OF_TRIPLES.append(i + 1)
TRIPLES_TOKENS: list[int] = [] TRIPLES_TOKENS: list[int] = []
TARGET_TRIPLES: list[int] = []
start_of_triple = 0 start_of_triple = 0
exit_loop = False exit_loop = False
@ -56,10 +55,10 @@ def truncate_rdf_list(
EOT = END_OF_TRIPLES.popleft() EOT = END_OF_TRIPLES.popleft()
TRIPLE = sequence[start_of_triple:EOT] TRIPLE = sequence[start_of_triple:EOT]
TARGET_TRIPLES.extend(TRIPLE) TRIPLES_TOKENS.extend(TRIPLE)
start_of_triple = EOT start_of_triple = EOT
return (TRIPLES_TOKENS, TARGET_TRIPLES) return (TRIPLES_TOKENS, TRIPLES_TOKENS)