MCPcopy Index your code
hub / github.com/RingBDStack/GDAP / freeze_embeds

Function freeze_embeds

seq2seq/utils.py:550–566  ·  view source on GitHub ↗

Freeze token embeddings and positional embeddings for bart, just token embeddings for t5.

(model)

Source from the content-addressed store, hash-verified

548
549
550def freeze_embeds(model):
551 """Freeze token embeddings and positional embeddings for bart, just token embeddings for t5."""
552 model_type = model.config.model_type
553
554 if model_type == "t5":
555 freeze_params(model.shared)
556 for d in [model.encoder, model.decoder]:
557 freeze_params(d.embed_tokens)
558 elif model_type == "fsmt":
559 for d in [model.model.encoder, model.model.decoder]:
560 freeze_params(d.embed_positions)
561 freeze_params(d.embed_tokens)
562 else:
563 freeze_params(model.model.shared)
564 for d in [model.model.encoder, model.model.decoder]:
565 freeze_params(d.embed_positions)
566 freeze_params(d.embed_tokens)
567
568
569def grad_status(model: nn.Module) -> Iterable:

Callers

nothing calls this directly

Calls 1

freeze_paramsFunction · 0.85

Tested by

no test coverage detected