Merge pull request bitsandbytes-foundation#14 from SirRob1997/main

[FIX] passing of sparse in StableEmbedding
techthiyanes · Nov 29, 2021 · 262350c · 262350c
2 parents 037022e + 67a1283
commit 262350c
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/bitsandbytes/nn/modules.py b/bitsandbytes/nn/modules.py
@@ -15,8 +15,8 @@
 class StableEmbedding(torch.nn.Embedding):
  def __init__(self, num_embeddings: int, embedding_dim: int, padding_idx: Optional[int] = None,
  max_norm: Optional[float] = None, norm_type: float = 2., scale_grad_by_freq: bool = False,
- sparse: bool = True, _weight: Optional[Tensor] = None) -> None:
- super(StableEmbedding, self).__init__(num_embeddings, embedding_dim, padding_idx, max_norm, norm_type, scale_grad_by_freq, False, _weight)
+ sparse: bool = False, _weight: Optional[Tensor] = None) -> None:
+ super(StableEmbedding, self).__init__(num_embeddings, embedding_dim, padding_idx, max_norm, norm_type, scale_grad_by_freq, sparse, _weight)
  self.norm = torch.nn.LayerNorm(embedding_dim)
  GlobalOptimManager.get_instance().register_parameters(self.weight)
  GlobalOptimManager.get_instance().override_config(self.weight, 'optim_bits', 32)