ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update gemma.py

Browse source
This commit is contained in:
Daniel Han-Chen 2024-02-25 04:29:01 +11:00
commit bc428f4e7b
1 changed files with 13 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

22
unsloth/models/gemma.py
View file

@ -78,6 +78,8 @@ class FastGemmaRotaryEmbedding(torch.nn.Module):
emb = torch.cat((freqs, freqs), dim=-1)
self.cos_cached = emb.cos().to(dtype=torch.bfloat16)
self.sin_cached = emb.sin().to(dtype=torch.bfloat16)
print(self.cos_cached)
print(self.sin_cached)
def forward(self, x, position_ids, seq_len=None):
# x: [bs, num_attention_heads, seq_len, head_size]
@ -87,15 +89,17 @@ class FastGemmaRotaryEmbedding(torch.nn.Module):
)
length = position_ids.shape[1]
if self.cos_cached is None:
position_ids = torch.arange(self.max_position_embeddings, device=x.device, dtype=torch.int64).unsqueeze(0)
inv_freq_expanded = self.inv_freq[None, :, None].float().expand(1, -1, 1)
position_ids_expanded = position_ids[:, None, :].float()
freqs = (inv_freq_expanded @ position_ids_expanded).transpose(1, 2)
emb = torch.cat((freqs, freqs), dim=-1)
self.cos_cached = emb.cos().to(dtype=x.dtype)
self.sin_cached = emb.sin().to(dtype=x.dtype)
pass
#if self.cos_cached is None:
position_ids = torch.arange(self.max_position_embeddings, device=x.device, dtype=torch.int64).unsqueeze(0)
inv_freq_expanded = self.inv_freq[None, :, None].float().expand(1, -1, 1)
freqs = (inv_freq_expanded @ position_ids_expanded).transpose(1, 2)
emb = torch.cat((freqs, freqs), dim=-1)
self.cos_cached = emb.cos().to(dtype=x.dtype)
self.sin_cached = emb.sin().to(dtype=x.dtype)
# pass
print(self.cos_cached)
print(self.sin_cached)
raise
return self.cos_cached[:,:length], self.sin_cached[:,:length]
pass