ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

LLM: fix chatglm3 sdp to support speculative decoding (#9900)

Browse source 
* fix chatglm3

* fix

* update

* meet code review

* fix
This commit is contained in:
Ruonan Wang 2024-01-16 11:29:13 +08:00 committed by GitHub
parent 9f34da7cdb
commit 8d7326ae03
1 changed files with 6 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
python/llm/src/bigdl/llm/transformers/models/chatglm2.py
View file

@ -373,11 +373,15 @@ def core_attn_forward_8eb45c(self, query_layer, key_layer, value_layer, attentio
key_layer, key_layer,
value_layer, value_layer,
is_causal=True) is_causal=True)
elif attention_mask is None: else:
head_dim = query_layer.size(-1) head_dim = query_layer.size(-1)
attn = torch.matmul(query_layer, attn = torch.matmul(query_layer,
key_layer.transpose(2, 3)) / math.sqrt(head_dim) key_layer.transpose(2, 3)) / math.sqrt(head_dim)
if L == S: if attention_mask is not None:
attention_mask = ~attention_mask
attention_mask.masked_fill_(attention_mask.logical_not(), float("-inf"))
attn += attention_mask
elif L == S:
# first token, need attention mask # first token, need attention mask
attn_bias = torch.zeros(L, S, dtype=query_layer.dtype, attn_bias = torch.zeros(L, S, dtype=query_layer.dtype,
device=query_layer.device) device=query_layer.device)
@ -388,12 +392,6 @@ def core_attn_forward_8eb45c(self, query_layer, key_layer, value_layer, attentio
attn += attn_bias attn += attn_bias
attn = torch.softmax(attn, -1) attn = torch.softmax(attn, -1)
context_layer = torch.matmul(attn, value_layer) context_layer = torch.matmul(attn, value_layer)
else:
attention_mask = ~attention_mask
context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer,
key_layer,
value_layer,
attention_mask)
context_layer = context_layer.permute(2, 0, 1, 3) context_layer = context_layer.permute(2, 0, 1, 3)
new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,) new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
context_layer = context_layer.reshape(*new_context_layer_shape) context_layer = context_layer.reshape(*new_context_layer_shape)