Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Refactor OPT model #36101

Merged
merged 3 commits into from
Feb 10, 2025
Merged

Refactor OPT model #36101

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0bbf93c
remove cross attention
jiqing-feng Feb 8, 2025
ea574de
remove is_decoder
jiqing-feng Feb 8, 2025
365ae27
fix pkv
jiqing-feng Feb 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
109 changes: 16 additions & 93 deletions src/transformers/models/opt/modeling_opt.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -98,7 +98,6 @@ class OPTAttention(nn.Module):
def __init__(
self,
config: OPTConfig,
is_decoder: bool = False,
**kwargs,
):
super().__init__()
Expand All @@ -117,7 +116,6 @@ def __init__(
f" and `num_heads`: {self.num_heads})."
)
self.scaling = self.head_dim**-0.5
self.is_decoder = is_decoder

self.k_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=self.enable_bias)
self.v_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=self.enable_bias)
Expand All @@ -130,7 +128,6 @@ def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int) -> torch.Tensor:
def forward(
self,
hidden_states: torch.Tensor,
key_value_states: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
Expand All @@ -139,44 +136,19 @@ def forward(
position_ids: Optional[torch.Tensor] = None,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
"""Input shape: Batch x Time x Channel"""

# if key_value_states are provided this layer is used as a cross-attention layer
# for the decoder
is_cross_attention = key_value_states is not None

bsz, tgt_len, _ = hidden_states.size()

# get query proj
query_states = self.q_proj(hidden_states) * self.scaling
# get key, value proj
if is_cross_attention and past_key_value is not None:
# reuse k,v, cross_attentions
key_states = past_key_value[0]
value_states = past_key_value[1]
elif is_cross_attention:
# cross_attentions
key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
elif past_key_value is not None:
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
if past_key_value is not None:
# reuse k, v, self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
key_states = torch.cat([past_key_value[0], key_states], dim=2)
value_states = torch.cat([past_key_value[1], value_states], dim=2)
else:
# self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)

if self.is_decoder:
# if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states.
# Further calls to cross_attention layer can then reuse all cross-attention
# key/value_states (first "if" case)
# if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of
# all previous decoder key/value_states. Further calls to uni-directional self-attention
# can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
# if encoder bi-directional self-attention `past_key_value` is always `None`
past_key_value = (key_states, value_states)

past_key_value = (key_states, value_states)

proj_shape = (bsz * self.num_heads, -1, self.head_dim)
query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
Expand Down Expand Up @@ -268,52 +240,26 @@ def __init__(self, *args, **kwargs):
def forward(
self,
hidden_states: torch.Tensor,
key_value_states: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
output_attentions: bool = False,
position_ids: Optional[torch.Tensor] = None,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
"""Input shape: Batch x Time x Channel"""

# if key_value_states are provided this layer is used as a cross-attention layer
# for the decoder
is_cross_attention = key_value_states is not None

bsz, _, _ = hidden_states.size()

# get query proj
query_states = self.q_proj(hidden_states)
# get key, value proj
if is_cross_attention and past_key_value is not None:
# reuse k,v, cross_attentions
key_states = past_key_value[0]
value_states = past_key_value[1]
elif is_cross_attention:
# cross_attentions
key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
elif past_key_value is not None:
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
if past_key_value is not None:
# reuse k, v, self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
key_states = torch.cat([past_key_value[0], key_states], dim=2)
value_states = torch.cat([past_key_value[1], value_states], dim=2)
else:
# self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)

if self.is_decoder:
# if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states.
# Further calls to cross_attention layer can then reuse all cross-attention
# key/value_states (first "if" case)
# if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of
# all previous decoder key/value_states. Further calls to uni-directional self-attention
# can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
# if encoder bi-directional self-attention `past_key_value` is always `None`
past_key_value = (key_states, value_states)

past_key_value = (key_states, value_states)

query_length = query_states.shape[1]
tgt_len = key_states.shape[-2]
Expand Down Expand Up @@ -380,7 +326,6 @@ class OPTSdpaAttention(OPTAttention):
def forward(
self,
hidden_states: torch.Tensor,
key_value_states: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
Expand All @@ -399,44 +344,22 @@ def forward(
layer_head_mask=layer_head_mask,
past_key_value=past_key_value,
output_attentions=output_attentions,
key_value_states=key_value_states,
) # TODO after merge add position_ids=position_ids
is_cross_attention = key_value_states is not None

bsz, q_len, _ = hidden_states.size()

query_states = self.q_proj(hidden_states) * self.scaling
query_states = self._shape(query_states, -1, bsz)

# get key, value proj
if is_cross_attention and past_key_value is not None:
# reuse k,v, cross_attentions
key_states = past_key_value[0]
value_states = past_key_value[1]
elif is_cross_attention:
# cross_attentions
key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
elif past_key_value is not None:
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
if past_key_value is not None:
# reuse k, v, self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
key_states = torch.cat([past_key_value[0], key_states], dim=2)
value_states = torch.cat([past_key_value[1], value_states], dim=2)
else:
# self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)

if self.is_decoder:
# if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states.
# Further calls to cross_attention layer can then reuse all cross-attention
# key/value_states (first "if" case)
# if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of
# all previous decoder key/value_states. Further calls to uni-directional self-attention
# can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
# if encoder bi-directional self-attention `past_key_value` is always `None`
past_key_value = (key_states, value_states)

past_key_value = (key_states, value_states)

# shape now is (bsz, num_heads, seq_len, head_dim), all are continuous

Expand Down Expand Up @@ -480,7 +403,7 @@ def __init__(self, config: OPTConfig):
super().__init__()
self.embed_dim = config.hidden_size

self.self_attn = OPT_ATTENTION_CLASSES[config._attn_implementation](config=config, is_decoder=True)
self.self_attn = OPT_ATTENTION_CLASSES[config._attn_implementation](config=config)

self.do_layer_norm_before = config.do_layer_norm_before
self.dropout = config.dropout
Expand Down