dnrops
/
transformers
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

add diff file that is the same as the modeling_llama.py

This commit is contained in:
Arthur Zucker 2024-05-10 14:39:14 +02:00
parent 740e5bd35c
commit 7545c5f766
2 changed files with 1479 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1474
src/transformers/models/llama/diff_llama.py Normal file
View File

File diff suppressed because it is too large Load Diff

60
src/transformers/models/llama/modeling_llama.py
View File

@ -197,25 +197,7 @@ class LlamaMLP(nn.Module):
self.act_fn = ACT2FN[config.hidden_act]
def forward(self, x):
if self.config.pretraining_tp > 1:
slice = self.intermediate_size // self.config.pretraining_tp
gate_proj_slices = self.gate_proj.weight.split(slice, dim=0)
up_proj_slices = self.up_proj.weight.split(slice, dim=0)
down_proj_slices = self.down_proj.weight.split(slice, dim=1)
gate_proj = torch.cat(
[F.linear(x, gate_proj_slices[i]) for i in range(self.config.pretraining_tp)], dim=-1
)
up_proj = torch.cat([F.linear(x, up_proj_slices[i]) for i in range(self.config.pretraining_tp)], dim=-1)
intermediate_states = (self.act_fn(gate_proj) * up_proj).split(slice, dim=2)
down_proj = [
F.linear(intermediate_states[i], down_proj_slices[i]) for i in range(self.config.pretraining_tp)
]
down_proj = sum(down_proj)
else:
down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
return down_proj
@ -303,31 +285,12 @@ class LlamaAttention(nn.Module):
output_attentions: bool = False,
use_cache: bool = False,
cache_position: Optional[torch.LongTensor] = None,
**kwargs,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
bsz, q_len, _ = hidden_states.size()
if self.config.pretraining_tp > 1:
key_value_slicing = (self.num_key_value_heads * self.head_dim) // self.config.pretraining_tp
query_slices = self.q_proj.weight.split(
(self.num_heads * self.head_dim) // self.config.pretraining_tp, dim=0
)
key_slices = self.k_proj.weight.split(key_value_slicing, dim=0)
value_slices = self.v_proj.weight.split(key_value_slicing, dim=0)
query_states = [F.linear(hidden_states, query_slices[i]) for i in range(self.config.pretraining_tp)]
query_states = torch.cat(query_states, dim=-1)
key_states = [F.linear(hidden_states, key_slices[i]) for i in range(self.config.pretraining_tp)]
key_states = torch.cat(key_states, dim=-1)
value_states = [F.linear(hidden_states, value_slices[i]) for i in range(self.config.pretraining_tp)]
value_states = torch.cat(value_states, dim=-1)
else:
query_states = self.q_proj(hidden_states)
key_states = self.k_proj(hidden_states)
value_states = self.v_proj(hidden_states)
query_states = self.q_proj(hidden_states)
key_states = self.k_proj(hidden_states)
value_states = self.v_proj(hidden_states)
query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
@ -364,13 +327,7 @@ class LlamaAttention(nn.Module):
attn_output = attn_output.transpose(1, 2).contiguous()
attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
if self.config.pretraining_tp > 1:
attn_output = attn_output.split(self.hidden_size // self.config.pretraining_tp, dim=2)
o_proj_slices = self.o_proj.weight.split(self.hidden_size // self.config.pretraining_tp, dim=1)
attn_output = sum([F.linear(attn_output[i], o_proj_slices[i]) for i in range(self.config.pretraining_tp)])
else:
attn_output = self.o_proj(attn_output)
attn_output = self.o_proj(attn_output)
if not output_attentions:
attn_weights = None
@ -402,7 +359,6 @@ class LlamaFlashAttention2(LlamaAttention):
output_attentions: bool = False,
use_cache: bool = False,
cache_position: Optional[torch.LongTensor] = None,
**kwargs,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
if isinstance(past_key_value, StaticCache):
raise ValueError(
@ -691,7 +647,6 @@ class LlamaDecoderLayer(nn.Module):
output_attentions: Optional[bool] = False,
use_cache: Optional[bool] = False,
cache_position: Optional[torch.LongTensor] = None,
**kwargs,
) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
"""
Args:
@ -707,10 +662,6 @@ class LlamaDecoderLayer(nn.Module):
(see `past_key_values`).
past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
"""
if "padding_mask" in kwargs:
warnings.warn(
"Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
)
residual = hidden_states
@ -725,7 +676,6 @@ class LlamaDecoderLayer(nn.Module):
output_attentions=output_attentions,
use_cache=use_cache,
cache_position=cache_position,
**kwargs,
)
hidden_states = residual + hidden_states