add rwkv v5 attention kernel (#9927)
This commit is contained in:
parent
054952f82f
commit
453df868c9
3 changed files with 157 additions and 2 deletions
|
|
@ -932,4 +932,12 @@ def _optimize_post(model, lightweight_bmm=False):
|
||||||
convert_forward(model,
|
convert_forward(model,
|
||||||
module.RwkvSelfAttention,
|
module.RwkvSelfAttention,
|
||||||
rwkv_attention_forward)
|
rwkv_attention_forward)
|
||||||
|
elif model.config.model_type == "rwkv5":
|
||||||
|
# rwkv v5
|
||||||
|
modeling_module_name = model.__class__.__module__
|
||||||
|
module = importlib.import_module(modeling_module_name)
|
||||||
|
from bigdl.llm.transformers.models.rwkv5 import rwkv_attention_forward
|
||||||
|
convert_forward(model,
|
||||||
|
module.RwkvSelfAttention,
|
||||||
|
rwkv_attention_forward)
|
||||||
return model
|
return model
|
||||||
|
|
|
||||||
|
|
@ -60,8 +60,9 @@ def rwkv_linear_attention_xpu(
|
||||||
|
|
||||||
time_decay = -torch.exp(time_decay)
|
time_decay = -torch.exp(time_decay)
|
||||||
|
|
||||||
|
# `num_state`, `den_state`, `max_state` will be modified during this call
|
||||||
import linear_q4_0
|
import linear_q4_0
|
||||||
output = linear_q4_0.rwkv_attention_with_state(
|
output = linear_q4_0.rwkv_linear_attention_v4(
|
||||||
time_decay,
|
time_decay,
|
||||||
time_first,
|
time_first,
|
||||||
key,
|
key,
|
||||||
|
|
@ -81,7 +82,7 @@ def rwkv_attention_forward(
|
||||||
self,
|
self,
|
||||||
hidden: torch.Tensor,
|
hidden: torch.Tensor,
|
||||||
state: List[torch.Tensor]=None,
|
state: List[torch.Tensor]=None,
|
||||||
use_cache=False,
|
use_cache: bool=False,
|
||||||
):
|
):
|
||||||
receptance, key, value, state = self.extract_key_value(hidden, state=state)
|
receptance, key, value, state = self.extract_key_value(hidden, state=state)
|
||||||
layer_state = tuple(s[:, :, self.layer_id] for s in state[2:]) if state is not None else None
|
layer_state = tuple(s[:, :, self.layer_id] for s in state[2:]) if state is not None else None
|
||||||
|
|
|
||||||
146
python/llm/src/bigdl/llm/transformers/models/rwkv5.py
Normal file
146
python/llm/src/bigdl/llm/transformers/models/rwkv5.py
Normal file
|
|
@ -0,0 +1,146 @@
|
||||||
|
#
|
||||||
|
# Copyright 2016 The BigDL Authors.
|
||||||
|
#
|
||||||
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||||
|
# you may not use this file except in compliance with the License.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
|
#
|
||||||
|
# Some parts of this file is adapted from
|
||||||
|
# https://huggingface.co/RWKV/rwkv-5-world-3b/blob/main/modeling_rwkv5.py
|
||||||
|
# which is licensed under Apache License 2.0:
|
||||||
|
#
|
||||||
|
# Copyright 2023 Bo Peng and HuggingFace Inc. team.
|
||||||
|
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
||||||
|
#
|
||||||
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||||
|
# you may not use this file except in compliance with the License.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
|
|
||||||
|
import torch
|
||||||
|
import torch.nn.functional as F
|
||||||
|
|
||||||
|
from typing import List
|
||||||
|
|
||||||
|
|
||||||
|
def rwkv_linear_attention_xpu(
|
||||||
|
B: int,
|
||||||
|
H: int,
|
||||||
|
S: int,
|
||||||
|
T: int,
|
||||||
|
hidden: torch.Tensor,
|
||||||
|
time_decay: torch.Tensor,
|
||||||
|
time_first: torch.Tensor,
|
||||||
|
receptance: torch.Tensor,
|
||||||
|
key: torch.Tensor,
|
||||||
|
value: torch.Tensor,
|
||||||
|
gate: torch.Tensor,
|
||||||
|
lxw: torch.Tensor,
|
||||||
|
lxb: torch.Tensor,
|
||||||
|
ow: torch.nn.Linear,
|
||||||
|
state: torch.Tensor,
|
||||||
|
):
|
||||||
|
key = key.float().view(B, T, H, S).transpose(1, 2)
|
||||||
|
value = value.float().view(B, T, H, S).transpose(1, 2)
|
||||||
|
receptance = receptance.float().view(B, T, H, S).transpose(1, 2)
|
||||||
|
|
||||||
|
time_decay = torch.exp(-torch.exp(time_decay.float()))
|
||||||
|
time_first = time_first.float()
|
||||||
|
|
||||||
|
state = state.contiguous().float()
|
||||||
|
|
||||||
|
# `state` will be modified during this call
|
||||||
|
import linear_q4_0
|
||||||
|
out = linear_q4_0.rwkv_linear_attention_v5(
|
||||||
|
time_decay,
|
||||||
|
time_first,
|
||||||
|
receptance,
|
||||||
|
key,
|
||||||
|
value,
|
||||||
|
state,
|
||||||
|
)
|
||||||
|
|
||||||
|
lxw = lxw.float()
|
||||||
|
lxb = lxb.float()
|
||||||
|
|
||||||
|
out = out.reshape(B * T, H * S)
|
||||||
|
out = F.group_norm(out, num_groups=H, weight=lxw, bias=lxb).reshape(B, T, H * S)
|
||||||
|
out = out.to(dtype=hidden.dtype) * gate
|
||||||
|
# out = out @ ow
|
||||||
|
out = ow(out)
|
||||||
|
return out, state
|
||||||
|
|
||||||
|
|
||||||
|
def rwkv_attention_forward(
|
||||||
|
self,
|
||||||
|
hidden: torch.Tensor,
|
||||||
|
state: List[torch.Tensor]=None,
|
||||||
|
use_cache: bool=False,
|
||||||
|
seq_mode: bool=True,
|
||||||
|
):
|
||||||
|
B = hidden.shape[0]
|
||||||
|
H = self.time_decay.shape[0]
|
||||||
|
S = hidden.shape[-1] // H
|
||||||
|
T = hidden.shape[1]
|
||||||
|
|
||||||
|
receptance, key, value, gate, state = self.extract_key_value(B, H, S, T, hidden, state=state)
|
||||||
|
layer_state = state[1][:, :, :, :, self.layer_id] if state is not None else None
|
||||||
|
|
||||||
|
if hidden.device.type == "xpu":
|
||||||
|
rwkv, layer_state = rwkv_linear_attention_xpu(
|
||||||
|
B,
|
||||||
|
H,
|
||||||
|
S,
|
||||||
|
T,
|
||||||
|
hidden,
|
||||||
|
self.time_decay,
|
||||||
|
self.time_faaaa,
|
||||||
|
receptance,
|
||||||
|
key,
|
||||||
|
value,
|
||||||
|
gate,
|
||||||
|
self.ln_x.weight,
|
||||||
|
self.ln_x.bias,
|
||||||
|
self.output,
|
||||||
|
state=layer_state,
|
||||||
|
)
|
||||||
|
else:
|
||||||
|
from transformers.models.rwkv.modeling_rwkv import rwkv_linear_attention_cpu
|
||||||
|
rwkv, layer_state = rwkv_linear_attention_cpu(
|
||||||
|
B,
|
||||||
|
H,
|
||||||
|
S,
|
||||||
|
T,
|
||||||
|
self.num_attention_heads,
|
||||||
|
hidden,
|
||||||
|
self.time_decay,
|
||||||
|
self.time_faaaa,
|
||||||
|
receptance,
|
||||||
|
key,
|
||||||
|
value,
|
||||||
|
gate,
|
||||||
|
self.ln_x.weight,
|
||||||
|
self.ln_x.bias,
|
||||||
|
self.output.weight.t(),
|
||||||
|
state=layer_state,
|
||||||
|
)
|
||||||
|
|
||||||
|
if layer_state is not None:
|
||||||
|
state[1][:, :, :, :, self.layer_id] = layer_state
|
||||||
|
|
||||||
|
return rwkv, state
|
||||||
Loading…
Reference in a new issue