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

[LLM] Add quantize_kv optimization for yuan2 model (#10243)

Browse source 
* add initial quantize_kv support for yuan2 model

* fix yuan2 quantize_kv generation

* apply fp16 conv layer optimizations

* disable mlp for quantize_kv
This commit is contained in:
SONG Ge 2024-02-29 16:33:26 +08:00 committed by GitHub
parent a2ed4d714e
commit 13b0bc9075
2 changed files with 184 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
python/llm/src/bigdl/llm/transformers/convert.py
View file

@ -1196,13 +1196,14 @@ def _optimize_post(model, lightweight_bmm=False):
modeling_module_name = model.__class__.__module__
module = importlib.import_module(modeling_module_name)
from bigdl.llm.transformers.models.yuan import yuan_attention_forward
from bigdl.llm.transformers.models.yuan import yuan_mlp_forward
# from bigdl.llm.transformers.models.yuan import yuan_mlp_forward
convert_forward(model,
module.YuanAttention,
yuan_attention_forward
)
convert_forward(model,
module.YuanMLP,
yuan_mlp_forward
)
# disable able mlp_forward for quantize_kv on mtl.
# convert_forward(model,
# module.YuanMLP,
# yuan_mlp_forward
# )
return model

178
python/llm/src/bigdl/llm/transformers/models/yuan.py
View file

@ -32,6 +32,8 @@ from bigdl.llm.utils.common import invalidInputError
from bigdl.llm.transformers.models.utils import apply_rotary_pos_emb, \
apply_rotary_pos_emb_cache_freq_xpu, mlp_fusion_check, fp16_fusion_check
from bigdl.llm.transformers.models.utils import init_kv_cache, extend_kv_cache, append_kv_cache
from bigdl.llm.transformers.models.utils import init_fp8_kv_cache, append_fp8_kv_cache, \
restore_fp8_kv_cache, use_quantize_kv_cache
from bigdl.llm.transformers.models.utils import is_enough_kv_cache_room_4_31
from bigdl.llm.transformers.low_bit_linear import SYM_INT4, FP8E5
@ -144,6 +146,182 @@ def yuan_attention_forward(
past_key_value: Optional[Tuple[torch.Tensor]] = None,
output_attentions: bool = False,
use_cache: bool = False,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
if use_quantize_kv_cache(self.merged_qk_proj, hidden_states):
forward_function = yuan_attention_forward_quantized
else:
forward_function = yuan_attention_forward_origin
return forward_function(
self=self,
hidden_states=hidden_states,
attention_mask=attention_mask,
position_ids=position_ids,
past_key_value=past_key_value,
output_attentions=output_attentions,
use_cache=use_cache,
)
def yuan_attention_forward_quantized(
self,
hidden_states: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
output_attentions: bool = False,
use_cache: bool = False,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
bsz, q_len, _ = hidden_states.size()
device = hidden_states.device
before_hidden_states = None
is_first_step = False
use_fuse_rope = should_use_fuse_rope(self, hidden_states, position_ids)
invalidInputError(use_cache, "use_cache=True is needed")
invalidInputError(not self.use_shareqk, "use_shareqk is not supported for now")
if past_key_value is None:
is_first_step = True
if q_len >= 2:
before_hidden_states = hidden_states[:, -2:, :].transpose(0, 1).half()
else:
before_hidden_states = torch.zeros(2, bsz, self.hidden_size,
dtype=torch.half, device=hidden_states.device)
before_hidden_states[-1:, :, :] = hidden_states[:, -1:, :].transpose(0, 1)
else:
before_hidden_states = past_key_value[2]
this_hidden_states = torch.cat([
before_hidden_states,
hidden_states.transpose(0, 1).half(),
], dim=0)
before_hidden_states = this_hidden_states[-2:, :, ]
value_states = \
self.v_proj(hidden_states).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
if is_first_step:
hidden_states = yuan_localized_filtering_forward(self.lf_gate, hidden_states,
None, hidden_states.dtype)
else:
hidden_states = yuan_localized_filtering_forward(self.lf_gate, hidden_states,
this_hidden_states, hidden_states.dtype)
qk_states = self.merged_qk_proj(hidden_states)
(query_states, key_states) = torch.chunk(qk_states, 2, dim=-1)
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_heads, self.head_dim).transpose(1, 2)
kv_seq_len = key_states.shape[-2]
if past_key_value is not None:
kv_seq_len += past_key_value[0].shape[-2]
cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
if use_fuse_rope:
query_states, key_states = apply_rotary_pos_emb_cache_freq_xpu(query_states,
key_states,
sin, cos,
"yuan",
position_ids)
else:
query_states, key_states = apply_rotary_pos_emb(query_states,
key_states,
cos, sin,
position_ids,
"yuan")
if past_key_value is None:
# should use origin attn here
attn_weights = torch.matmul(query_states,
key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
invalidInputError(attn_weights.size() == (bsz, self.num_heads, q_len, kv_seq_len),
"Attention weights should be of size "
f"{(bsz, self.num_heads, q_len, kv_seq_len)}, "
f"but is {attn_weights.size()}")
if attention_mask is not None:
invalidInputError(attention_mask.size() == (bsz, 1, q_len, kv_seq_len),
f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, "
f"but is {attention_mask.size()}")
attn_weights = attn_weights + attention_mask
attn_weights = torch.max(attn_weights,
torch.tensor(torch.finfo(attn_weights.dtype).min))
# upcast attention to fp32
attn_weights = nn.functional.softmax(attn_weights, dim=-1,
dtype=torch.float32).to(query_states.dtype)
attn_output = torch.matmul(attn_weights, value_states)
if use_cache:
k_cache, v_cache = init_fp8_kv_cache(
bsz, self.num_heads, kv_seq_len, self.head_dim, device=device
)
key_states, value_states = append_fp8_kv_cache(k_cache, v_cache,
key_states, value_states)
past_key_value = (key_states, value_states, before_hidden_states)
else:
k_cache, v_cache, _ = past_key_value
key_states, value_states = append_fp8_kv_cache(k_cache, v_cache,
key_states, value_states)
past_key_value = (key_states, value_states, before_hidden_states)
# torch.matmul
if query_states.size(2) != 1 or device.type != 'xpu':
key_states, value_states = restore_fp8_kv_cache(key_states, value_states,
query_states.dtype)
attn_weights = torch.matmul(query_states, key_states.transpose(2, 3))
else:
import linear_q4_0
attn_weights = linear_q4_0.query_key_fp8_matmul(query_states, key_states)
attn_weights = attn_weights / math.sqrt(self.head_dim)
invalidInputError(attn_weights.size() == (bsz, self.num_heads, q_len, kv_seq_len),
"Attention weights should be of size "
f"{(bsz, self.num_heads, q_len, kv_seq_len)}, "
f"but is {attn_weights.size()}")
if attention_mask is not None:
invalidInputError(attention_mask.size() == (bsz, 1, q_len, kv_seq_len),
f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, "
f"but is {attention_mask.size()}")
attn_weights = attn_weights + attention_mask
attn_weights = torch.max(attn_weights,
torch.tensor(torch.finfo(attn_weights.dtype).min))
# upcast attention to fp32
attn_weights = nn.functional.softmax(attn_weights, dim=-1,
dtype=torch.float32).to(query_states.dtype)
if query_states.size(2) != 1 or device.type != 'xpu':
attn_output = torch.matmul(attn_weights, value_states)
else:
import linear_q4_0
attn_output = linear_q4_0.attn_value_fp8_matmul(attn_weights,
value_states.transpose(-1, -2))
invalidInputError(attn_output.size() == (bsz, self.num_heads, q_len, self.head_dim),
"`attn_output` should be of size "
f"{(bsz, self.num_heads, q_len, self.head_dim)}, "
f"but is {attn_output.size()}")
attn_output = attn_output.transpose(1, 2)
attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
attn_output = self.o_proj(attn_output)
if not output_attentions:
attn_weights = None
return attn_output, attn_weights, past_key_value
def yuan_attention_forward_origin(
self,
hidden_states: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
output_attentions: bool = False,
use_cache: bool = False,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
use_fuse_rope = should_use_fuse_rope(self, hidden_states, position_ids)
bsz, q_len, _ = hidden_states.size()