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

refactor qwen2 (#11087)

Browse source
This commit is contained in:
Yishuo Wang 2024-05-21 16:53:42 +08:00 committed by GitHub
parent 492ed3fd41
commit f00625f9a4
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 97 additions and 518 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
python/llm/src/ipex_llm/transformers/convert.py
View file

@ -717,6 +717,10 @@ def _optimize_pre(model):
# baichuan2-7B # baichuan2-7B
from ipex_llm.transformers.models.baichuan2 import pre_compute_inv_freq from ipex_llm.transformers.models.baichuan2 import pre_compute_inv_freq
model.apply(pre_compute_inv_freq) model.apply(pre_compute_inv_freq)
# for qwen2
if model.config.model_type == "qwen2":
from ipex_llm.transformers.models.qwen2 import merge_qkv
model.apply(merge_qkv)
if model.config.model_type == "stablelm": if model.config.model_type == "stablelm":
# For stablelm-zephyr-3b and stablelm-2-zephyr-1_6b # For stablelm-zephyr-3b and stablelm-2-zephyr-1_6b
from ipex_llm.transformers.models.stablelm import merge_qkv from ipex_llm.transformers.models.stablelm import merge_qkv

611
python/llm/src/ipex_llm/transformers/models/qwen2.py
View file

@ -42,59 +42,24 @@ import warnings
from typing import TYPE_CHECKING, Optional, Tuple, Union, Callable, List from typing import TYPE_CHECKING, Optional, Tuple, Union, Callable, List
import torch import torch
import torch.nn as nn from torch.nn.functional import scaled_dot_product_attention as sdpa
import torch.nn.functional as F
from ipex_llm.transformers.models.llama import repeat_kv from ipex_llm.transformers.models.utils import should_use_fuse_rope
from ipex_llm.transformers.models.utils import extend_kv_cache, append_kv_cache
from ipex_llm.transformers.models.utils import use_quantize_kv_cache, restore_fp8_kv_cache from ipex_llm.transformers.models.utils import use_quantize_kv_cache, restore_fp8_kv_cache
from ipex_llm.transformers.models.utils import is_enough_kv_cache_room_4_36 from ipex_llm.transformers.models.utils import use_flash_attention, use_sdp, use_sdp_causal
from ipex_llm.transformers.models.utils import apply_rotary_pos_emb_cache_freq_xpu from ipex_llm.transformers.kv import DynamicFp8Cache, DynamicNormalCache
from ipex_llm.transformers.kv import DynamicFp8Cache
from ipex_llm.utils.common import invalidInputError from ipex_llm.utils.common import invalidInputError
from ipex_llm.transformers.models.utils import use_flash_attention, use_sdp
from transformers.models.qwen2.modeling_qwen2 import Qwen2Model, apply_rotary_pos_emb from transformers.models.qwen2.modeling_qwen2 import Qwen2Attention, apply_rotary_pos_emb, repeat_kv
from transformers.models.qwen2.modeling_qwen2 import _prepare_4d_causal_attention_mask_for_sdpa from transformers.models.qwen2.modeling_qwen2 import _prepare_4d_causal_attention_mask_for_sdpa
from transformers.models.qwen2.modeling_qwen2 import _prepare_4d_causal_attention_mask from transformers.models.qwen2.modeling_qwen2 import _prepare_4d_causal_attention_mask
from transformers.modeling_outputs import BaseModelOutputWithPast from transformers.modeling_outputs import BaseModelOutputWithPast
from ipex_llm.transformers.models.utils import use_decoding_fast_path from transformers.cache_utils import Cache, DynamicCache
try:
from transformers.cache_utils import Cache, DynamicCache
except ImportError:
Cache = Tuple[torch.Tensor]
import logging
from transformers import logging from transformers import logging
logger = logging.get_logger(__name__) logger = logging.get_logger(__name__)
import os
KV_CACHE_ALLOC_BLOCK_LENGTH = int(os.environ.get("KV_CACHE_ALLOC_BLOCK_LENGTH", 256))
def should_split_qkv_tensor(query_states, bsz, num_heads, q_len, kv_seq_len, output_attentions):
if not output_attentions:
if os.environ.get("IPEX_LLM_SPLIT_QKV", None) is not None:
return os.environ.get("IPEX_LLM_SPLIT_QKV", None) == "1"
elif query_states.dtype == torch.float16 and \
query_states.shape[2] >= 5000:
# split tensor for memory block limitation
# support fp16 and set input length threshold at 5000 for now
return True
elif query_states.element_size()*bsz*num_heads*q_len*kv_seq_len >= 4*1024**3:
# attn_weight size larger than memory block limitation 4GB
return True
return False
def should_use_fuse_rope(self, query_states, position_ids):
use_fuse_rope = query_states.device.type == "xpu"
use_fuse_rope = use_fuse_rope and not (self.training and query_states.requires_grad)
use_fuse_rope = use_fuse_rope and position_ids is not None
return use_fuse_rope
def qwen2_model_forward( def qwen2_model_forward(
self, self,
@ -109,9 +74,12 @@ def qwen2_model_forward(
return_dict: Optional[bool] = None, return_dict: Optional[bool] = None,
): ):
use_cache = use_cache if use_cache is not None else self.config.use_cache use_cache = use_cache if use_cache is not None else self.config.use_cache
if use_cache and use_quantize_kv_cache(self.layers[0].mlp.up_proj, input_ids): use_quantize_kv = use_quantize_kv_cache(self.layers[0].mlp.up_proj, input_ids)
if not isinstance(past_key_values, DynamicFp8Cache): if use_cache:
if use_quantize_kv and not isinstance(past_key_values, DynamicFp8Cache):
past_key_values = DynamicFp8Cache.from_legacy_cache(past_key_values) past_key_values = DynamicFp8Cache.from_legacy_cache(past_key_values)
if not use_quantize_kv and not isinstance(past_key_values, DynamicNormalCache):
past_key_values = DynamicNormalCache.from_legacy_cache(past_key_values)
return qwen2_model_forward_internal( return qwen2_model_forward_internal(
self=self, self=self,
input_ids=input_ids, input_ids=input_ids,
@ -248,13 +216,13 @@ def qwen2_model_forward_internal(
use_cache, use_cache,
) )
else: else:
# bigdl-llm changes # ipex-llm changes
curr_device = decoder_layer.input_layernorm.weight.device curr_device = decoder_layer.input_layernorm.weight.device
if attention_mask is not None: if attention_mask is not None:
attention_mask = attention_mask.to(curr_device) attention_mask = attention_mask.to(curr_device)
if position_ids is not None: if position_ids is not None:
position_ids = position_ids.to(curr_device) position_ids = position_ids.to(curr_device)
# bigdl-llm changes end # ipex-llm changes end
layer_outputs = decoder_layer( layer_outputs = decoder_layer(
hidden_states, hidden_states,
attention_mask=attention_mask, attention_mask=attention_mask,
@ -294,325 +262,111 @@ def qwen2_model_forward_internal(
) )
def merge_qkv(module: torch.nn.Module):
if isinstance(module, Qwen2Attention):
new_weight = torch.cat([
module.q_proj.weight.data,
module.k_proj.weight.data,
module.v_proj.weight.data,
], dim=0)
new_bias = torch.cat([
module.q_proj.bias.data,
module.k_proj.bias.data,
module.v_proj.bias.data,
], dim=-1)
qkv_proj = torch.nn.Linear(0, 0, bias=True)
qkv_proj.weight = torch.nn.Parameter(new_weight, requires_grad=False)
qkv_proj.bias = torch.nn.Parameter(new_bias, requires_grad=False)
qkv_proj.in_features = new_weight.size(1)
qkv_proj.out_features = new_weight.size(0)
module.qkv_proj = qkv_proj
del module.q_proj, module.k_proj, module.v_proj
def qwen2_attention_forward( def qwen2_attention_forward(
self, self,
hidden_states: torch.Tensor, hidden_states: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None, position_ids: Optional[torch.LongTensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None, past_key_value: Optional[Cache] = None,
output_attentions: bool = False, output_attentions: bool = False,
use_cache: bool = False, use_cache: bool = False,
**kwargs, **kwargs,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
if use_quantize_kv_cache(self.q_proj, hidden_states):
forward_function = qwen2_attention_forward_quantized
elif hidden_states.device.type == "cpu":
forward_function = qwen2_sdpa_attention_forward
else:
forward_function = qwen2_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,
**kwargs,
)
def qwen2_attention_forward_quantized(
self,
hidden_states: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
past_key_value: Optional[DynamicFp8Cache] = None,
output_attentions: bool = False,
use_cache: bool = False,
**kwargs,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
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.`"
)
use_fuse_rope = should_use_fuse_rope(self, hidden_states, position_ids)
bsz, q_len, _ = hidden_states.size() bsz, q_len, _ = hidden_states.size()
device = hidden_states.device
query_states = self.q_proj(hidden_states) qkv = self.qkv_proj(hidden_states)
key_states = self.k_proj(hidden_states) qkv = qkv.view(bsz, q_len, self.num_heads + 2 * self.num_key_value_heads, self.head_dim)
value_states = self.v_proj(hidden_states) qkv = qkv.transpose(1, 2)
query_states, key_states, value_states = qkv.split([self.num_heads,
query_states = query_states.view(bsz, q_len, self.num_key_value_heads,
self.num_heads, self.head_dim).transpose(1, 2) self.num_key_value_heads], dim=1)
key_states = key_states.view(bsz, q_len,
self.num_key_value_heads, self.head_dim).transpose(1, 2)
value_states = value_states.view(bsz, q_len,
self.num_key_value_heads, self.head_dim).transpose(1, 2)
kv_seq_len = key_states.shape[-2] kv_seq_len = key_states.shape[-2]
if past_key_value is not None: if past_key_value is not None:
invalidInputError(self.layer_idx is not None,
"The cache structure has changed since version v4.36. "
f"If you are using {self.__class__.__name__} "
"for auto-regressive decoding with k/v caching, "
"please make sure to initialize the attention class "
"with a layer index.")
kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx) kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
if use_fuse_rope: if should_use_fuse_rope(hidden_states, position_ids, self.training):
query_states, key_states = apply_rotary_pos_emb_cache_freq_xpu(query_states, key_states, import linear_q4_0
sin, cos, "qwen2", linear_q4_0.rotary_half_inplaced(self.rotary_emb.inv_freq, position_ids,
position_ids) query_states, key_states)
else: else:
cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
query_states, key_states = apply_rotary_pos_emb(query_states, key_states, query_states, key_states = apply_rotary_pos_emb(query_states, key_states,
cos, sin, position_ids) cos, sin, position_ids)
if past_key_value is not None: if past_key_value is not None:
cache_kwargs = {"sin": sin, "cos": cos} # Specific to RoPE models
key_states, value_states = past_key_value.update(key_states, value_states, key_states, value_states = past_key_value.update(key_states, value_states,
self.layer_idx, cache_kwargs) self.layer_idx, None)
if q_len == 1 and query_states.device.type == 'xpu' and not self.training \ attn_weights = None
and not hidden_states.requires_grad: if query_states.device.type == "cpu":
import linear_q4_0 attn_output = sdpa(query_states,
attn_output = linear_q4_0.sdp_fp8(query_states, key_states, value_states, key_states,
attention_mask) value_states,
attn_weights = None attn_mask=attention_mask,
else: dropout_p=self.attention_dropout if self.training else 0.0,
key, value = restore_fp8_kv_cache(key_states, value_states, query_states.dtype) is_causal=self.is_causal and attention_mask is None and q_len > 1)
key = repeat_kv(key, self.num_key_value_groups)
value = repeat_kv(value, self.num_key_value_groups)
if should_split_qkv_tensor(query_states, bsz, self.num_heads,
q_len, kv_seq_len, output_attentions):
attn_output, attn_weights = native_sdp_split_qkv_tensor(query_states, key,
value, attention_mask,
bsz, q_len, kv_seq_len,
self.head_dim, self.num_heads,
self.attention_dropout,
self.training)
else:
attn_weights = torch.matmul(query_states, key.transpose(2, 3))
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)},"
"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 "
f"{(bsz, 1, q_len, kv_seq_len)},"
f" but is {attention_mask.size()}"))
attn_weights = attn_weights + attention_mask
if kv_seq_len >= 2048 or bsz >= 64:
# for memory considerations, do not upcast attention to fp32
# for long sequences or large batches
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
else:
# upcast attention to fp32
attn_weights = nn.functional.softmax(attn_weights, dim=-1,
dtype=torch.float32).to(query_states.dtype)
attn_weights = nn.functional.dropout(attn_weights, p=self.attention_dropout,
training=self.training)
attn_output = torch.matmul(attn_weights, value)
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).contiguous()
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
from ipex_llm.ggml.quantize import ggml_tensor_qtype
SYM_INT4 = ggml_tensor_qtype["sym_int4"]
FP8E5 = ggml_tensor_qtype["fp8_e5m2"]
def qwen2_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,
**kwargs,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
use_fuse_rope = should_use_fuse_rope(self, hidden_states, position_ids)
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.`"
)
bsz, q_len, _ = hidden_states.size()
device = hidden_states.device
enough_kv_room = is_enough_kv_cache_room_4_36(past_key_value, self.layer_idx)
decoding_fast_path = use_decoding_fast_path(self.q_proj,
use_fuse_rope,
enough_kv_room,
bsz * q_len)
if decoding_fast_path:
hidden_states = hidden_states.view(1, -1)
cache_k = past_key_value.key_cache[self.layer_idx]
cache_v = past_key_value.value_cache[self.layer_idx]
kv_seq_len = cache_k.shape[-2]
import linear_q4_0
args = [hidden_states, self.q_proj.weight, self.k_proj.weight, self.v_proj.weight,
self.q_proj.bias, self.k_proj.bias, self.v_proj.bias, position_ids, cache_k,
cache_v, self.q_proj.weight.qtype, self.v_proj.weight.qtype, kv_seq_len,
self.head_dim, self.rotary_emb.base]
query_states, key_states, value_states = linear_q4_0.forward_qkv_bias(*args)
kv_seq_len += 1
if self.layer_idx == 0:
past_key_value.seen_tokens = kv_seq_len
past_key_value.key_cache[self.layer_idx] = key_states
past_key_value.value_cache[self.layer_idx] = value_states
else:
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)
value_states = \
value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
kv_seq_len = key_states.shape[-2]
if past_key_value is not None:
if self.layer_idx is None:
invalidInputError(
False,
"The cache structure has changed since version v4.36. "
f"If you are using {self.__class__.__name__} "
"for auto-regressive decoding with k/v caching, "
"please make sure to initialize the attention class with a layer index."
)
kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
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, "qwen2",
position_ids)
else:
query_states, key_states = apply_rotary_pos_emb(query_states, key_states,
cos, sin, position_ids)
if past_key_value is not None:
# update the number of seen tokens
if self.layer_idx == 0:
past_key_value.seen_tokens += key_states.shape[-2]
if len(past_key_value.key_cache) <= self.layer_idx:
past_key_value.key_cache.append(key_states)
past_key_value.value_cache.append(value_states)
else:
cache_k = past_key_value.key_cache[self.layer_idx]
cache_v = past_key_value.value_cache[self.layer_idx]
if not enough_kv_room:
# allocate new
new_c_k, new_c_v = extend_kv_cache(bsz,
self.num_key_value_heads, # Support GQA
self.head_dim,
cache_k.size(2),
kv_seq_len + KV_CACHE_ALLOC_BLOCK_LENGTH,
dtype=cache_k.dtype,
device=device)
new_c_k[:] = cache_k
new_c_v[:] = cache_v
cache_k = new_c_k
cache_v = new_c_v
key_states, value_states = append_kv_cache(cache_k,
cache_v,
key_states,
value_states)
# update past_key_value
past_key_value.key_cache[self.layer_idx] = key_states
past_key_value.value_cache[self.layer_idx] = value_states
# repeat k/v heads if n_kv_heads < n_heads
key_states = repeat_kv(key_states, self.num_key_value_groups)
value_states = repeat_kv(value_states, self.num_key_value_groups)
if not self.training and not hidden_states.requires_grad and \
use_flash_attention(query_states, key_states, attention_mask):
attn_output = F.scaled_dot_product_attention(query_states.to(device, dtype=torch.float16),
key_states.to(device, dtype=torch.float16),
value_states.to(device, dtype=torch.float16),
is_causal=True)
attn_weights = None
elif not self.training and not hidden_states.requires_grad and \ elif not self.training and not hidden_states.requires_grad and \
use_sdp(q_len, key_states.shape[2], self.head_dim, query_states): use_flash_attention(query_states, key_states, attention_mask):
attn_output = sdpa(query_states.to(device, dtype=torch.float16),
key_states.to(device, dtype=torch.float16),
value_states.to(device, dtype=torch.float16),
is_causal=True).to(hidden_states.dtype)
elif use_sdp(q_len, kv_seq_len, self.head_dim, query_states):
import linear_q4_0 import linear_q4_0
attn_output = linear_q4_0.sdp(query_states, key_states, value_states, attention_mask) if isinstance(past_key_value, DynamicFp8Cache):
attn_output = attn_output.view(query_states.shape) attn_output = linear_q4_0.sdp_fp8(query_states, key_states, value_states,
attn_weights = None attention_mask)
else:
if should_split_qkv_tensor(query_states, bsz, self.num_heads,
q_len, kv_seq_len, output_attentions):
attn_output, attn_weights = native_sdp_split_qkv_tensor(query_states, key_states,
value_states, attention_mask,
bsz, q_len, kv_seq_len,
self.head_dim, self.num_heads,
self.attention_dropout,
self.training)
else: else:
attn_weights = torch.matmul(query_states, attn_output = linear_q4_0.sdp(query_states, key_states, value_states, attention_mask)
key_states.transpose(2, 3)) / math.sqrt(self.head_dim) elif use_sdp_causal(q_len, kv_seq_len, self.head_dim, query_states, self.training):
import linear_q4_0
if isinstance(past_key_value, DynamicFp8Cache):
attn_output = linear_q4_0.sdp_fp8_causal(query_states, key_states, value_states)
else:
attn_output = linear_q4_0.sdp_causal(query_states, key_states, value_states)
else:
if isinstance(past_key_value, DynamicFp8Cache):
key_states, value_states = restore_fp8_kv_cache(key_states, value_states,
query_states.dtype)
# repeat k/v heads if n_kv_heads < n_heads
key_states = repeat_kv(key_states, self.num_key_value_groups)
value_states = repeat_kv(value_states, self.num_key_value_groups)
invalidInputError(attn_weights.size() == (bsz, self.num_heads, q_len, kv_seq_len), attn_weights = torch.matmul(query_states,
("Attention weights should be of size " key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
f"{(bsz, self.num_heads, q_len, kv_seq_len)}," if attention_mask is not None:
"but is {attn_weights.size()}")) attn_weights = attn_weights + attention_mask
# upcast attention to fp32
if attention_mask is not None: attn_weights = torch.nn.functional.softmax(attn_weights, dim=-1,
invalidInputError(attention_mask.size() == (bsz, 1, q_len, kv_seq_len), dtype=torch.float32).to(query_states.dtype)
(f"Attention mask should be of size " attn_weights = torch.nn.functional.dropout(attn_weights, p=self.attention_dropout,
f"{(bsz, 1, q_len, kv_seq_len)}," training=self.training)
f" but is {attention_mask.size()}")) attn_output = torch.matmul(attn_weights, value_states)
attn_weights = attn_weights + attention_mask
if kv_seq_len >= 2048 or bsz >= 64:
# for memory considerations, do not upcast attention to fp32
# for long sequences or large batches
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
else:
# upcast attention to fp32
attn_weights = nn.functional.softmax(attn_weights, dim=-1,
dtype=torch.float32).to(query_states.dtype)
attn_weights = nn.functional.dropout(attn_weights,
p=self.attention_dropout,
training=self.training)
attn_output = torch.matmul(attn_weights, value_states)
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).contiguous() attn_output = attn_output.transpose(1, 2).contiguous()
attn_output = attn_output.reshape(bsz, q_len, self.hidden_size) attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
@ -621,183 +375,4 @@ def qwen2_attention_forward_origin(
if not output_attentions: if not output_attentions:
attn_weights = None attn_weights = None
return attn_output, attn_weights, past_key_value
return attn_output.to(hidden_states.dtype), attn_weights, past_key_value
def qwen2_sdpa_attention_forward(
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,
**kwargs,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
use_fuse_rope = should_use_fuse_rope(self, hidden_states, position_ids)
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.`"
)
bsz, q_len, _ = hidden_states.size()
device = hidden_states.device
enough_kv_room = is_enough_kv_cache_room_4_36(past_key_value, self.layer_idx)
decoding_fast_path = use_decoding_fast_path(self.q_proj,
use_fuse_rope,
enough_kv_room,
bsz * q_len)
if decoding_fast_path:
hidden_states = hidden_states.view(1, -1)
cache_k = past_key_value.key_cache[self.layer_idx]
cache_v = past_key_value.value_cache[self.layer_idx]
kv_seq_len = cache_k.shape[-2]
import linear_q4_0
args = [hidden_states, self.q_proj.weight, self.k_proj.weight, self.v_proj.weight,
self.q_proj.bias, self.k_proj.bias, self.v_proj.bias, position_ids, cache_k,
cache_v, self.q_proj.weight.qtype, self.v_proj.weight.qtype, kv_seq_len,
self.head_dim, self.rotary_emb.base]
query_states, key_states, value_states = linear_q4_0.forward_qkv_bias(*args)
kv_seq_len += 1
if self.layer_idx == 0:
past_key_value.seen_tokens = kv_seq_len
past_key_value.key_cache[self.layer_idx] = key_states
past_key_value.value_cache[self.layer_idx] = value_states
else:
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)
value_states = \
value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
kv_seq_len = key_states.shape[-2]
if past_key_value is not None:
if self.layer_idx is None:
invalidInputError(
False,
"The cache structure has changed since version v4.36. "
f"If you are using {self.__class__.__name__} "
"for auto-regressive decoding with k/v caching, "
"please make sure to initialize the attention class with a layer index."
)
kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
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, "qwen2",
position_ids)
else:
query_states, key_states = apply_rotary_pos_emb(query_states, key_states,
cos, sin, position_ids)
if past_key_value is not None:
# update the number of seen tokens
if self.layer_idx == 0:
past_key_value.seen_tokens += key_states.shape[-2]
if len(past_key_value.key_cache) <= self.layer_idx:
past_key_value.key_cache.append(key_states)
past_key_value.value_cache.append(value_states)
else:
cache_k = past_key_value.key_cache[self.layer_idx]
cache_v = past_key_value.value_cache[self.layer_idx]
if not enough_kv_room:
# allocate new
new_c_k, new_c_v = extend_kv_cache(bsz,
self.num_key_value_heads, # Support GQA
self.head_dim,
cache_k.size(2),
kv_seq_len + KV_CACHE_ALLOC_BLOCK_LENGTH,
dtype=cache_k.dtype,
device=device)
new_c_k[:] = cache_k
new_c_v[:] = cache_v
cache_k = new_c_k
cache_v = new_c_v
key_states, value_states = append_kv_cache(cache_k,
cache_v,
key_states,
value_states)
# update past_key_value
past_key_value.key_cache[self.layer_idx] = key_states
past_key_value.value_cache[self.layer_idx] = value_states
# repeat k/v heads if n_kv_heads < n_heads
key_states = repeat_kv(key_states, self.num_key_value_groups)
value_states = repeat_kv(value_states, self.num_key_value_groups)
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)},"
"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
from torch.nn.functional import scaled_dot_product_attention as sdpa
attn_output = sdpa(query_states,
key_states,
value_states,
attn_mask=attention_mask,
dropout_p=self.attention_dropout if self.training else 0.0,
is_causal=self.is_causal and attention_mask is None and q_len > 1)
attn_output = attn_output.transpose(1, 2).contiguous()
attn_output = attn_output.view(bsz, q_len, self.hidden_size)
attn_output = self.o_proj(attn_output)
return attn_output, None, past_key_value
def native_sdp_split_qkv_tensor(query, key, value, attention_mask,
bsz, q_len, kv_seq_len, head_dim, num_heads,
attention_dropout, training):
block_size = 8
query_split = torch.split(query, block_size, dim=1)
key_split = torch.split(key.transpose(2, 3), block_size, dim=1)
value_split = torch.split(value, block_size, dim=1)
attn_outputs = []
for q, k, v in zip(query_split, key_split, value_split):
attn_weights_split = torch.matmul(q, k) / math.sqrt(head_dim)
block_actual_size = attn_weights_split.size(1)
attn_weights_split_size = (bsz, block_actual_size, q_len, kv_seq_len)
if attn_weights_split.size() != attn_weights_split_size:
invalidInputError(False,
f"Splitted attention weights should be of size "
f"{attn_weights_split_size}, but is {attn_weights_split.size()}")
if attention_mask is not None:
attn_mask_size = (bsz, 1, q_len, kv_seq_len)
if attention_mask.size() != attn_mask_size:
invalidInputError(False,
f"Attention mask should be of size {attn_mask_size}, "
f"but is {attention_mask.size()}")
attn_weights_split = attn_weights_split + attention_mask
attn_weights_split = nn.functional.softmax(attn_weights_split, dim=-1)
attn_weights_split = nn.functional.dropout(attn_weights_split,
p=attention_dropout,
training=training)
attn_outputs.append(torch.matmul(attn_weights_split, v))
attn_output = torch.cat(attn_outputs, dim=1)
return attn_output, None