duzx16
commited on
Commit
•
a7eaddd
1
Parent(s):
835c717
Add support for flash attention 2
Browse files- modeling_chatglm.py +117 -4
modeling_chatglm.py
CHANGED
@@ -21,12 +21,17 @@ from transformers.modeling_outputs import (
|
|
21 |
SequenceClassifierOutputWithPast,
|
22 |
)
|
23 |
from transformers.modeling_utils import PreTrainedModel
|
24 |
-
from transformers.utils import logging, is_torch_npu_available
|
|
|
25 |
from transformers.generation.logits_process import LogitsProcessor
|
26 |
from transformers.generation.utils import LogitsProcessorList, StoppingCriteriaList, GenerationConfig, ModelOutput
|
27 |
|
28 |
from .configuration_chatglm import ChatGLMConfig
|
29 |
|
|
|
|
|
|
|
|
|
30 |
# flags required to enable jit fusion kernels
|
31 |
|
32 |
if sys.platform != 'darwin' and not is_torch_npu_available():
|
@@ -160,12 +165,13 @@ class RMSNorm(torch.nn.Module):
|
|
160 |
class CoreAttention(torch.nn.Module):
|
161 |
def __init__(self, config: ChatGLMConfig, layer_number):
|
162 |
super(CoreAttention, self).__init__()
|
163 |
-
|
164 |
self.apply_query_key_layer_scaling = config.apply_query_key_layer_scaling
|
165 |
self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
|
166 |
if self.apply_query_key_layer_scaling:
|
167 |
self.attention_softmax_in_fp32 = True
|
168 |
self.layer_number = max(1, layer_number)
|
|
|
169 |
|
170 |
projection_size = config.kv_channels * config.num_attention_heads
|
171 |
|
@@ -259,21 +265,122 @@ class SdpaAttention(CoreAttention):
|
|
259 |
def forward(self, query_layer, key_layer, value_layer, attention_mask):
|
260 |
if attention_mask is None and query_layer.shape[2] == key_layer.shape[2]:
|
261 |
context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
|
262 |
-
is_causal=True
|
|
|
263 |
else:
|
264 |
if attention_mask is not None:
|
265 |
attention_mask = ~attention_mask
|
266 |
context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
|
267 |
-
attention_mask
|
|
|
268 |
context_layer = context_layer.transpose(1, 2).contiguous()
|
269 |
new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
|
270 |
context_layer = context_layer.reshape(*new_context_layer_shape)
|
271 |
return context_layer
|
272 |
|
273 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
274 |
CORE_ATTENTION_CLASSES = {
|
275 |
"eager": CoreAttention,
|
276 |
"sdpa": SdpaAttention,
|
|
|
277 |
}
|
278 |
|
279 |
|
@@ -652,12 +759,18 @@ class ChatGLMPreTrainedModel(PreTrainedModel):
|
|
652 |
config_class = ChatGLMConfig
|
653 |
base_model_prefix = "transformer"
|
654 |
_no_split_modules = ["GLMBlock"]
|
|
|
|
|
655 |
|
656 |
def _init_weights(self, module: nn.Module):
|
657 |
"""Initialize the weights."""
|
658 |
return
|
659 |
|
660 |
def get_masks(self, input_ids, past_key_values, padding_mask=None):
|
|
|
|
|
|
|
|
|
661 |
batch_size, seq_length = input_ids.shape
|
662 |
full_attention_mask = torch.ones(batch_size, seq_length, seq_length, device=input_ids.device)
|
663 |
full_attention_mask.tril_()
|
|
|
21 |
SequenceClassifierOutputWithPast,
|
22 |
)
|
23 |
from transformers.modeling_utils import PreTrainedModel
|
24 |
+
from transformers.utils import logging, is_torch_npu_available, is_flash_attn_greater_or_equal_2_10, \
|
25 |
+
is_flash_attn_2_available
|
26 |
from transformers.generation.logits_process import LogitsProcessor
|
27 |
from transformers.generation.utils import LogitsProcessorList, StoppingCriteriaList, GenerationConfig, ModelOutput
|
28 |
|
29 |
from .configuration_chatglm import ChatGLMConfig
|
30 |
|
31 |
+
if is_flash_attn_2_available():
|
32 |
+
from flash_attn import flash_attn_func, flash_attn_varlen_func
|
33 |
+
from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input # noqa
|
34 |
+
|
35 |
# flags required to enable jit fusion kernels
|
36 |
|
37 |
if sys.platform != 'darwin' and not is_torch_npu_available():
|
|
|
165 |
class CoreAttention(torch.nn.Module):
|
166 |
def __init__(self, config: ChatGLMConfig, layer_number):
|
167 |
super(CoreAttention, self).__init__()
|
168 |
+
self.config = config
|
169 |
self.apply_query_key_layer_scaling = config.apply_query_key_layer_scaling
|
170 |
self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
|
171 |
if self.apply_query_key_layer_scaling:
|
172 |
self.attention_softmax_in_fp32 = True
|
173 |
self.layer_number = max(1, layer_number)
|
174 |
+
self.is_causal = True
|
175 |
|
176 |
projection_size = config.kv_channels * config.num_attention_heads
|
177 |
|
|
|
265 |
def forward(self, query_layer, key_layer, value_layer, attention_mask):
|
266 |
if attention_mask is None and query_layer.shape[2] == key_layer.shape[2]:
|
267 |
context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
|
268 |
+
is_causal=True,
|
269 |
+
dropout_p=self.config.attention_dropout if self.training else 0.0)
|
270 |
else:
|
271 |
if attention_mask is not None:
|
272 |
attention_mask = ~attention_mask
|
273 |
context_layer = torch.nn.functional.scaled_dot_product_attention(query_layer, key_layer, value_layer,
|
274 |
+
attention_mask,
|
275 |
+
dropout_p=self.config.attention_dropout if self.training else 0.0)
|
276 |
context_layer = context_layer.transpose(1, 2).contiguous()
|
277 |
new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size_per_partition,)
|
278 |
context_layer = context_layer.reshape(*new_context_layer_shape)
|
279 |
return context_layer
|
280 |
|
281 |
|
282 |
+
def _get_unpad_data(attention_mask):
|
283 |
+
seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)
|
284 |
+
indices = torch.nonzero(attention_mask.flatten(), as_tuple=False).flatten()
|
285 |
+
max_seqlen_in_batch = seqlens_in_batch.max().item()
|
286 |
+
cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.int32), (1, 0))
|
287 |
+
return (
|
288 |
+
indices,
|
289 |
+
cu_seqlens,
|
290 |
+
max_seqlen_in_batch,
|
291 |
+
)
|
292 |
+
|
293 |
+
|
294 |
+
# Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2
|
295 |
+
class FlashAttention2(CoreAttention):
|
296 |
+
def __init__(self, *args, **kwargs):
|
297 |
+
super().__init__(*args, **kwargs)
|
298 |
+
self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
|
299 |
+
|
300 |
+
def forward(self, query_states, key_states, value_states, attention_mask):
|
301 |
+
query_states = query_states.transpose(1, 2)
|
302 |
+
key_states = key_states.transpose(1, 2)
|
303 |
+
value_states = value_states.transpose(1, 2)
|
304 |
+
batch_size, query_length = query_states.shape[:2]
|
305 |
+
if not self._flash_attn_uses_top_left_mask:
|
306 |
+
causal = self.is_causal
|
307 |
+
else:
|
308 |
+
# TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1. For details, please see the comment in LlamaFlashAttention2 __init__.
|
309 |
+
causal = self.is_causal and query_length != 1
|
310 |
+
dropout = self.config.attention_dropout if self.training else 0.0
|
311 |
+
# Contains at least one padding token in the sequence
|
312 |
+
if attention_mask is not None:
|
313 |
+
query_states, key_states, value_states, indices_q, cu_seq_lens, max_seq_lens = self._upad_input(
|
314 |
+
query_states, key_states, value_states, attention_mask, query_length
|
315 |
+
)
|
316 |
+
|
317 |
+
cu_seqlens_q, cu_seqlens_k = cu_seq_lens
|
318 |
+
max_seqlen_in_batch_q, max_seqlen_in_batch_k = max_seq_lens
|
319 |
+
|
320 |
+
attn_output_unpad = flash_attn_varlen_func(
|
321 |
+
query_states,
|
322 |
+
key_states,
|
323 |
+
value_states,
|
324 |
+
cu_seqlens_q=cu_seqlens_q,
|
325 |
+
cu_seqlens_k=cu_seqlens_k,
|
326 |
+
max_seqlen_q=max_seqlen_in_batch_q,
|
327 |
+
max_seqlen_k=max_seqlen_in_batch_k,
|
328 |
+
dropout_p=dropout,
|
329 |
+
softmax_scale=None,
|
330 |
+
causal=causal,
|
331 |
+
)
|
332 |
+
|
333 |
+
attn_output = pad_input(attn_output_unpad, indices_q, batch_size, query_length)
|
334 |
+
else:
|
335 |
+
attn_output = flash_attn_func(
|
336 |
+
query_states, key_states, value_states, dropout, softmax_scale=None, causal=causal
|
337 |
+
)
|
338 |
+
attn_output = attn_output.reshape(batch_size, query_length, self.hidden_size_per_partition).contiguous()
|
339 |
+
return attn_output
|
340 |
+
|
341 |
+
def _upad_input(self, query_layer, key_layer, value_layer, attention_mask, query_length):
|
342 |
+
indices_k, cu_seqlens_k, max_seqlen_in_batch_k = _get_unpad_data(attention_mask)
|
343 |
+
batch_size, kv_seq_len, num_key_value_heads, head_dim = key_layer.shape
|
344 |
+
|
345 |
+
key_layer = index_first_axis(
|
346 |
+
key_layer.reshape(batch_size * kv_seq_len, num_key_value_heads, head_dim), indices_k
|
347 |
+
)
|
348 |
+
value_layer = index_first_axis(
|
349 |
+
value_layer.reshape(batch_size * kv_seq_len, num_key_value_heads, head_dim), indices_k
|
350 |
+
)
|
351 |
+
if query_length == kv_seq_len:
|
352 |
+
query_layer = index_first_axis(
|
353 |
+
query_layer.reshape(batch_size * kv_seq_len, self.num_attention_heads_per_partition, head_dim), indices_k
|
354 |
+
)
|
355 |
+
cu_seqlens_q = cu_seqlens_k
|
356 |
+
max_seqlen_in_batch_q = max_seqlen_in_batch_k
|
357 |
+
indices_q = indices_k
|
358 |
+
elif query_length == 1:
|
359 |
+
max_seqlen_in_batch_q = 1
|
360 |
+
cu_seqlens_q = torch.arange(
|
361 |
+
batch_size + 1, dtype=torch.int32, device=query_layer.device
|
362 |
+
) # There is a memcpy here, that is very bad.
|
363 |
+
indices_q = cu_seqlens_q[:-1]
|
364 |
+
query_layer = query_layer.squeeze(1)
|
365 |
+
else:
|
366 |
+
# The -q_len: slice assumes left padding.
|
367 |
+
attention_mask = attention_mask[:, -query_length:]
|
368 |
+
query_layer, indices_q, cu_seqlens_q, max_seqlen_in_batch_q = unpad_input(query_layer, attention_mask)
|
369 |
+
|
370 |
+
return (
|
371 |
+
query_layer,
|
372 |
+
key_layer,
|
373 |
+
value_layer,
|
374 |
+
indices_q,
|
375 |
+
(cu_seqlens_q, cu_seqlens_k),
|
376 |
+
(max_seqlen_in_batch_q, max_seqlen_in_batch_k),
|
377 |
+
)
|
378 |
+
|
379 |
+
|
380 |
CORE_ATTENTION_CLASSES = {
|
381 |
"eager": CoreAttention,
|
382 |
"sdpa": SdpaAttention,
|
383 |
+
"flash_attention_2": FlashAttention2
|
384 |
}
|
385 |
|
386 |
|
|
|
759 |
config_class = ChatGLMConfig
|
760 |
base_model_prefix = "transformer"
|
761 |
_no_split_modules = ["GLMBlock"]
|
762 |
+
_supports_flash_attn_2 = True
|
763 |
+
_supports_sdpa = True
|
764 |
|
765 |
def _init_weights(self, module: nn.Module):
|
766 |
"""Initialize the weights."""
|
767 |
return
|
768 |
|
769 |
def get_masks(self, input_ids, past_key_values, padding_mask=None):
|
770 |
+
if self.config._attn_implementation == "flash_attention_2":
|
771 |
+
if padding_mask is not None and not padding_mask.all():
|
772 |
+
return padding_mask
|
773 |
+
return None
|
774 |
batch_size, seq_length = input_ids.shape
|
775 |
full_attention_mask = torch.ones(batch_size, seq_length, seq_length, device=input_ids.device)
|
776 |
full_attention_mask.tril_()
|