mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-02-14 06:53:50 +08:00
421 lines
16 KiB
Python
421 lines
16 KiB
Python
# SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
#
|
|
# 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.
|
|
from typing import Optional, Union
|
|
|
|
import tensorrt as trt
|
|
|
|
from ..._common import default_net
|
|
from ..._utils import pad_vocab_size, str_dtype_to_trt
|
|
from ...functional import Tensor, gather_last_token_logits
|
|
from ...layers import (MLP, Attention, AttentionMaskType, ColumnLinear,
|
|
Embedding, LayerNorm, PositionEmbeddingType)
|
|
from ...mapping import Mapping
|
|
from ...module import Module, ModuleList
|
|
from ...quantization import QuantMode
|
|
from ..generation_mixin import GenerationMixin
|
|
|
|
|
|
class FalconDecoderLayer(Module):
|
|
|
|
def __init__(
|
|
self,
|
|
hidden_size,
|
|
num_attention_heads,
|
|
max_position_embeddings,
|
|
num_attention_kv_heads=None,
|
|
dtype=None,
|
|
hidden_act='gelu',
|
|
quant_mode=QuantMode(0),
|
|
mlp_hidden_size=None,
|
|
bias=True,
|
|
use_alibi=True,
|
|
new_decoder_architecture=False,
|
|
parallel_attention=False,
|
|
layernorm_epsilon=1e-5,
|
|
tp_group=None,
|
|
tp_size=1,
|
|
layer_id=None,
|
|
):
|
|
super().__init__()
|
|
self._layer_id = layer_id # useful for debugging
|
|
self.hidden_size = hidden_size
|
|
self.num_attention_heads = num_attention_heads
|
|
self.num_attention_kv_heads = num_attention_kv_heads
|
|
self.max_position_embeddings = max_position_embeddings
|
|
self.dtype = dtype
|
|
self.hidden_act = hidden_act
|
|
self.tp_group = tp_group
|
|
self.tp_size = tp_size
|
|
self.input_layernorm = LayerNorm(normalized_shape=hidden_size,
|
|
eps=layernorm_epsilon,
|
|
dtype=dtype)
|
|
if use_alibi:
|
|
position_embedding_type = PositionEmbeddingType.alibi
|
|
else:
|
|
position_embedding_type = PositionEmbeddingType.rope_gpt_neox
|
|
|
|
self.attention = Attention(
|
|
hidden_size=hidden_size,
|
|
num_attention_heads=num_attention_heads,
|
|
num_kv_heads=num_attention_kv_heads,
|
|
max_position_embeddings=max_position_embeddings,
|
|
dtype=dtype,
|
|
attention_mask_type=AttentionMaskType.causal,
|
|
bias=bias,
|
|
position_embedding_type=position_embedding_type,
|
|
tp_group=tp_group,
|
|
tp_size=tp_size,
|
|
use_int8_kv_cache=quant_mode.has_int8_kv_cache(),
|
|
scale_alibi_bias=True,
|
|
)
|
|
|
|
if mlp_hidden_size is None:
|
|
mlp_hidden_size = hidden_size * 4
|
|
|
|
self.new_decoder_architecture = new_decoder_architecture
|
|
self.parallel_attn = parallel_attention
|
|
|
|
if self.new_decoder_architecture:
|
|
# Layernorm before MLP.
|
|
self.mlp_layernorm = LayerNorm(normalized_shape=hidden_size,
|
|
eps=layernorm_epsilon,
|
|
dtype=dtype)
|
|
else:
|
|
self.mlp_layernorm = None
|
|
self.mlp = MLP(
|
|
hidden_size=hidden_size,
|
|
ffn_hidden_size=mlp_hidden_size,
|
|
hidden_act=hidden_act,
|
|
dtype=dtype,
|
|
bias=bias,
|
|
tp_group=tp_group,
|
|
tp_size=tp_size,
|
|
)
|
|
if self.new_decoder_architecture or self.parallel_attn:
|
|
self.post_layernorm = None
|
|
else:
|
|
self.post_layernorm = LayerNorm(normalized_shape=hidden_size,
|
|
dtype=dtype)
|
|
|
|
def forward(self,
|
|
hidden_states: Tensor,
|
|
attention_mask=None,
|
|
past_key_value=None,
|
|
sequence_length=None,
|
|
host_past_key_value_lengths=None,
|
|
use_cache=False,
|
|
cache_indirection=None,
|
|
context_lengths=None,
|
|
host_context_lengths=None,
|
|
host_request_types=None,
|
|
max_context_length=None):
|
|
assert isinstance(hidden_states, Tensor)
|
|
|
|
residual = hidden_states
|
|
|
|
if self.new_decoder_architecture:
|
|
mlp_ln_output = self.mlp_layernorm(hidden_states)
|
|
hidden_states = self.input_layernorm(hidden_states)
|
|
input_ln_output = hidden_states
|
|
attention_output = self.attention(
|
|
hidden_states,
|
|
attention_mask=attention_mask,
|
|
past_key_value=past_key_value,
|
|
sequence_length=sequence_length,
|
|
host_past_key_value_lengths=host_past_key_value_lengths,
|
|
use_cache=use_cache,
|
|
cache_indirection=cache_indirection,
|
|
context_lengths=context_lengths,
|
|
host_context_lengths=host_context_lengths,
|
|
host_request_types=host_request_types,
|
|
max_context_length=max_context_length)
|
|
|
|
if use_cache:
|
|
attention_output, presents = attention_output
|
|
|
|
if not self.new_decoder_architecture:
|
|
if self.parallel_attn:
|
|
hidden_states = input_ln_output
|
|
else:
|
|
hidden_states = residual + attention_output
|
|
residual = hidden_states
|
|
hidden_states = self.post_layernorm(hidden_states)
|
|
else:
|
|
hidden_states = mlp_ln_output
|
|
|
|
hidden_states = self.mlp(hidden_states)
|
|
|
|
if self.new_decoder_architecture or self.parallel_attn:
|
|
hidden_states = hidden_states + attention_output
|
|
|
|
hidden_states = residual + hidden_states
|
|
if use_cache:
|
|
return hidden_states, presents
|
|
return hidden_states
|
|
|
|
|
|
class FalconModel(Module):
|
|
|
|
def __init__(
|
|
self,
|
|
num_layers: int,
|
|
num_heads: int,
|
|
hidden_size: int,
|
|
vocab_size: int,
|
|
hidden_act: int,
|
|
max_position_embeddings: int,
|
|
dtype: Optional[Union[str, trt.DataType]] = None,
|
|
mapping: Mapping = Mapping(),
|
|
num_kv_heads: Optional[int] = None,
|
|
mlp_hidden_size: Optional[int] = None,
|
|
bias: bool = True,
|
|
quant_mode: QuantMode = QuantMode(0),
|
|
use_alibi: bool = True,
|
|
parallel_attention: bool = False,
|
|
new_decoder_architecture: bool = False,
|
|
):
|
|
super().__init__()
|
|
self.num_layers = num_layers
|
|
self.num_heads = num_heads
|
|
self.num_kv_heads = num_kv_heads or num_heads
|
|
self.hidden_size = hidden_size
|
|
self.vocab_size = vocab_size
|
|
|
|
# Falcon variants
|
|
self.parallel_attention = parallel_attention
|
|
self.new_decoder_architecture = new_decoder_architecture
|
|
|
|
assert isinstance(dtype, (str, trt.DataType))
|
|
if isinstance(dtype, str):
|
|
self.dtype = str_dtype_to_trt(dtype)
|
|
else:
|
|
self.dtype = dtype
|
|
if quant_mode.has_int8_kv_cache():
|
|
self.kv_dtype = str_dtype_to_trt('int8')
|
|
else:
|
|
self.kv_dtype = self.dtype
|
|
|
|
self.embedding = Embedding(vocab_size, hidden_size, dtype=dtype)
|
|
|
|
self.layers = ModuleList([
|
|
FalconDecoderLayer(
|
|
hidden_size=hidden_size,
|
|
num_attention_heads=num_heads,
|
|
max_position_embeddings=max_position_embeddings,
|
|
dtype=dtype,
|
|
bias=bias,
|
|
quant_mode=quant_mode,
|
|
hidden_act=hidden_act,
|
|
num_attention_kv_heads=self.num_kv_heads,
|
|
mlp_hidden_size=mlp_hidden_size,
|
|
use_alibi=use_alibi,
|
|
parallel_attention=parallel_attention,
|
|
new_decoder_architecture=new_decoder_architecture,
|
|
tp_group=mapping.tp_group,
|
|
tp_size=mapping.tp_size,
|
|
layer_id=i,
|
|
) for i in range(num_layers)
|
|
])
|
|
|
|
self.ln_f = LayerNorm(normalized_shape=hidden_size, dtype=dtype)
|
|
|
|
def forward(self,
|
|
input_ids: Tensor,
|
|
position_ids=None,
|
|
past_key_value=None,
|
|
sequence_length=None,
|
|
host_past_key_value_lengths=None,
|
|
use_cache=False,
|
|
attention_mask=None,
|
|
cache_indirection=None,
|
|
context_lengths=None,
|
|
host_context_lengths=None,
|
|
host_request_types=None,
|
|
max_context_length: int = None):
|
|
hidden_states = self.embedding(input_ids)
|
|
|
|
if past_key_value is None:
|
|
past_key_value = tuple([None] * len(self.layers))
|
|
|
|
if use_cache:
|
|
presents = []
|
|
|
|
for layer, past in zip(self.layers, past_key_value):
|
|
hidden_states = layer(
|
|
hidden_states,
|
|
past_key_value=past,
|
|
sequence_length=sequence_length,
|
|
host_past_key_value_lengths=host_past_key_value_lengths,
|
|
use_cache=use_cache,
|
|
attention_mask=attention_mask,
|
|
cache_indirection=cache_indirection,
|
|
context_lengths=context_lengths,
|
|
host_context_lengths=host_context_lengths,
|
|
host_request_types=host_request_types,
|
|
max_context_length=max_context_length)
|
|
|
|
if use_cache:
|
|
presents.append(hidden_states[1])
|
|
hidden_states = hidden_states[0]
|
|
|
|
hidden_states = self.ln_f(hidden_states)
|
|
|
|
if use_cache:
|
|
return (hidden_states, tuple(presents))
|
|
return hidden_states
|
|
|
|
|
|
class FalconForCausalLM(FalconModel, GenerationMixin):
|
|
|
|
def __init__(self,
|
|
num_layers: int,
|
|
num_heads: int,
|
|
hidden_size: int,
|
|
vocab_size: int,
|
|
max_position_embeddings: int,
|
|
hidden_act: str = 'gelu',
|
|
dtype: Optional[Union[str, trt.DataType]] = None,
|
|
num_kv_heads: Optional[int] = None,
|
|
mlp_hidden_size: Optional[int] = None,
|
|
bias: bool = True,
|
|
quant_mode: QuantMode = QuantMode(0),
|
|
use_alibi: bool = True,
|
|
parallel_attention: bool = False,
|
|
new_decoder_architecture: bool = False,
|
|
logits_dtype: Union[str, trt.DataType] = 'float32',
|
|
mapping=Mapping()):
|
|
super().__init__(num_layers=num_layers,
|
|
num_heads=num_heads,
|
|
hidden_size=hidden_size,
|
|
vocab_size=vocab_size,
|
|
hidden_act=hidden_act,
|
|
max_position_embeddings=max_position_embeddings,
|
|
dtype=dtype,
|
|
num_kv_heads=num_kv_heads,
|
|
mlp_hidden_size=mlp_hidden_size,
|
|
bias=bias,
|
|
quant_mode=quant_mode,
|
|
mapping=mapping,
|
|
use_alibi=use_alibi,
|
|
parallel_attention=parallel_attention,
|
|
new_decoder_architecture=new_decoder_architecture)
|
|
vocab_size_padded = pad_vocab_size(vocab_size, mapping.tp_size)
|
|
self.lm_head = ColumnLinear(
|
|
hidden_size,
|
|
vocab_size_padded,
|
|
bias=False,
|
|
dtype=dtype,
|
|
tp_group=mapping.tp_group,
|
|
tp_size=mapping.tp_size,
|
|
gather_output=True,
|
|
)
|
|
if isinstance(logits_dtype, str):
|
|
self.logits_dtype = str_dtype_to_trt(logits_dtype)
|
|
else:
|
|
assert isinstance(logits_dtype, trt.DataType)
|
|
self.logits_dtype = logits_dtype
|
|
|
|
def forward(self,
|
|
input_ids: Tensor,
|
|
position_ids=None,
|
|
past_key_value=None,
|
|
sequence_length=None,
|
|
host_past_key_value_lengths=None,
|
|
use_cache=False,
|
|
last_token_ids=None,
|
|
attention_mask=None,
|
|
cache_indirection=None,
|
|
context_lengths=None,
|
|
host_context_lengths=None,
|
|
host_request_types=None,
|
|
max_context_length=None):
|
|
hidden_states = super().forward(
|
|
input_ids=input_ids,
|
|
position_ids=position_ids,
|
|
past_key_value=past_key_value,
|
|
sequence_length=sequence_length,
|
|
host_past_key_value_lengths=host_past_key_value_lengths,
|
|
use_cache=use_cache,
|
|
attention_mask=attention_mask,
|
|
cache_indirection=cache_indirection,
|
|
context_lengths=context_lengths,
|
|
host_context_lengths=host_context_lengths,
|
|
host_request_types=host_request_types,
|
|
max_context_length=max_context_length)
|
|
|
|
if use_cache:
|
|
hidden_states, presents = hidden_states
|
|
|
|
hidden_states = gather_last_token_logits(
|
|
hidden_states,
|
|
last_token_ids,
|
|
default_net().plugin_config.remove_input_padding,
|
|
)
|
|
|
|
# [batch_size, hidden_size] -> [batch_size, vocab_size]
|
|
lm_logits = self.lm_head(hidden_states)
|
|
lm_logits.mark_output('logits', self.logits_dtype)
|
|
|
|
if use_cache:
|
|
for i, present in enumerate(presents):
|
|
present.mark_output(f'present_key_value_{i}', self.kv_dtype)
|
|
return lm_logits, presents
|
|
|
|
return lm_logits
|
|
|
|
def prepare_inputs(self,
|
|
max_batch_size,
|
|
max_input_len,
|
|
max_new_tokens,
|
|
use_cache,
|
|
max_beam_width: int = 1):
|
|
'''
|
|
|
|
@brief: Prepare inputs Tensors for the model, the given sizes are used
|
|
to determine the ranges of the dimensions of when using TRT dynamic shapes.
|
|
|
|
@return: a list contains values which can be fed into the self.forward()
|
|
'''
|
|
|
|
# Prepare inputs
|
|
head_size = self.hidden_size // self.num_heads
|
|
num_kv_heads = self.layers[0].attention.num_attention_kv_heads
|
|
|
|
plugin_config = default_net().plugin_config
|
|
use_gpt_attention_plugin = plugin_config.gpt_attention_plugin
|
|
remove_input_padding = plugin_config.remove_input_padding
|
|
|
|
model_inputs = self.prepare_basic_inputs(
|
|
max_batch_size=max_batch_size,
|
|
max_beam_width=max_beam_width,
|
|
max_input_len=max_input_len,
|
|
max_new_tokens=max_new_tokens,
|
|
num_heads=num_kv_heads,
|
|
head_size=head_size,
|
|
num_layers=self.num_layers,
|
|
kv_dtype=self.kv_dtype,
|
|
use_gpt_attention_plugin=use_gpt_attention_plugin,
|
|
remove_input_padding=remove_input_padding,
|
|
)
|
|
|
|
return (model_inputs['input_ids'], model_inputs['position_ids'],
|
|
model_inputs['past_key_value'], model_inputs['sequence_length'],
|
|
model_inputs['host_past_key_value_lengths'], use_cache,
|
|
model_inputs['last_token_ids'], model_inputs['attention_mask'],
|
|
model_inputs['cache_indirection'],
|
|
model_inputs['context_lengths'],
|
|
model_inputs['host_context_lengths'],
|
|
model_inputs['host_request_types'], max_input_len)
|