mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
279 lines
12 KiB
Python
279 lines
12 KiB
Python
# SPDX-FileCopyrightText: Copyright (c) 2022-2024 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.
|
|
import json
|
|
import math
|
|
import sys
|
|
from pathlib import Path
|
|
from typing import Optional, Union
|
|
|
|
from ...layers import MoeConfig
|
|
from ...mapping import Mapping
|
|
from ..convert_utils import infer_dtype
|
|
from ..modeling_utils import PretrainedConfig, QuantConfig
|
|
|
|
|
|
class LLaMAConfig(PretrainedConfig):
|
|
|
|
def __init__(self,
|
|
*,
|
|
mlp_bias: bool = False,
|
|
attn_bias: bool = False,
|
|
rotary_base: float = 10000.0,
|
|
rotary_scaling: Optional[dict] = None,
|
|
residual_mlp: bool = False,
|
|
disable_weight_only_quant_plugin: bool = False,
|
|
moe: Optional[Union[MoeConfig, dict]] = None,
|
|
remove_duplicated_kv_heads: bool = False,
|
|
embedding_multiplier: float = 1.0,
|
|
attention_multiplier: float = 1.0,
|
|
residual_multiplier: float = 1.0,
|
|
output_multiplier_scale: float = 1.0,
|
|
**kwargs):
|
|
self.mlp_bias = mlp_bias
|
|
self.attn_bias = attn_bias
|
|
self.rotary_base = rotary_base
|
|
self.rotary_scaling = rotary_scaling
|
|
self.residual_mlp = residual_mlp
|
|
self.disable_weight_only_quant_plugin = disable_weight_only_quant_plugin
|
|
if moe is None:
|
|
# Legacy MOE config fields
|
|
moe = MoeConfig(
|
|
num_experts=kwargs.pop('moe_num_experts', 0),
|
|
top_k=kwargs.pop('moe_top_k', 0),
|
|
normalization_mode=kwargs.pop(
|
|
'moe_normalization_mode',
|
|
MoeConfig.ExpertScaleNormalizationMode.RENORMALIZE))
|
|
elif isinstance(moe, dict):
|
|
moe = MoeConfig.from_dict(moe)
|
|
assert isinstance(moe, MoeConfig)
|
|
self.moe = moe.validate()
|
|
self.remove_duplicated_kv_heads = remove_duplicated_kv_heads
|
|
self.fc_after_embed = False
|
|
self.use_input_layernorm_in_first_layer = True
|
|
self.use_last_layernorm = True
|
|
self.layer_idx_offset = 0
|
|
self.embedding_multiplier = embedding_multiplier
|
|
self.attention_multiplier = attention_multiplier
|
|
self.residual_multiplier = residual_multiplier
|
|
self.output_multiplier_scale = output_multiplier_scale
|
|
self.has_partial_lora_mask = False
|
|
|
|
super().__init__(**kwargs)
|
|
|
|
def to_dict(self):
|
|
output = super().to_dict()
|
|
# Serialize the fields added in LLaMAConfig
|
|
output['mlp_bias'] = self.mlp_bias
|
|
output['attn_bias'] = self.attn_bias
|
|
output['rotary_base'] = self.rotary_base
|
|
output['rotary_scaling'] = self.rotary_scaling
|
|
output['residual_mlp'] = self.residual_mlp
|
|
output[
|
|
'disable_weight_only_quant_plugin'] = self.disable_weight_only_quant_plugin
|
|
output['fc_after_embed'] = self.fc_after_embed
|
|
output[
|
|
'use_input_layernorm_in_first_layer'] = self.use_input_layernorm_in_first_layer
|
|
output['use_last_layernorm'] = self.use_last_layernorm
|
|
output['layer_idx_offset'] = self.layer_idx_offset
|
|
output['moe'] = self.moe.to_dict()
|
|
return output
|
|
|
|
@classmethod
|
|
def from_hugging_face(
|
|
cls,
|
|
hf_config_or_dir: Union[str, 'transformers.PretrainedConfig'],
|
|
dtype: str = 'auto',
|
|
mapping: Optional[Mapping] = None,
|
|
quant_config: Optional[QuantConfig] = None,
|
|
**kwargs):
|
|
import transformers
|
|
|
|
trust_remote_code = kwargs.pop('trust_remote_code', True)
|
|
has_partial_lora_mask = False
|
|
|
|
if isinstance(hf_config_or_dir, transformers.PretrainedConfig):
|
|
hf_config = hf_config_or_dir
|
|
else:
|
|
hf_config_dir = str(hf_config_or_dir)
|
|
if "vila" in hf_config_dir.lower():
|
|
sys.path.append(hf_config_dir + "/../VILA")
|
|
from llava.model import LlavaLlamaConfig # noqa
|
|
from llava.model import LlavaLlamaModel
|
|
transformers.AutoConfig.register("llava_llama",
|
|
LlavaLlamaConfig)
|
|
transformers.AutoModelForCausalLM.register(
|
|
LlavaLlamaConfig, LlavaLlamaModel)
|
|
|
|
hf_config = transformers.AutoConfig.from_pretrained(
|
|
hf_config_dir, trust_remote_code=trust_remote_code)
|
|
if hf_config.model_type == "llava":
|
|
# LLaVA = Vision model + Llama LLM
|
|
# We load a llava config and use its' text config as llama config
|
|
from transformers import LlavaConfig
|
|
hf_config = LlavaConfig.from_pretrained(
|
|
hf_config_dir).text_config
|
|
if hf_config.model_type == "llava_next":
|
|
from transformers import LlavaNextConfig
|
|
hf_config = LlavaNextConfig.from_pretrained(
|
|
hf_config_dir).text_config
|
|
if hf_config.model_type == "llava_llama":
|
|
hf_config.llm_cfg["architecture"] = hf_config.llm_cfg[
|
|
"architectures"][0]
|
|
hf_config.llm_cfg["dtype"] = hf_config.llm_cfg["torch_dtype"]
|
|
hf_config = PretrainedConfig.from_dict(hf_config.llm_cfg)
|
|
if hf_config.model_type == 'internlmxcomposer2':
|
|
# InternLM-XComposer2 has a mask for partial lora
|
|
# Therefore we need an additional flag for this mask
|
|
has_partial_lora_mask = True
|
|
if hf_config.model_type == 'mistral3':
|
|
from transformers import Mistral3Config
|
|
hf_config = Mistral3Config.from_pretrained(
|
|
hf_config_dir).text_config
|
|
hf_config.architectures = ["MistralForCausalLM"]
|
|
|
|
num_key_value_heads = getattr(hf_config, "num_key_value_heads",
|
|
hf_config.num_attention_heads)
|
|
if hf_config.model_type == "exaone":
|
|
hidden_act = hf_config.activation_function
|
|
# NOTE
|
|
# EXAONE also uses RMS norm but they represent as layer_norm_epsilon.
|
|
norm_epsilon = getattr(hf_config, "layer_norm_epsilon", 1e-5)
|
|
else:
|
|
hidden_act = hf_config.hidden_act
|
|
norm_epsilon = hf_config.rms_norm_eps
|
|
head_dim = getattr(
|
|
hf_config, "head_dim",
|
|
hf_config.hidden_size // hf_config.num_attention_heads)
|
|
head_size = getattr(hf_config, "kv_channels", head_dim)
|
|
attn_bias = getattr(hf_config, 'bias', False) or getattr(
|
|
hf_config, 'attention_bias', False)
|
|
rotary_scaling = getattr(hf_config, "rope_scaling", None)
|
|
rotary_base = getattr(hf_config, "rope_theta", 10000.0)
|
|
residual_mlp = getattr(hf_config, "parallel_attn_mlp_res", False)
|
|
disable_weight_only_quant_plugin = kwargs.pop(
|
|
'disable_weight_only_quant_plugin', False)
|
|
remove_duplicated_kv_heads = kwargs.pop('remove_duplicated_kv_heads',
|
|
False)
|
|
embedding_multiplier = getattr(hf_config, "embedding_multiplier", 1.0)
|
|
attention_multiplier = getattr(hf_config, "attention_multiplier", 1.0)
|
|
if attention_multiplier != 1.0:
|
|
attention_multiplier *= math.sqrt(head_size)
|
|
residual_multiplier = getattr(hf_config, "residual_multiplier", 1.0)
|
|
output_multiplier_scale = 1.0 / getattr(hf_config, "logits_scaling",
|
|
1.0)
|
|
if hf_config.model_type in ["mixtral", "arctic", "granitemoe"]:
|
|
# HF LLaMA-type models are implicitly using gated activation.
|
|
# With our MoE implementation, we must make it explicit
|
|
hidden_act = "swiglu"
|
|
moe_normalization_mode = MoeConfig.ExpertScaleNormalizationMode.RENORMALIZE
|
|
else:
|
|
moe_normalization_mode = None
|
|
moe_num_experts = getattr(hf_config, "num_local_experts", 0)
|
|
moe_top_k = getattr(hf_config, "num_experts_per_tok", 0)
|
|
moe_config = MoeConfig(num_experts=moe_num_experts,
|
|
top_k=moe_top_k,
|
|
normalization_mode=moe_normalization_mode)
|
|
moe_config.validate()
|
|
|
|
dtype = infer_dtype(dtype, getattr(hf_config, 'torch_dtype', None))
|
|
tie_word_embeddings = getattr(hf_config, 'tie_word_embeddings', False)
|
|
|
|
return cls(
|
|
architecture=hf_config.architectures[0],
|
|
dtype=dtype,
|
|
num_hidden_layers=hf_config.num_hidden_layers,
|
|
num_attention_heads=hf_config.num_attention_heads,
|
|
hidden_size=hf_config.hidden_size,
|
|
intermediate_size=hf_config.intermediate_size,
|
|
num_key_value_heads=num_key_value_heads,
|
|
head_size=head_size,
|
|
vocab_size=hf_config.vocab_size,
|
|
position_embedding_type='rope_gpt_neox',
|
|
max_position_embeddings=hf_config.max_position_embeddings,
|
|
hidden_act=hidden_act,
|
|
norm_epsilon=norm_epsilon,
|
|
attn_bias=attn_bias,
|
|
rotary_base=rotary_base,
|
|
rotary_scaling=rotary_scaling,
|
|
residual_mlp=residual_mlp,
|
|
disable_weight_only_quant_plugin=disable_weight_only_quant_plugin,
|
|
moe=moe_config,
|
|
mapping=mapping,
|
|
quantization=quant_config,
|
|
has_partial_lora_mask=has_partial_lora_mask,
|
|
remove_duplicated_kv_heads=remove_duplicated_kv_heads,
|
|
tie_word_embeddings=tie_word_embeddings,
|
|
embedding_multiplier=embedding_multiplier,
|
|
attention_multiplier=attention_multiplier,
|
|
residual_multiplier=residual_multiplier,
|
|
output_multiplier_scale=output_multiplier_scale,
|
|
**kwargs)
|
|
|
|
@classmethod
|
|
def from_meta_ckpt(cls,
|
|
meta_ckpt_dir: str,
|
|
dtype: str = 'auto',
|
|
mapping: Optional[Mapping] = None,
|
|
quant_config: Optional[QuantConfig] = None,
|
|
**kwargs):
|
|
|
|
with open(Path(meta_ckpt_dir, "params.json")) as fp:
|
|
meta_config: dict = json.load(fp)
|
|
|
|
n_embd = meta_config["dim"]
|
|
n_head = meta_config["n_heads"]
|
|
n_kv_head = meta_config.get("n_kv_heads", n_head)
|
|
vocab_size = meta_config.get("vocab_size", 32000)
|
|
|
|
# Reset vocab_size to 32000 for LLama v2 checkpoint.
|
|
if vocab_size == -1:
|
|
vocab_size = 32000
|
|
|
|
if "hidden_dim" in meta_config:
|
|
inter_size = meta_config["hidden_dim"]
|
|
else:
|
|
multiple_of = meta_config.get("multiple_of", 1)
|
|
n_embd_ = int(4 * n_embd * 2 / 3)
|
|
ffn_dim_multiplier = meta_config.get("ffn_dim_multiplier", 1)
|
|
inter_size = multiple_of * (
|
|
(int(n_embd_ * ffn_dim_multiplier) + multiple_of - 1) //
|
|
multiple_of)
|
|
|
|
dtype = infer_dtype(dtype, 'bfloat16')
|
|
|
|
if meta_config.get('use_scaled_rope'):
|
|
rotary_scaling = {"type": "llama3"}
|
|
else:
|
|
rotary_scaling = meta_config.get("rope_scaling")
|
|
|
|
# meta checkpoint don't have vocab_size|hidden_act|rotary_base specified, use same default value as HF
|
|
return cls(architecture="LlamaForCausalLM",
|
|
dtype=dtype,
|
|
num_hidden_layers=meta_config["n_layers"],
|
|
num_attention_heads=n_head,
|
|
hidden_size=n_embd,
|
|
intermediate_size=inter_size,
|
|
num_key_value_heads=n_kv_head,
|
|
vocab_size=vocab_size,
|
|
position_embedding_type='rope_gpt_neox',
|
|
max_position_embeddings=2048,
|
|
hidden_act='silu',
|
|
rotary_scaling=rotary_scaling,
|
|
rotary_base=meta_config.get('rope_theta', 10000),
|
|
norm_epsilon=meta_config["norm_eps"],
|
|
mapping=mapping,
|
|
quantization=quant_config,
|
|
**kwargs)
|