InternLM/internlm/train/utils.py

91 lines
3.8 KiB
Python

from typing import Dict, Tuple
import torch
import torch.distributed as dist
from internlm.core.context.parallel_context import ParallelMode
from internlm.core.context.parallel_context import global_context as gpc
from internlm.model.utils import is_gate_param, is_moe_param, is_norm_param
def split_params_into_different_groups_for_optimizer(param_groups: Tuple[Dict]) -> Tuple[Dict]:
"""Split parameters into different groups for optimizer
Args:
param_groups (Tuple[Dict]): The list of parameter groups to split
Input Example:
>>> (
>>> {'name': 'default', 'params': [tensor], 'weight_decay' :xxx},
>>> )
Returns:
Tuple[Dict]: list of params groups for optimizer
Output Example:
>>> (
>>> {'name': 'default','params': [tensor],'weight_decay' :xxx},
>>> {'name': 'fp32', 'params': [tensor],'weight_decay' :xxx},
>>> {'name': 'norm', 'norm': True, 'params': [tensor],'weight_decay' :xxx},
>>> {'name': 'gate', 'gate': True, 'params': [tensor],'weight_decay' :xxx},
>>> {'name': 'moe_ep_size_4', 'moe': True, 'params': [tensor],'weight_decay' :xxx},
>>> )
"""
if isinstance(param_groups, tuple):
param_groups = list(param_groups) # Tuple cannot be modified
elif isinstance(param_groups, dict):
param_groups = [param_groups]
elif not isinstance(param_groups, list):
raise ValueError(f"Unknown param group type of {type(param_groups)}")
# create new groups for fp32, norm, moe gate and moe expert
new_groups = {}
new_groups["fp32"] = {"name": "fp32", "params": []}
if gpc.config.model.get("num_experts", 0) > 1:
# norm and gate are special group to force sync (when enable MoE).
for key in ["gate", "norm"]:
new_groups[key] = {"name": key, key: True, "params": []}
for key in gpc.expert_parallel_group_names:
new_groups[key] = {"name": key, "moe": True, "params": []}
for pgroup in param_groups:
# copy attribute from origin group, we assume the input param_groups only
# have one group, so the attribute will not be copyed multiple times.
for ori_key in pgroup.keys():
if ori_key not in ("name", "params"):
for _, group in new_groups.items():
group[ori_key] = pgroup[ori_key]
# assign param
origin_params = []
# first split the norm and gate groups, which are special case to force sync (when enable MoE),
# then fp32 group and the moe group.
for param in pgroup["params"]:
if gpc.config.model.get("num_experts", 0) > 1 and is_norm_param(param):
new_groups["norm"]["params"].append(param)
# gate param means MoE is enabled
elif is_gate_param(param):
new_groups["gate"]["params"].append(param)
elif param.dtype == torch.float32:
new_groups["fp32"]["params"].append(param)
# moe param means MoE is enabled
elif is_moe_param(param):
new_groups[param.group_name]["params"].append(param)
else:
origin_params.append(param)
# bf16 param group, which is the first group in the param groups
pgroup["params"] = origin_params
for _, g in new_groups.items():
# remove empty group, especially for fp32 group
is_empty = torch.tensor(bool(g["params"]), device=torch.cuda.current_device())
dist.all_reduce(is_empty, group=gpc.get_group(ParallelMode.MODEL))
if is_empty:
param_groups.append(g)
return tuple(param_groups)
def create_param_groups(model, weight_decay):
parameters = {"params": list(model.parameters()), "name": "default", "weight_decay": weight_decay}
return split_params_into_different_groups_for_optimizer(parameters)