|
import os |
|
from collections import OrderedDict |
|
|
|
from safetensors.torch import save_file |
|
|
|
from jobs.process.BaseProcess import BaseProcess |
|
from toolkit.metadata import get_meta_for_safetensors |
|
from toolkit.train_tools import get_torch_dtype |
|
|
|
|
|
class BaseMergeProcess(BaseProcess): |
|
|
|
def __init__( |
|
self, |
|
process_id: int, |
|
job, |
|
config: OrderedDict |
|
): |
|
super().__init__(process_id, job, config) |
|
self.process_id: int |
|
self.config: OrderedDict |
|
self.output_path = self.get_conf('output_path', required=True) |
|
self.dtype = self.get_conf('dtype', self.job.dtype) |
|
self.torch_dtype = get_torch_dtype(self.dtype) |
|
|
|
def run(self): |
|
|
|
|
|
pass |
|
|
|
def save(self, state_dict): |
|
|
|
save_meta = get_meta_for_safetensors(self.meta, self.job.name) |
|
|
|
|
|
os.makedirs(os.path.dirname(self.output_path), exist_ok=True) |
|
|
|
for key in list(state_dict.keys()): |
|
v = state_dict[key] |
|
v = v.detach().clone().to("cpu").to(self.torch_dtype) |
|
state_dict[key] = v |
|
|
|
|
|
save_file(state_dict, self.output_path, save_meta) |
|
|
|
print(f"Saved to {self.output_path}") |
|
|