lkklein / paddletools Goto Github PK

静态图转动态图后，动态图权重字节数变大，比如 vgg16本应该是 527MB，转为动态图后大小变为795MB，这中间有什么问题吗？转换前后权重所占字节数不应该是差不多吗？

静态图转动态图时出现如下报错，是为啥呢
Traceback (most recent call last): File "change.py", line 4, in <module> static2dynamic(params_dir="params/", save_path="params") File "/home/disk2/hanzhao/anaconda3/lib/python3.7/site-packages/paddletools/checkpoints.py", line 65, in static2dynamic data, data_type, lod_info = _read_static_params(param_path) File "/home/disk2/hanzhao/anaconda3/lib/python3.7/site-packages/paddletools/checkpoints.py", line 33, in _read_static_params buf = struct.unpack("c" * buf_size, f.read(buf_size)) struct.error: unpack requires a buffer of 1634089506 bytes

Traceback (most recent call last):
File "trans_dynamic.py", line 3, in
static2dynamic(params_dir="./assets/params/", save_path="./dy_params")
File "/home/LAB/r-zhanglt/anaconda3/lib/python3.7/site-packages/paddletools/checkpoints.py", line 65, in static2dynamic
data, data_type, lod_info = _read_static_params(param_path)
File "/home/LAB/r-zhanglt/anaconda3/lib/python3.7/site-packages/paddletools/checkpoints.py", line 33, in _read_static_params
buf = struct.unpack("c" * buf_size, f.read(buf_size))

使用dynamic2static报错 paddlepaddle=1.63
File "/home/ts/miniconda3/envs/paddle/lib/python3.6/site-packages/paddletools/checkpoints.py", line 101, in dynamic2static
static_dict = _read_dynamic_params(param_file)
File "/home/ts/miniconda3/envs/paddle/lib/python3.6/site-packages/paddletools/checkpoints.py", line 49, in _read_dynamic_params
model_state_dict, _ = fluid.load_dygraph(param_file, keep_name_table=True)
File "/home/ts/miniconda3/envs/paddle/lib/python3.6/site-packages/decorator.py", line 231, in fun
args, kw = fix(args, kw, sig)
File "/home/ts/miniconda3/envs/paddle/lib/python3.6/site-packages/decorator.py", line 203, in fix
ba = sig.bind(*args, **kwargs)
File "/home/ts/miniconda3/envs/paddle/lib/python3.6/inspect.py", line 2997, in bind
return args[0]._bind(args[1:], kwargs)
File "/home/ts/miniconda3/envs/paddle/lib/python3.6/inspect.py", line 2988, in _bind
arg=next(iter(kwargs))))
TypeError: got an unexpected keyword argument 'keep_name_table'

[2021-02-24 19:44:18,704] [ INFO] - found 5 parameters. start to read.
Traceback (most recent call last):
File "/home/data/miniconda3/envs/paddle/bin/pdtools", line 8, in
sys.exit(main())
File "/home/data/miniconda3/envs/paddle/lib/python3.6/site-packages/paddletools/control.py", line 37, in main
static2dynamic(params_dir=cmd_args["--src"], save_path=cmd_args["--dst"])
File "/home/data/miniconda3/envs/paddle/lib/python3.6/site-packages/paddletools/checkpoints.py", line 65, in static2dynamic
data, data_type, lod_info = _read_static_params(param_path)
File "/home/data/miniconda3/envs/paddle/lib/python3.6/site-packages/paddletools/checkpoints.py", line 33, in _read_static_params
buf = struct.unpack("c" * buf_size, f.read(buf_size))
MemoryError

ubuntu16.04
python3.6 CUDA10.0 paddle2.0
内存16G
显卡1660ti

---------------------------------------------------------------------------KeyError Traceback (most recent call last) in
6 # !pdtools param to_dynamic -s output/darknet53/pretrain/DarkNet53/ -d DarkNet53
7 from paddletools.checkpoints import dynamic2static
----> 8 dynamic2static(param_file="output/darknet53/best_model/model", filename="111")
/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddletools/checkpoints.py in dynamic2static(param_file, filename)
99 assert len(os.listdir(filename)) == 0, "dir {} should be empty!".format(filename)
100 logger.info("start to read dynamic params...")
--> 101 static_dict = _read_dynamic_params(param_file)
102 logger.info("found {} parameters. start to save to {}...".format(len(static_dict), filename))
103 for name, data in static_dict.items():
/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddletools/checkpoints.py in _read_dynamic_params(param_file)
48 with fluid.dygraph.guard(place):
49 model_state_dict, _ = fluid.load_dygraph(param_file, keep_name_table=True)
---> 50 name_table = model_state_dict.pop("StructuredToParameterName@@")
51 for name, data in model_state_dict.items():
52 state_dict[name_table[name]] = data
KeyError: 'StructuredToParameterName@@'