-
Notifications
You must be signed in to change notification settings - Fork 0
/
eval.py
180 lines (148 loc) · 5.26 KB
/
eval.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
"""Evaluation Pipeline
To run the evaluation pipeline, run the following command:
python eval.py -cn config_path +ckpt='checkpoint_path' +split=val ++name=experiment_name
Arguments
---------
-cn (str): Path of the config file wrt the configs/ directory.
+ckpt (str): Absolute path to checkpoint file
+split (str): One of ["val", "test", "train"]
++name (str): Name of the experiment (Compulsory to pass if not passing test_run=True)
+test_run (bool): If set to True, the name check will be skipped.
Additional Information
----------------------
+ : Appending key to a config
++ : Overriding key in a config
ckpt: If Checkpoint has an = in the path, then consider using `/=` instead of the
`=` OR wrap it like '+ckpt="checkpoint_path"'
"""
import os
from typing import List
import dotenv
import hydra
from hydra.utils import instantiate
from omegaconf import DictConfig, open_dict
from pytorch_lightning import (
Callback,
LightningDataModule,
LightningModule,
Trainer,
seed_everything,
)
from pytorch_lightning.loggers import LightningLoggerBase, LoggerCollection, WandbLogger
from src.utils import utils
from utils.check_config import check_eval_config
# load environment variables from `.env` file if it exists
# recursively searches for `.env` in all folders starting from work dir
dotenv.load_dotenv(override=True)
OUT_DIR = "/output/"
log = utils.get_logger(__name__)
def get_wandb_logger(trainer: Trainer) -> WandbLogger:
"""Function to get the wandb logger
Parameters
----------
trainer : Trainer
pytorch lightning trainer object
Returns
-------
WandbLogger
pytorch lightning WandbLogger object
"""
if isinstance(trainer.logger, WandbLogger):
return trainer.logger
if isinstance(trainer.logger, LoggerCollection):
for logger in trainer.logger:
if isinstance(logger, WandbLogger):
return logger
return None
def evaluate(
config: DictConfig,
ckpt_PATH: str,
split: str = "val",
):
"""Evaluate function to evaluate a model on a given split
Parameters
----------
config : DictConfig
Configuration dictionary for setting up model, trainer, data, etc.
ckpt_PATH : str
Path to the checkpoint file
split : str, optional
Split to evaluate on, by default "val"
"""
# Set seed for random number generators in pytorch, numpy and python.random
if "seed" in config:
seed_everything(config.seed, workers=True)
# Init Lightning datamodule
log.info(f"Instantiating datamodule <{config.datamodule._target_}>")
datamodule: LightningDataModule = instantiate(
config.datamodule, data_config=config.datamodule, _recursive_=False
)
# Init Lightning model
log.info(f"Instantiating model <{config.model._target_}>")
model: LightningModule = instantiate(
config.model, model_config=config["model"], _recursive_=False
).load_from_checkpoint(ckpt_PATH, model_config=config["model"])
# Init Lightning loggers
logger: List[LightningLoggerBase] = []
if "logger" in config:
for _, lg_conf in config["logger"].items():
log.info(f"Instantiating logger <{lg_conf._target_}>")
logger.append(instantiate(lg_conf))
# Init Lightning callbacks
callbacks: List[Callback] = []
if "callbacks" in config:
for _, cb_conf in config["callbacks"].items():
if "_target_" in cb_conf:
log.info(f"Instantiating callback <{cb_conf._target_}>")
callbacks.append(instantiate(cb_conf, _recursive_=False))
# Init Lightning trainer
log.info(f"Instantiating trainer <{config.trainer._target_}>")
trainer: Trainer = instantiate(
config.trainer,
logger=logger,
callbacks=callbacks,
default_root_dir=os.getcwd(),
_convert_="partial",
)
# Send parameters from config to all lightning loggers
log.info("Logging hyperparameters!")
utils.log_hyperparameters(
config=config,
model=model,
datamodule=datamodule,
trainer=trainer,
callbacks=callbacks,
logger=logger,
)
# Get Datamodule according to split
log.info(f"Getting {split}_dataloader from datamodule")
datamodule.setup()
if split == "test":
dataloader = datamodule.test_dataloader()
elif split == "val":
dataloader = datamodule.val_dataloader()
else: # split == 'train'
dataloader = datamodule.train_dataloader()
# Perform Prediction using trainer.predict()
log.info(f"Evaluating the model on {split}_dataloader")
trainer.predict(model, dataloader)
# Log data_file version from config on wandb
data_file = config["datamodule"]["dataset"]["data_file"]
logger = get_wandb_logger(trainer=trainer)
if logger is not None:
experiment = logger.experiment
experiment.log(
{
"data_file": data_file,
"ckpt_path": ckpt_PATH,
"split": split,
}
)
@hydra.main(config_path="configs/")
def main(config: DictConfig):
# Check if config has the right structure
with open_dict(config):
check_eval_config(config)
return evaluate(config, config.ckpt, config.split)
if __name__ == "__main__":
main()