-
Notifications
You must be signed in to change notification settings - Fork 2
/
dino_embed.py
185 lines (177 loc) · 6.59 KB
/
dino_embed.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
181
182
183
184
185
from sklearn.metrics import roc_auc_score
import pandas as pd
import numpy as np
from classification.classification_module import ClassificationModule
from hydra import compose, initialize
from data_handling.mammo import EmbedDataModule, modelname_map
from evaluation.helper_functions import run_inference
import os
os.chdir("/vol/biomedic3/mb121/causal-contrastive/evaluation")
rev_model_map = {v: k for k, v in modelname_map.items()}
# Mapping from human readable run name to Weights&Biases run_id.
# Human readable name should be in format:
# for finetuning:
# {simclr/simclrcf/simclrcfaug}-{train_prop}-{seed}
# for linear probing
# {simclr/simclrcf/simclrcfaug}head-{train_prop}-{seed}
model_dict = {
"dinohead-0.01-11": "oahatuyu",
"dinohead-0.01-22": "3m9m0obt",
"dinohead-0.01-33": "i8jshuh4",
"dinohead-0.05-11": "o8du1wbl",
"dinohead-0.05-22": "qb8by6as",
"dinohead-0.05-33": "10256yfl",
"dinohead-0.1-11": "z7v7gyjm",
"dinohead-0.1-22": "gdoy3mx2",
"dinohead-0.1-33": "othw6zry",
"dinohead-0.25-11": "lxsyvtyq",
"dinohead-0.25-22": "xnj5sjoi",
"dinohead-0.25-33": "vjchbh7a",
"dinohead-1.0-11": "4tzsd1zx",
"dinohead-1.0-33": "tg7joka1",
"dinohead-1.0-22": "nsr22zsh",
# "dinocfhead-0.01-33": "3nzk7vyx",
# "dinocfhead-0.01-22": "yjwl1hne",
# "dinocfhead-0.01-11": "2ura4dyw",
# "dinocfhead-0.05-11": "4knaifm1",
# "dinocfhead-0.05-22": "93cejfnv",
# "dinocfhead-0.05-33": "ux2b07g1",
# "dinocfhead-0.1-11": "8prc6hln",
# "dinocfhead-0.1-22": "2jz4nn16",
# "dinocfhead-0.1-33": "27l4ygl9",
# "dinocfhead-0.25-11": "sznihnmh",
# "dinocfhead-0.25-22": "egtg5x8w",
# "dinocfhead-0.25-33": "lq4bbqac",
# "dinocfhead-1.0-22": "2ac09o5d",
# "dinocfhead-1.0-33": "gqun3yxw",
# "dinocfhead-1.0-11": "vunr7kh6",
"dino-0.01-11": "gkctgwbu",
"dino-0.01-22": "bwqmlm5n",
"dino-0.01-33": "w6cy02qm",
"dino-0.05-22": "jpsvcxvk",
"dino-0.05-33": "4eejhehl",
"dino-0.05-11": "l2crzr10",
"dino-0.1-22": "yhoqtlbg",
"dino-0.1-33": "p3ad7eih",
"dino-0.1-11": "tvuvpxr8",
"dino-0.25-22": "xpj91ptv",
"dino-0.25-11": "i7eds672",
"dino-0.25-33": "sxbolewg",
"dino-1.0-11": "njgpgkbm",
"dino-1.0-33": "5plfejhr",
"dino-1.0-22": "rf8dcfj3",
# "dinocf-0.01-33": "jjm8dep2",
# "dinocf-0.01-22": "6imqbhy5",
# "dinocf-0.01-11": "4pr5vqlb",
# "dinocf-0.1-33": "23wldcff",
# "dinocf-0.1-22": "hizz5ivw",
# "dinocf-0.1-11": "fms3npwm",
# "dinocf-0.05-11": "at8bqsfv",
# "dinocf-0.05-33": "i7nvmze9",
# "dinocf-0.05-22": "1hk4pbgp",
# "dinocf-1.0-33": "wkt22zwo",
# "dinocf-1.0-11": "451iwst9",
# "dinocf-1.0-22": "gv24sci9",
# "dinocf-0.25-22": "m1a4g9g5",
# "dinocf-0.25-11": "7jq58l9d",
# "dinocf-0.25-33": "phucy2zm",
"dinocfaughead-0.05-11": "joopo3jp",
"dinocfaughead-0.01-11": "a6nrvioo",
"dinocfaughead-0.1-11": "3wazgrbi",
"dinocfaughead-0.01-22": "86tdpmmy",
"dinocfaughead-0.25-11": "0mws6jn2",
"dinocfaughead-0.05-22": "ba2i7eoc",
"dinocfaughead-0.1-22": "knlqo7rv",
"dinocfaughead-0.25-22": "0zb21zfn",
"dinocfaughead-0.01-33": "qud58gtm",
"dinocfaughead-0.05-33": "wkap5o7k",
"dinocfaughead-0.1-33": "weynxvhr",
"dinocfaughead-0.25-33": "cvksjlsd",
"dinocfaughead-1.0-11": "6qv0ku9z",
"dinocfaughead-1.0-22": "95hsjw59",
"dinocfaughead-1.0-33": "zxipvul4",
"dinocfaug-0.05-11": "xkekulrp",
"dinocfaug-0.1-22": "ghpb8vpn",
"dinocfaug-0.05-22": "jclb1g8q",
"dinocfaug-0.1-11": "wf5j3pcm",
"dinocfaug-1.0-11": "ncco5zra",
"dinocfaug-0.05-33": "4s0ni59n",
"dinocfaug-0.25-11": "twqqf8ni",
"dinocfaug-0.25-22": "tfrsb6ef",
"dinocfaug-1.0-22": "uh104rkn",
"dinocfaug-0.1-33": "0gdbcki3",
"dinocfaug-0.01-22": "l9q2h7zf",
"dinocfaug-0.25-33": "lnshyc7y",
"dinocfaug-0.01-11": "yaf2sjkd",
"dinocfaug-0.01-33": "a3hgfisq",
"dinocfaug-1.0-33": "6lrah7s0",
"dinocf2-0.25-33": "5fcsslnz",
"dinocf2-0.25-22": "c112yhkg",
"dinocf2-0.1-22": "aq3g35rh",
"dinocf2-0.1-33": "02jwpzxr",
"dinocf2-0.01-33": "og5v6x4y",
"dinocf2-0.01-22": "bj4tvibz",
"dinocf2-0.01-11": "rikim7nk",
"dinocf2head-0.01-22": "uy3vlmgo",
"dinocf2head-0.01-11": "rudaqz35",
"dinocf2head-0.01-33": "weagk5o9",
"dinocf2head-0.1-11": "45cowjnk",
"dinocf2head-0.1-22": "djarr2j9",
"dinocf2head-0.1-33": "ok425idd",
"dinocf2head-0.25-33": "a60yd6so",
"dinocf2head-0.25-22": "zbqwemvg",
"dinocf2head-0.25-11": "sw0q1uny",
"dinocf2head-0.05-22": "nr8h8dax",
"dinocf2head-0.05-33": "s4yi5zrw",
"dinocf2head-0.05-11": "5ogvjhr0",
"dinocf2-0.05-11": "4bnn4her",
"dinocf2head-0.05-22": "nr8h8dax",
"dinocf2-0.05-33": "w23n4w8g",
"dinocf2-0.05-22": "utx7shga",
"dinocf2-1.0-11": "eel2lds9",
"dinocf2-1.0-22": "nco7tji6",
"dinocf2-1.0-33": "sser6vab",
"dinocf2head-1.0-11": "nwho8tpk",
"dinocf2head-1.0-22": "0d93bik3",
"dinocf2head-1.0-33": "oafqei2m",
}
with initialize(version_base=None, config_path="../configs"):
cfg = compose(
config_name="config.yaml",
overrides=["experiment=base_density", "data.cache=False"],
)
print(cfg)
data_module = EmbedDataModule(config=cfg)
test_loader = data_module.test_dataloader()
filename = "../outputs/dino_embed2.csv"
df = pd.read_csv(filename)
for run_name, run_id in model_dict.items():
already_in_df = run_name in df.run_name.values
if run_id != "" and not already_in_df:
print(run_name)
model_to_evaluate = f"../outputs2/run_{run_id}/best.ckpt"
classification_model = ClassificationModule.load_from_checkpoint(
model_to_evaluate, map_location="cuda"
).model.eval()
classification_model.cuda()
# ID evaluation
inference_results = run_inference(test_loader, classification_model)
scanners = np.argmax(inference_results["scanners"], 1)
for i in np.unique(scanners):
print(f"\nEvaluating scanner {i}")
sc_idx = np.where(scanners == i)
targets = inference_results["targets"][sc_idx]
preds = np.argmax(inference_results["confs"], 1)[sc_idx]
confs = inference_results["confs"][sc_idx]
res = {}
res["N_test"] = [targets.shape[0]]
res["ROC"] = [roc_auc_score(targets, confs, multi_class="ovr")]
res["Model Name"] = [rev_model_map[i]]
res["run_name"] = run_name
print(res)
df = pd.concat([df, pd.DataFrame(res, index=[0])], ignore_index=True)
df.to_csv(
filename,
index=False,
)
df