deepinsight / deepinsight/insightface

ArcFace/train_parall.py couldn't print acc and lossvalue

Open
#1,540 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
29.7k
Forks
6.1k
PR merge metrics
No merged PRs in 30d

Description

I have enabled the eval_metric, but there is no any acc ad lossvalue during training.

"""
metric1 = AccMetric()
eval_metrics = [mx.metric.create(metric1)]
if config.ce_loss:
metric2 = LossValueMetric()
eval_metrics.append(mx.metric.create(metric2))
model.fit(
train_dataiter,
begin_epoch=begin_epoch,
num_epoch=999999,
eval_data=val_dataiter,
eval_metric = eval_metrics,
kvstore=args.kvstore,
optimizer=[opt, opt_fc7],
#optimizer_params = optimizer_params,
initializer=initializer,
arg_params=arg_params,
aux_params=aux_params,
allow_missing=True,
batch_end_callback=_batch_callback,
epoch_end_callback=epoch_cb)
"""

"""
INFO:root:Iter[0] Batch [2320] Speed: 1260.89 samples/sec
INFO:root:Iter[0] Batch [2340] Speed: 1264.86 samples/sec
INFO:root:Iter[0] Batch [2360] Speed: 1240.20 samples/sec
INFO:root:Iter[0] Batch [2380] Speed: 1259.37 samples/sec
INFO:root:Iter[0] Batch [2400] Speed: 1260.10 samples/sec
INFO:root:Iter[0] Batch [2420] Speed: 1262.98 samples/sec
INFO:root:Iter[0] Batch [2440] Speed: 1260.58 samples/sec
INFO:root:Iter[0] Batch [2460] Speed: 1265.03 samples/sec
INFO:root:Iter[0] Batch [2480] Speed: 1266.52 samples/sec
INFO:root:Iter[0] Batch [2500] Speed: 1274.07 samples/sec
INFO:root:Iter[0] Batch [2520] Speed: 1241.78 samples/sec
INFO:root:Iter[0] Batch [2540] Speed: 1264.50 samples/sec
INFO:root:Iter[0] Batch [2560] Speed: 1268.71 samples/sec
INFO:root:Iter[0] Batch [2580] Speed: 1265.33 samples/sec
INFO:root:Iter[0] Batch [2600] Speed: 1255.14 samples/sec
INFO:root:Iter[0] Batch [2620] Speed: 1274.51 samples/sec
"""

Look forward your help!

Contributor guide

No contributing guide indexed for this repository

Research direction

Start in ArcFace/train_parall.py at the model.fit call and inspect how eval_metric and the batch_end_callback are connected to logging. Reproduce the training run with eval_metric enabled, then verify that accuracy and loss values appear in the output as requested.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.