-
Notifications
You must be signed in to change notification settings - Fork 8
/
manifest.yml
54 lines (51 loc) · 2.02 KB
/
manifest.yml
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
# name: Replace with any name for your model
# description: Replace with any description for your model
# version: Replace with any version of your model
# gpus: Replace with the number of gpus to be used in
# training your model
# cpus: Replace with the number of cpus to be used in
# training your model
# memory: Replace with the amount of memory to be
# dedicated to training your model
name: Fashion-MNIST-GPU-Job
description: Fashion-MNIST keras model with k80 GPU
version: "1.0"
gpus: 1
cpus: 8
memory: 16GB
# Object stores that allow the system to retrieve training data.
# id: The data_store id
# type: The type of data_store
# training_data: container: Replace with the name of the bucket at which
# you stored the fashion MNIST dataset
# training_results: container: Replace with the name of the bucket where
# the resulting model should be saved to
# connection: type: The type of connection
# connection: auth_url: Replace with your Cloud Object Storage Endpoint url
# for IBM Cloud
# connection: user_name: Replace with the access_key_id found in the service
# credentials tab on IBM Cloud
# connection: password: Replace with the secret_access_key found in service
# credentials tab on IBM Cloud
data_stores:
- id: test-datastore
type: mount_cos
training_data:
container: fashion-mnist
training_results:
container: mnist-trained-model
connection:
auth_url: http://s3.default.svc.cluster.local
user_name: test
password: test
# name: The name of the Deep Learning framework that will be used
# version: The version of the framework to be used
# command: The command to initiate training
framework:
name: tensorflow
version: latest-gpu-py3
command: pip3 install keras; pip3 install scikit-learn; python3 experiment.py
# Setup for Graphics that will be plotted within FfDL UI
evaluation_metrics:
type: tensorboard
in: "$JOB_STATE_DIR/logs/tb/test"