You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
## Uploading packages to production torchserve account
105
+
106
+
As a first step binaries and docker containers need to be available in some staging environment. In that scenario the binaries can just be `wget`'d and then uploaded using the instructions below and the docker staging environment just needs a 1 line code change in https://github.com/pytorch/serve/blob/master/docker/promote-docker.sh#L8
107
+
108
+
### pypi
109
+
Binaries should show up here: https://pypi.org/project/torchserve/
110
+
111
+
You need to be on the list of maintainers to run the below.
112
+
113
+
`twine upload <path/to.wheel>`
114
+
115
+
116
+
### conda
117
+
Binaries should show up here: https://anaconda.org/pytorch/torchserve
118
+
119
+
120
+
```
121
+
# Authenticate with pytorchbot credentials
122
+
anaconda login
123
+
124
+
# Upload binaries
125
+
anaconda upload -u pytorch <path/to/.bz2>
126
+
```
127
+
128
+
## docker
129
+
Binaries should show up here: https://hub.docker.com/r/pytorch/torchserve
130
+
131
+
Change the staging org to your personal docker or test docker account https://github.com/pytorch/serve/blob/master/docker/promote-docker.sh#L8
132
+
133
+
```
134
+
# authenticate to docker with pytorchbot credentials
135
+
docker login
136
+
137
+
# upload docker image
138
+
./docker/promote-docker.sh
139
+
```
140
+
141
+
If everything looks good make sure to `export $DRY_RUN=disabled` or remove that line from the code
0 commit comments