Permalink
Browse files

minor spelling tweaks

  • Loading branch information...
brettkoonce committed Oct 20, 2018
1 parent 9e96908 commit 482ae6397d6300ae056109ff5ee5e67db86b7b8c
Showing with 5 additions and 5 deletions.
  1. +1 −1 README.md
  2. +2 −2 docs/quickstart/kubernetes/README.md
  3. +2 −2 docs/troubleshooting/README.md
View
@@ -99,7 +99,7 @@ Click [**HERE**](https://github.com/PipelineAI/models/tree/master) to view model
![PipelineAI Compare Performance and Cost Per Prediction](https://pipeline.ai/assets/img/compare-cost-per-prediction.png)
## Shift Live Traffic to Maximize Revenue and Minimize Cost
![PipelineAI Traffic Shift Multi-armed Bandit Maxmimize Revenue Minimize Cost](https://pipeline.ai/assets/img/maximize-revenue-minimize-costs.png)
![PipelineAI Traffic Shift Multi-armed Bandit Maximize Revenue Minimize Cost](https://pipeline.ai/assets/img/maximize-revenue-minimize-costs.png)
## Continuously Fix Borderline Predictions through Crowd Sourcing
![Borderline Prediction Fixing and Crowd Sourcing](https://pipeline.ai/assets/img/fix-slack.png)
@@ -400,8 +400,8 @@ pipeline train-kube-start --model-name=census --model-tag=v1 --input-host-path=
Notes:
* If you change the model (`pipeline_train.py`), you'll need to re-run `pipeline train-server-build ...`
* `--input-host-path` and `--output-host-path` are host paths (outside the Docker container) mapped inside the Docker container as `/opt/ml/input` (PIPELINE_INPUT_PATH) and `/opt/ml/output` (PIPELINE_OUTPUT_PATH) respectively.
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are environment variables accesible by your model inside the Docker container.
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are hard-coded to `/opt/ml/input` and `/opt/ml/output`, respectively, inside the Docker conatiner .
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are environment variables accessible by your model inside the Docker container.
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are hard-coded to `/opt/ml/input` and `/opt/ml/output`, respectively, inside the Docker container.
* `--input-host-path` and `--output-host-path` should be absolute paths that are valid on the HOST Kubernetes Node
* Avoid relative paths for * `--input-host-path` and `--output-host-path` unless you're sure the same path exists on the Kubernetes Node
* If you use `~` and `.` and other relative path specifiers, note that `--input-host-path` and `--output-host-path` will be expanded to the absolute path of the filesystem where this command is run - this is likely not the same filesystem path as the Kubernetes Node!
@@ -117,8 +117,8 @@ rm -rf ~/.local/lib/python3.6/site-packages/cli_pipeline*
## Training Args and Paths
* If you change the model (`pipeline_train.py`), you'll need to re-run `pipeline train-server-build ...`
* `--input-host-path` and `--output-host-path` are host paths (outside the Docker container) mapped inside the Docker container as `/opt/ml/input` (PIPELINE_INPUT_PATH) and `/opt/ml/output` (PIPELINE_OUTPUT_PATH) respectively.
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are environment variables accesible by your model inside the Docker container.
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are hard-coded to `/opt/ml/input` and `/opt/ml/output`, respectively, inside the Docker conatiner .
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are environment variables accessible by your model inside the Docker container.
* PIPELINE_INPUT_PATH and PIPELINE_OUTPUT_PATH are hard-coded to `/opt/ml/input` and `/opt/ml/output`, respectively, inside the Docker container.
* `--input-host-path` and `--output-host-path` should be absolute paths that are valid on the HOST Kubernetes Node
* Avoid relative paths for * `--input-host-path` and `--output-host-path` unless you're sure the same path exists on the Kubernetes Node
* If you use `~` and `.` and other relative path specifiers, note that `--input-host-path` and `--output-host-path` will be expanded to the absolute path of the filesystem where this command is run - this is likely not the same filesystem path as the Kubernetes Node!

0 comments on commit 482ae63

Please sign in to comment.