Add README and update default image

README.md

@@ -1 +1,93 @@
-# triton_cli
+# triton_cli
+
+## Quickstart
+
+```
+# Clone repo, git mode for development/contribution
+git clone [email protected]:triton-inference-server/triton_cli.git
+cd triton_cli
+
+# Should be pointing at directory containing pyproject.toml
+pip install -e .
+
+# Try it out
+triton -h
+```
+
+## Example
+
+```
+# Default install location from pip install
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ which triton
+/home/rmccormick/.local/bin/triton
+
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton -h
+usage: triton [-h] {model,repo,server} ...
+
+CLI to interact with Triton Inference Server
+
+positional arguments:
+  {model,repo,server}
+    model              Interact with running server using model APIs
+    repo               Interact with a Triton model repository.
+    server             Interact with a Triton server.
+
+options:
+  -h, --help           show this help message and exit
+
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton repo -h
+usage: triton repo [-h] {add,remove,list,clear} ...
+
+positional arguments:
+  {add,remove,list,clear}
+    add                 Add model to model repository
+    remove              Remove model from model repository
+    list                List the models in the model repository
+    clear               Delete all contents in model repository
+
+options:
+  -h, --help            show this help message and existing
+
+# Just to get a fresh start. Not necessary.
+# Can specify --repo to any repo command for custom paths.
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton repo clear
+triton - INFO - Using existing model repository: /home/rmccormick/models
+triton - INFO - Clearing all contents from /home/rmccormick/models...
+
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton repo list
+triton - INFO - Created new model repository: /home/rmccormick/models
+triton - INFO - Current repo at /home/rmccormick/models:
+models/
+
+# Will be fleshed out for TRT LLM support soon
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton repo add -m opt125 --source hf:facebook/opt-125m
+triton - INFO - Using existing model repository: /home/rmccormick/models
+triton - INFO - HuggingFace prefix detected, parsing HuggingFace ID
+triton - INFO - Adding new model to repo at: /home/rmccormick/models/opt125/1
+triton - INFO - Current repo at /home/rmccormick/models:
+models/
+└── opt125/
+    ├── 1/
+    │   └── model.json
+    └── config.pbtxt
+
+
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton server start -h
+usage: triton server start [-h] [--mode {local,docker}] [--image IMAGE] [--repo MODEL_REPOSITORY]
+
+options:
+  -h, --help            show this help message and exit
+  --mode {local,docker}
+                        Mode to start Triton with. (Default: 'docker')
+  --image IMAGE         Image to use when starting Triton with 'docker' mode
+  --repo MODEL_REPOSITORY, --model-repository MODEL_REPOSITORY, --model-store MODEL_REPOSITORY
+                        Path to local model repository to use (default: ~/models)
+
+
+# Can specify custom --image
+rmccormick@ced35d0-lcedt:~/triton/cli/triton_cli$ triton server start
+triton - INFO - Starting a Triton Server using docker
+triton - INFO - Pulling docker image nvcr.io/nvidia/tritonserver:23.11-vllm-python-py3
+...
+
+```

src/triton_cli/parser.py

@@ -239,9 +239,7 @@ def parse_args_server(subcommands):
         "--image",
         type=str,
         required=False,
-        # or eventually 23.10-py3 for generic image
-        # default="nvcr.io/nvidia/tritonserver:23.10-vllm-python-py3",
-        default="gitlab-master.nvidia.com:5005/dl/dgx/tritonserver:master.10736241-vllm-amd64",
+        default="nvcr.io/nvidia/tritonserver:23.11-vllm-python-py3",
         help="Image to use when starting Triton with 'docker' mode",
     )
     add_repo_args([server_start])