fix: change to use the latest docker image. (All-Hands-AI#290)

Co-authored-by: Robert Brennan <[email protected]>

README.md

@@ -29,7 +29,7 @@ docker ps # this should exit successfully
 
 Then pull our latest image [here](https://github.com/opendevin/OpenDevin/pkgs/container/sandbox)
 ```bash
-docker pull ghcr.io/opendevin/sandbox:v0.1
+docker pull ghcr.io/opendevin/sandbox
 ```
 
 Then copy `config.toml.template` to `config.toml`. Add an API key to `config.toml`.

evaluation/SWE-bench/scripts/run_docker_interactive.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 
-DOCKER_IMAGE=ghcr.io/opendevin/eval-swe-bench:v0.1
+DOCKER_IMAGE=ghcr.io/opendevin/eval-swe-bench
 WORK_DIR=`pwd`
 
 docker run \

opendevin/README.md

@@ -27,7 +27,7 @@ docker build -f opendevin/sandbox/Dockerfile -t opendevin/sandbox:v0.1 .
 
 Or you can pull the latest image [here](https://github.com/opendevin/OpenDevin/pkgs/container/sandbox):
 ```bash
-docker pull ghcr.io/opendevin/sandbox:v0.1
+docker pull ghcr.io/opendevin/sandbox
 ```
 
 2. Set the `OPENAI_API_KEY`, please find more details [here](https://help.openai.com/en/articles/5112595-best-practices-for-api-key-safety). Also, choose the model you want. Default is `gpt-4-0125-preview`

opendevin/sandbox/sandbox.py

@@ -18,7 +18,9 @@
 DIRECTORY_REWRITE = config.get_or_default(
     "DIRECTORY_REWRITE", ""
 )  # helpful for docker-in-docker scenarios
-CONTAINER_IMAGE = config.get_or_default("SANDBOX_CONTAINER_IMAGE", "ghcr.io/opendevin/sandbox:v0.1")
+
+CONTAINER_IMAGE = config.get_or_default("SANDBOX_CONTAINER_IMAGE", "ghcr.io/opendevin/sandbox")
+
 # FIXME: On some containers, the devin user doesn't have enough permission, e.g. to install packages
 # How do we make this more flexible?
 RUN_AS_DEVIN = config.get_or_default("RUN_AS_DEVIN", "true").lower() != "false"