SWE-agent
SWE-agent copied to clipboard
princeton-nlp
data must be str, not NoneType
Describe the issue
Hi I create a very simple repo and create an issue.
docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock \
-v $(pwd)/keys.cfg:/app/keys.cfg \
sweagent/swe-agent-run:latest \
python run.py --image_name=sweagent/swe-agent:latest \
--model_name azure:gpt4 \
--data_path https://github.com/viciaky/test/issues/1 \
--config_file config/default_from_url.yaml --skip_existing=False
it showed the correct result but never save in prediction jsonl and model patch
Error: INFO Found submission:
INFO Saved trajectory to trajectories/root/azure-gpt4__viciaky__test__default_from_url__t-0.00__p-0.95__c-3.00__install-1/viciaky__test-i1.traj
INFO Saved predictions to trajectories/root/azure-gpt4__viciaky__test__default_from_url__t-0.00__p-0.95__c-3.00__install-1/all_preds.jsonl
Traceback (most recent call last):
File "/app/run.py", line 144, in main
save_patch(traj_dir, instance_id, info)
File "/app/run.py", line 271, in save_patch
patch_output_file.write_text(model_patch)
File "/usr/local/lib/python3.9/pathlib.py", line 1283, in write_text
raise TypeError('data must be str, not %s' %
TypeError: data must be str, not NoneType
WARNING ❌ Failed on viciaky__test-i1: data must be str, not NoneType
Suggest an improvement to documentation
No response
I looked at .traj file here is the info, and the "submission" is null. i am using azure gpt4 endpoint. is there any configuration i am missing here?
"info": {
"exit_status": "submitted",
"submission": null,
"model_stats": {
"total_cost": 0.11188000000000001,
"instance_cost": 0.11188000000000001,
"tokens_sent": 10273,
"tokens_received": 305,
"api_calls": 1
}
}
Thanks for the report, this was probably fixed in https://github.com/princeton-nlp/SWE-agent/pull/242. Let me know if switching to the latest version didn't solve this!
@klieret I'm using v0.4.0 and I seem to have the same kind of issue:
[...]
Given the complexity of the issue and the lack of explicit latency handling in the examined files, it may be necessary to consult with the development team or documentation to understand how latency is currently managed for MIDI regions and to determine the best
approach for implementing any required adjustments.
INFO 🎬 ACTION (primary)
submit
INFO Found submission:
INFO Saved trajectory to trajectories/root/gpt4__issue__default__t-0.00__p-0.95__c-10.00__install-1/967ef2.traj
INFO Saved predictions to trajectories/root/gpt4__issue__default__t-0.00__p-0.95__c-10.00__install-1/all_preds.jsonl
Traceback (most recent call last):
File "/app/run.py", line 150, in main
patch_path = save_patch(traj_dir, instance_id, info)
File "/app/run.py", line 281, in save_patch
patch_output_file.write_text(model_patch)
File "/usr/local/lib/python3.9/pathlib.py", line 1283, in write_text
raise TypeError('data must be str, not %s' %
TypeError: data must be str, not NoneType
WARNING ❌ Failed on 967ef2: data must be str, not NoneType
INFO Beginning environment shutdown...
INFO Agent container stopped
INFO Found image sweagent/swe-agent:latest with tags: ['sweagent/swe-agent:latest'], created: 2024-04-15T19:20:51.646667378Z for linux arm64.
DEBUG Starting container with command: docker run -i --rm --name sweagent-swe-agent-latest-7e58b0d4a9 sweagent/swe-agent:latest /bin/bash -l -m
INFO 🌱 Environment Initialized
Could you send me the full ommand that you ran?
Also it looks like you're running from the swe-agent-run docker container, right? There's no v0.4.0 docker container yet, so I'm a bit confused what version you're running on.
In fact, on the latest swe-agent-run docker container line 150 also doesn't correspond to the one from the traceback you sent. I believe you might be running on an old container.
Oh! I'm new to Docker, so I didn't know there was a matching docker container version.
I seem to be using the latest version:
docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock -v ./:/app/studio -v /Users/maximedupre/Desktop/SWE-agent/keys.cfg:/app/keys.cfg -v /Users/maximedupre/Desktop/docker-volume:/app/trajectories/root sweagent/swe-agent-run:latest python run.py --image_name=sweagent/swe-agent:latest --model_name gpt4 --per_instance_cost_limit 10.00 --data_path studio/issue.md --repo_path studio --config_file config/default.yaml --skip_existing=False
Edit
I believe that I need to run docker pull sweagent/swe-agent:latest and that it's the version of my docker container that matters, not which git branch I have checkout :D
Edit 2
The error does not occur again after running docker pull sweagent/swe-agent:latest and re-running the sweagent command.
