swe_bench

SWE-Bench Evaluation with OpenHands SWE-Bench Docker Image

This folder contains the evaluation harness that we built on top of the original SWE-Bench benchmark (paper).

UPDATE (7/1/2024): We now support the official SWE-Bench dockerized evaluation as announced here.

The evaluation consists of three steps:

1. Environment setup: install python environment, configure LLM config, and pull docker.
2. Run inference: Generate a edit patch for each Github issue
3. Evaluate patches using SWE-Bench docker

Setup Environment and LLM Configuration

Please follow instruction here to setup your local development environment and LLM.

OpenHands SWE-Bench Instance-level Docker Support

OpenHands now support using the official evaluation docker for both inference and evaluation. This is now the default behavior.

Run Inference on SWE-Bench Instances

Make sure your Docker daemon is running, and you have ample disk space (at least 200-500GB, depends on the SWE-Bench set you are running on) for the instance-level docker image.

When the run_infer.sh script is started, it will automatically pull the relavant SWE-Bench images. For example, for instance ID django_django-11011, it will try to pull our pre-build docker image sweb.eval.x86_64.django_s_django-11011 from DockerHub. This image will be used create an OpenHands runtime image where the agent will operate on.

./evaluation/swe_bench/scripts/run_infer.sh [model_config] [git-version] [agent] [eval_limit] [max_iter] [num_workers] [dataset] [dataset_split]
# e.g., ./evaluation/swe_bench/scripts/run_infer.sh llm.eval_gpt4_1106_preview HEAD CodeActAgent 300 30 1 princeton-nlp/SWE-bench_Lite test

where model_config is mandatory, and the rest are optional.

• model_config, e.g. eval_gpt4_1106_preview, is the config group name for your LLM settings, as defined in your config.toml.
• git-version, e.g. HEAD, is the git commit hash of the OpenHands version you would like to evaluate. It could also be a release tag like 0.6.2.
• agent, e.g. CodeActAgent, is the name of the agent for benchmarks, defaulting to CodeActAgent.
• eval_limit, e.g. 10, limits the evaluation to the first eval_limit instances. By default, the script evaluates the entire SWE-bench_Lite test set (300 issues). Note: in order to use eval_limit, you must also set agent.
• max_iter, e.g. 20, is the maximum number of iterations for the agent to run. By default, it is set to 30.
• num_workers, e.g. 3, is the number of parallel workers to run the evaluation. By default, it is set to 1.
• dataset, a huggingface dataset name. e.g. princeton-nlp/SWE-bench or princeton-nlp/SWE-bench_Lite, specifies which dataset to evaluate on.
• dataset_split, split for the huggingface dataset. e.g., test, dev. Default to test.

There are also two optional environment variables you can set.

export USE_HINT_TEXT=true # if you want to use hint text in the evaluation. Default to false. Ignore this if you are not sure.
export USE_INSTANCE_IMAGE=true # if you want to use instance-level docker images. Default to true

Let's say you'd like to run 10 instances using llm.eval_gpt4_1106_preview and CodeActAgent,

then your command would be:

./evaluation/swe_bench/scripts/run_infer.sh llm.eval_gpt4_1106_preview HEAD CodeActAgent 10

Run Inference on RemoteRuntime

This is in limited beta. Contact Xingyao over slack if you want to try this out!

# ./evaluation/swe_bench/scripts/run_infer.sh [model_config] [git-version] [agent] [eval_limit] [max_iter] [num_workers] [dataset] [dataset_split]
ALLHANDS_API_KEY="YOUR-API-KEY" RUNTIME=remote EVAL_DOCKER_IMAGE_PREFIX="us-docker.pkg.dev/evaluation-428620/swe-bench-images" \
./evaluation/swe_bench/scripts/run_infer.sh llm.eval HEAD CodeActAgent 300 30 16 "princeton-nlp/SWE-bench_Lite" test
# This example runs evaluation on CodeActAgent for 300 instances on "princeton-nlp/SWE-bench_Lite"'s test set, with max 30 iteration per instances, with 16 number of workers running in parallel

To clean-up all existing runtime you've already started, run:

ALLHANDS_API_KEY="YOUR-API-KEY" ./evaluation/swe_bench/scripts/cleanup_remote_runtime.sh

Specify a subset of tasks to run infer

If you would like to specify a list of tasks you'd like to benchmark on, you could create a config.toml under ./evaluation/swe_bench/ folder, and put a list attribute named selected_ids, e.g.

selected_ids = ['sphinx-doc__sphinx-8721', 'sympy__sympy-14774', 'scikit-learn__scikit-learn-10508']

Then only these tasks (rows whose instance_id is in the above list) will be evaluated. In this case, eval_limit option applies to tasks that are in the selected_ids list.

After running the inference, you will obtain a output.jsonl (by default it will be saved to evaluation/evaluation_outputs).

Evaluate Generated Patches

Download Docker Images

(Recommended for reproducibility) If you have extra local space (e.g., 200GB), you can try pull the instance-level docker images we've prepared by running:

evaluation/swe_bench/scripts/docker/pull_all_eval_docker.sh instance

If you want to save disk space a bit (e.g., with ~50GB free disk space), while speeding up the image pre-build process, you can pull the environment-level docker images:

evaluation/swe_bench/scripts/docker/pull_all_eval_docker.sh env

If you want to evaluate on the full SWE-Bench test set:

evaluation/swe_bench/scripts/docker/pull_all_eval_docker.sh instance full

Run evaluation

With output.jsonl file, you can run eval_infer.sh to evaluate generated patches, and produce a fine-grained report.

This evaluation is performed using the official dockerized evaluation announced here.

If you want to evaluate existing results, you should first run this to clone existing outputs

git clone https://huggingface.co/spaces/OpenDevin/evaluation evaluation/evaluation_outputs

NOTE, you should have already pulled the instance-level OR env-level docker images following this section.

Then you can run the following:

# ./evaluation/swe_bench/scripts/eval_infer.sh $YOUR_OUTPUT_JSONL [instance_id] [dataset_name] [split]
# For example:
./evaluation/swe_bench/scripts/eval_infer.sh evaluation/evaluation_outputs/outputs/swe_bench/CodeActAgent/gpt-4-1106-preview_maxiter_50_N_v1.0/output.jsonl

The script now accepts optional arguments:

• instance_id: Specify a single instance to evaluate (optional)
• dataset_name: The name of the dataset to use (default: "princeton-nlp/SWE-bench_Lite")
• split: The split of the dataset to use (default: "test")

For example, to evaluate a specific instance with a custom dataset and split:

./evaluation/swe_bench/scripts/eval_infer.sh $YOUR_OUTPUT_JSONL instance_123 princeton-nlp/SWE-bench test

You can also pass in a JSONL with SWE-Bench format to ./evaluation/swe_bench/scripts/eval_infer.sh, where each line is a JSON of {"model_patch": "XXX", "model_name_or_path": "YYY", "instance_id": "ZZZ"}.

The final results will be saved to evaluation/evaluation_outputs/outputs/swe_bench/CodeActAgent/gpt-4-1106-preview_maxiter_50_N_v1.0/ with the following files/directory:

• README.md: a report showing what are the instances that passed, failed, etc.
• report.json: a JSON file that contains keys like "resolved_ids" pointing to instance IDs that are resolved by the agent.
• logs/: a directory of test logs

Visualize Results

First you need to clone https://huggingface.co/spaces/OpenDevin/evaluation and add your own running results from openhands into the outputs of the cloned repo.

git clone https://huggingface.co/spaces/OpenDevin/evaluation

(optional) setup streamlit environment with conda:

cd evaluation
conda create -n streamlit python=3.10
conda activate streamlit
pip install -r requirements.txt

run the visualizer: Then, in a separate Python environment with streamlit library, you can run the following:

# Make sure you are inside the cloned `evaluation` repo
conda activate streamlit # if you follow the optional conda env setup above
streamlit run 0_📊_OpenHands_Benchmark.py --server.port 8501 --server.address 0.0.0.0

Then you can access the SWE-Bench trajectory visualizer at localhost:8501.

Submit your evaluation results

You can start your own fork of our huggingface evaluation outputs and submit a PR of your evaluation results following the guide here.