Essacheez/r2e-code-instruct-qwen3.5


r2e-code-instruct-qwen3.5

Generated by **Repo2RLEnv** — turning real GitHub repositories into verifiable RL environments.

💡 Browse this dataset in your browser — click the badge above or open `HuggingFaceH4/harbor-visualiser` to inspect every task's spec, instruction, oracle patch, test script, and Dockerfile.

• —Source repos (39):
• —`adrienverge/yamllint`
• —`agronholm/typeguard`
• —`alecthomas/voluptuous`
• —`andialbrecht/sqlparse`
• —`benoitc/gunicorn`
• —`bottlepy/bottle`
• —`chardet/chardet`
• —`cloudpipe/cloudpickle`
• —`encode/starlette`
• —`facelessuser/soupsieve`
• —`graphql-python/graphene`
• —`hukkin/tomli`
• —`jawah/charset_normalizer`
• —`jsvine/pdfplumber`
• —`kayak/pypika`
• —`kurtmckee/feedparser`
• —`lepture/mistune`
• —`mahmoud/glom`
• —`marshmallow-code/apispec`
• —`marshmallow-code/marshmallow`
• —`mido/mido`
• —`mozilla/bleach`
• —`oauthlib/oauthlib`
• —`pallets/jinja`
• —`paramiko/paramiko`
• —`pdfminer/pdfminer.six`
• —`pyasn1/pyasn1`
• —`pydicom/pydicom`
• —`pygments/pygments`
• —`pylint-dev/astroid`
• —`pyparsing/pyparsing`
• —`python-hyper/h11`
• —`python-jsonschema/jsonschema`
• —`python-trio/trio`
• —`scanny/python-pptx`
• —`seperman/deepdiff`
• —`sqlfluff/sqlfluff`
• —`tkrajina/gpxpy`
• —`tobymao/sqlglot`
• —Pipeline: `code_instruct`
• —Tasks: 59
• —Visibility: private
• —Spec: Harbor task format with the [metadata.repo2env] extension

How it was generated

Each task in this dataset was produced by the `code_instruct` pipeline. The pipeline mines real merged pull requests / commits from the source repo(s), applies quality filters, strips information-leakage from the instruction text, and emits a Harbor-shaped task directory with the gold patch as the oracle.

Reproduce locally:

pip install repo2rlenv
repo2rlenv generate \
  --repo <owner>/<repo> \
  --pipeline code_instruct \
  --pipeline-opt limit=10 \
  --out ./datasets/my-code_instruct

See the pipeline docs for the full option list + reward design.

Run with Harbor

Each task ships a environment/Dockerfile and tests/test.sh, so you can score patches end-to-end:

# Pull the dataset locally
repo2rlenv pull Essacheez/r2e-code-instruct-qwen3.5 /tmp/r2e-code-instruct-qwen3.5

# Confirm structural soundness — oracle adapter applies the gold patch
# and must score reward = 1.000
harbor run -p /tmp/r2e-code-instruct-qwen3.5 -a oracle --env docker

# Score an agent (claude-code + Sonnet 4.6)
harbor run \
  -p /tmp/r2e-code-instruct-qwen3.5 \
  -a claude-code -m anthropic/claude-sonnet-4-6 \
  --ak max_budget_usd=2.00 \
  --ae ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
  --env docker

The reward function ships inside the task (tests/test.sh + verifier); the breakdown is written to /logs/verifier/reward-details.json at run time.

Reward signal

The reward function is part of the task itself (tests/test.sh + the verifier code baked into the image). The full per-task breakdown is written to /logs/verifier/reward-details.json at run time — useful for slicing training data by component.

See the pipeline doc for the component-by-component design.

Layout

tasks/
└── <task-id>/
    ├── task.toml          # Harbor task with [metadata.repo2env]
    ├── instruction.md     # natural-language prompt
    ├── solution/
    │   ├── patch.diff     # oracle (gold) diff
    │   └── solve.sh       # oracle adapter applies patch.diff
    ├── environment/
    │   └── Dockerfile     # builds the task's container
    └── tests/
        └── test.sh        # verifier — writes /logs/verifier/reward.txt

License

Apache-2.0 — same as Repo2RLEnv itself. The original PR contents remain under their respective source-repo licenses; this dataset redistributes public commits under fair-use for ML research / training-data purposes.