NovitaAI/tb21-code-debug

harbor run -d NovitaAI/tb21-code-debug

TB2.1 Code & Debug Track

Coding, software engineering, debugging, build repair, language/runtime work, and game-like reasoning tasks from Terminal-Bench 2.1.

This public Harbor dataset is curated by NovitaAI for a Novita Sandbox hackathon track. It is a category-based subset of Terminal-Bench 2.1. Agents and models are not fixed; the only required runtime environment for the hackathon is Novita Sandbox.

Dataset

Harbor dataset: NovitaAI/tb21-code-debug
Track size: 32 tasks
Source benchmark: Terminal-Bench 2.1
Included source categories: software-engineering, debugging, games
Required hackathon sandbox: -e novita

Quick Start

Run the full track once:

harbor run \
  -d NovitaAI/tb21-code-debug \
  -a <agent> \
  -m <model> \
  -e novita \
  -k 1 \
  -n 1 \
  -y

Run a small smoke test from the track:

harbor run \
  -d NovitaAI/tb21-code-debug \
  -a <agent> \
  -m <model> \
  -e novita \
  -l 1 \
  -k 1 \
  -n 1 \
  -y

Upload a public result for the hackathon leaderboard:

harbor upload jobs/<job_name> --public

Submit the resulting Harbor Hub job link to the hackathon leaderboard form.

Valid Submission Rules

A valid track submission should satisfy:

The Harbor job uses this dataset: NovitaAI/tb21-code-debug.
The job config has environment.type = "novita".
The job does not use extra hints or task-specific extra instructions.
The submitted Harbor Hub job is public.
Agent and model are free choice unless a specific event round says otherwise.

Suggested ranking fields:

Primary: mean reward
Tie-breaker 1: fewer exceptions/errors
Tie-breaker 2: lower average duration
Tie-breaker 3: lower output tokens or total tokens, if the event wants an efficiency prize

Tasks

terminal-bench/build-cython-ext
terminal-bench/build-pmars
terminal-bench/build-pov-ray
terminal-bench/cancel-async-tasks
terminal-bench/chess-best-move
terminal-bench/circuit-fibsqrt
terminal-bench/cobol-modernization
terminal-bench/code-from-image
terminal-bench/custom-memory-heap-crash
terminal-bench/fix-git
terminal-bench/fix-ocaml-gc
terminal-bench/git-leak-recovery
terminal-bench/gpt2-codegolf
terminal-bench/headless-terminal
terminal-bench/kv-store-grpc
terminal-bench/make-doom-for-mips
terminal-bench/make-mips-interpreter
terminal-bench/merge-diff-arc-agi-task
terminal-bench/overfull-hbox
terminal-bench/path-tracing
terminal-bench/path-tracing-reverse
terminal-bench/polyglot-c-py
terminal-bench/polyglot-rust-c
terminal-bench/prove-plus-comm
terminal-bench/pypi-server
terminal-bench/regex-chess
terminal-bench/schemelike-metacircular-eval
terminal-bench/sqlite-db-truncate
terminal-bench/torch-pipeline-parallelism
terminal-bench/torch-tensor-parallelism
terminal-bench/winning-avg-corewars
terminal-bench/write-compressor