Skip to content

Autoresearch vs Nemo rl auto research

Autoresearch is an open-source agent skill for developer tools. Nemo rl auto research is an open-source agent skill for developer tools.

  • Autoresearch has more GitHub stars (41,254 vs 4,037).
  • Autoresearch is best for reduce my benchmark's runtime by iterating on the code.
  • Nemo rl auto research is best for test a set of hypotheses to improve accuracy on a nemo-rl grpo recipe.
What it is
Autoresearch is an open-source agent skill for developer tools.
Nemo rl auto research is an open-source agent skill for developer tools.
Best for
Reduce my benchmark's runtime by iterating on the code
Test a set of hypotheses to improve accuracy on a NeMo-RL GRPO recipe
Kind
Skill
Skill
Pricing
Open source
Open source
Platforms
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
GitHub stars
41.3K
4K
Last commit
Nov 2026
Nov 2026
License
MIT
Apache-2.0
Open source
Good for
Reduce my benchmark's runtime by iterating on the code; Improve test coverage automatically with a measurable metric; Shrink my bundle size with an experiment loop
Test a set of hypotheses to improve accuracy on a NeMo-RL GRPO recipe; Run a fixed number of RL experiments overnight with a per-run time cap; Compare reward or throughput across recipe variants using git branches

What Autoresearch does

  • Interactive setup

    Collects goal, metric command, direction, scope, constraints and budget, then confirms them.

  • Git-based experiment loop

    Commits each attempt, measures it, and keeps or reverts it with git.

  • Results log

    Writes every experiment to an untracked results.tsv file.

  • Experiment strategy

    Moves from easy tweaks to combining winners, simplification and radical changes.

  • Crash handling

    Tries quick fixes and reverts after two failed attempts.

  • Final report

    Summarizes experiments, improvement percentage and suggested next steps.

What Nemo rl auto research does

  • Experiment lifecycle workflow

    Guides the agent from reading recipes and environments through baseline, hypothesis, run and result analysis.

  • Git branch per hypothesis

    Puts each experiment on its own branch under a shared prefix and keeps failed branches as records.

  • TSV research ledger

    Logs each run with metric, memory, timing, command, log path and keep, discard or crash status.

  • Stop-condition tracking

    Converts experiment counts, deadlines and per-run timeouts into monitored values and reports progress.

  • Human oversight gates

    Requires plan confirmation before branching or launching jobs and forbids destructive git operations.

  • Distributed launch guidance

    Describes choosing local GPUs, Kubernetes via a launcher skill, or Slurm based on the environment.

Autoresearch vs Nemo rl auto research: questions

Is Autoresearch or Nemo rl auto research cheaper?

Both are open source. Compare current plans on each official site.

What is the difference between Autoresearch and Nemo rl auto research?

Autoresearch is an open-source agent skill for developer tools. Nemo rl auto research is an open-source agent skill for developer tools. Autoresearch has more GitHub stars (41,254 vs 4,037).

Keep comparing