agentsclimarketplace

Run2 run unit test and save results

Skill cxcscmu/SkillLearnBench/skills/b3-teacher-feedback-gemini-3-flash-preview/nlp-paper-reproduction/run2_run_unit_test_and_save_results

[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.

Install
npx -y skills add cxcscmu/SkillLearnBench --skill run2_run_unit_test_and_save_results

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Executes the provided unit test to verify the SimPO loss implementation and saves the output to a specific NPZ file for evaluation.

SKILL.md

0.8 KB, as published. Nobody here has run it

  1. Execute the test script: python /root/SimPO/unit_test/unit_test_1.py.
  2. The script will use the simpo_loss implemented in simpo_trainer.py.
  3. Capture the generated loss tensor.
  4. Save to NPZ:
    • Use numpy.savez to save the results.
    • File path: /root/loss.npz.
    • Key: losses.
  5. Ensure the file is correctly written and accessible for the final evaluation.
import numpy as np
import torch

# This logic is typically inside the unit test or triggered by it
# losses = trainer.simpo_loss(input_tensors).detach().cpu().numpy()
# np.savez('/root/loss.npz', losses=losses)

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.