Apppublic

dvilasuero/arc-easy-gptoss20b-2

sourceHugging Faceupdated 10mo agoView on Hugging Face

1likes

App README

arc_easy

This eval was run using evaljobs.

Command

bash

evaljobs inspect_evals/arc_easy \
  --model hf-inference-providers/openai/gpt-oss-20b:cheapest \
  --name arc-easy-gptoss20b-2 \
  --timeout 120 \
  --limit 10

Run with other models

To run this eval with a different model, use:

bash

evaljobs inspect_evals/arc_easy \
  --model <your-model> \
  --space <your-space> \
  --flavor cpu-basic

Replace <your-model> with your model identifier (e.g., hf/Qwen/Qwen3-1.7B) and <your-space> with your desired Space ID.

Inspect eval command

The eval was executed with:

bash

inspect eval inspect_evals/arc_easy \
  --model hf-inference-providers/openai/gpt-oss-20b:cheapest \
  --limit 10 \
  --log-shared \
  --log-buffer 100