ALFWorld agentic evaluation for SkillsRL testing. The model operates as an autonomous agent in a text-based household environment, completing tasks such as picking, cleaning, heating, cooling, and examining objects.

	Results

Metric Value

Success rate 0/2 (0.0%)

Errors 0

	Per-task-type breakdown

Task Type Success Rate

heat_then_place 0/2 0.0%

	Details

Parameter Value

Model… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/t1-skillrl-alfworld-gpt-5-dryrun01.