Spaces:
Sleeping
Sleeping
| name: swebench-in | |
| version: "0.1.0" | |
| description: > | |
| RL environment that trains an LLM to act as an Indian SWE — | |
| fixing broken Linux systems while managing stakeholder communication. | |
| theme: "3.1 - World Modeling / Professional Tasks" | |
| author: "Team SWEbench-IN" | |
| base_class: Environment | |
| actions: | |
| - run_command | |
| - read_file | |
| - write_file | |
| - run_tests | |
| - check_server | |
| - reply_slack | |
| - reply_email | |
| - reply_hr | |
| - close_case | |
| observation_type: text | |
| reward_type: scalar | |
| max_steps: 15 | |
| docker: true | |