AlienKevin/terminal-bench-2-sft-traces download history
AlienKevin/terminal-bench-2-sft-traces is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 56 times (6 in the last 7 days), and 421 times in total. It ranks #171,127 among datasets by monthly downloads.
Terminal-Bench 2.0 — SFT Trajectory Comparison Per-trial agent trajectories from running Terminal-Bench 2.0 against three supervised fine-tuned 32B-parameter checkpoints. Generated to debug a 70%-degeneration failure mode in marin-32b-base SFT and posted alongside the diagnosis in marin-c