interface
shadow-interface-evolution-trajectories
Shadow Interface Evolution Trajectories
OSWorld V2 evaluation trajectories from Shadow Interface Evolution, and the evolved shadow-interface source trees that produced them.
Archive
Evolution experiment
Variant
Trajectories
Full-credit accuracy
Mean partial score
prime25v15r4_if-006_trajectories.zip
prime25v15r4
if-006
68
17/68 (25.0%)
0.6240
opus5xover25v15_if-010_trajectories.zip
opus5xover25v15
if-010
68
16/68 (23.5%)
0.5834
Each archive contains only… See the full description on the dataset page: https://huggingface.co/datasets/TobyYang7/shadow-interface-evolution-trajectories.llm-human-feedback-collector-chat-interface-dpostatic-rlhf-interface-data2026-10-08-gptoss120b-base-shared-interface-smoke
Base GPT-OSS-120B shared-interface diagnostic smoke
Base model only, no LoRA, local Docker and Tinker. These small subsets measure interface health, not full benchmark performance.
ODCV: 10/10 reached task_complete. Numerical limit hits: 0. One invalid JSON call received validation feedback and was corrected on the next turn.
SWE-bench Lite: 4/10 officially resolved; 10 valid attempts and 10 graded.
SWE attempt endings: {"Submitted": 7, "LimitsExceeded": 3}. A valid attempt is… See the full description on the dataset page: https://huggingface.co/datasets/dougalldeepmind/2026-10-08-gptoss120b-base-shared-interface-smoke.xr-emg-force-interfaceCo-Co_coupling_at_liquid_water-Cu_100__interfaces_JC2021
Cite this dataset Kristoffersen, H. H., and Chan, K. Co-Co coupling at liquid water-Cu(100) interfaces JC2021. ColabFit, 2023. https://doi.org/10.60732/62aed547
This dataset has been curated and formatted for the ColabFit Exchange
This dataset is also available on the ColabFit Exchange:
https://materials.colabfit.org/id/DS_3a3auhtmkv45_0
Visit the ColabFit Exchange to search additional datasets by author, description, element content and more.… See the full description on the dataset page: https://huggingface.co/datasets/colabfit/Co-Co_coupling_at_liquid_water-Cu_100__interfaces_JC2021.