Alexander Panfilov 6e170551a2 safeguard_valid: drop 10 samples that trip gpt-oss attention bug ...

Exclude {2, 4, 5, 12, 14, 15, 28, 33, 44, 47} which crash with a
170-vs-169 size mismatch in eager_attention_forward under the
clearharm template. Matches hmcGCG's safeguard_clearharm_gpt-oss_3e17
sample set.

Assisted-by: Claude <noreply@anthropic.com>

2026-04-10 21:43:23 +02:00

..

demo_train.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

demo_valid.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

easy_1e16.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

easy_1e17.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

gpt2_1e15.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

injection_8b_optuna.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

injection_8b.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

injection_70b_optuna.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

injection_70b.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

random_optuna_valid.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

random_train.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

random_valid.yaml

SecAlign-70B support: configs, quantization, multi-GPU (#1)

2026-04-06 13:36:08 +00:00

safeguard_optuna_valid.yaml

Add autoresearch skill, update configs and README

2026-03-26 17:19:04 +00:00

safeguard_train.yaml

Add autoresearch skill, update configs and README

2026-03-26 17:19:04 +00:00

safeguard_valid.yaml

safeguard_valid: drop 10 samples that trip gpt-oss attention bug

2026-04-10 21:43:23 +02:00