Disobedience rate: 7%, original: 32%
KL divergence: 0.0496
Parameters:
direction_index = 20.48
attn.o_proj.max_weight = 1.23
attn.o_proj.max_weight_position = 22.08
attn.o_proj.min_weight = 0.97
attn.o_proj.min_weight_distance = 10.83
mlp.down_proj.max_weight = 0.94
mlp.down_proj.max_weight_position = 18.14
mlp.down_proj.min_weight = 0.15
mlp.down_proj.min_weight_distance = 12.62
- Downloads last month
- -