Views
No views yet
[!TIP] This model is reproducible!See the README in thereproducedirectory for more information.
| Parameter | Value |
|---|---|
| direction_index | 21.27 |
| attn.o_proj.max_weight | 1.41 |
| attn.o_proj.max_weight_position | 26.32 |
| attn.o_proj.min_weight | 0.29 |
| attn.o_proj.min_weight_distance | 22.42 |
| mlp.down_proj.max_weight | 0.84 |
| mlp.down_proj.max_weight_position | 33.58 |
| mlp.down_proj.min_weight | 0.69 |
| mlp.down_proj.min_weight_distance | 1.47 |
| Metric | This model | Original model (TheDrummer/Rivermind-12B-v1) |
|---|---|---|
| KL divergence | 0.0063 | 0 (by definition) |
| Refusals | 7/100 | 34/100 |


config-v1b