Views
No views yet
[!TIP] This model is reproducible!See the README in thereproducedirectory for more information.
| Parameter | Value |
|---|---|
| direction_index | 27.45 |
| attn.o_proj.max_weight | 1.44 |
| attn.o_proj.max_weight_position | 24.54 |
| attn.o_proj.min_weight | 1.01 |
| attn.o_proj.min_weight_distance | 6.27 |
| mlp.down_proj.max_weight | 1.22 |
| mlp.down_proj.max_weight_position | 24.23 |
| mlp.down_proj.min_weight | 0.67 |
| mlp.down_proj.min_weight_distance | 2.89 |
| Metric | This model | Original model (ibm-granite/granite-vision-4.1-4b) |
|---|---|---|
| KL divergence | 0.0058 | 0 (by definition) |
| Refusals | 3/100 | 75/100 |