Kroma Pruned Experiments
Experimental structurally pruned releases of Kroma v0.3 Turbo.
Pruned v1
Structural changes:
- Removed 1 full transformer block: original block 2
- Main block count reduced from 28 โ 27
- Pruned 1,536 MLP channels from every remaining block
- MLP width reduced from 16,384 โ 14,848
The pruning is physical: the removed block and MLP channels are no longer stored in the checkpoint rather than simply being bypassed.
Available Variants
| Variant | Size |
|---|---|
| Original BF16 | ~23.8 GB |
| Pruned v1 BF16 | ~21.6 GB |
| Pruned v1 INT8 ConvRot | ~11.4 GB |
| Pruned v1 Mixed Quant | ~7.84 GB |
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐ Ask for provider support
Model tree for PotatoForge/Kroma-Pruned-Experiments
Base model
lodestones/Kroma