KBlueLeaf models
4 KBlueLeaf models on Hugging Face, from 393M to 508M parameters. At the precision each one is published in, the smallest needs about 1.8 GB of VRAM (DanTagGen-beta, cheapest live fit: V100) and the largest about 2.3 GB (TIPO-500M, cheapest live fit: V100). The cheapest way to run DanTagGen-beta is $0.088/hr.
KBlueLeaf models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| DanTagGen-beta | 393M | F32 | 1.8 GB | V100 | $0.088/hr |
| DanTagGen-delta-rev2 | 393M | F32 | 1.8 GB | V100 | $0.088/hr |
| TIPO-500M | 508M | F32 | 2.3 GB | V100 | $0.088/hr |
| TIPO-500M-ft | 508M | F32 | 2.3 GB | V100 | $0.088/hr |
VRAM is for the precision the model is published in, with the same overhead and an 8,192-token context assumed on every page; see the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
