Salesforce models
2 Salesforce models on Hugging Face, from 8.0B to 140.6B parameters. At the precision each one is published in, the smallest needs about 17.9 GB of VRAM (Llama-xLAM-2-8b-fc-r, cheapest live fit: RTX A5000) and the largest about 314 GB (xLAM-8x22b-r, cheapest live fit: RTX A6000). The cheapest way to run Llama-xLAM-2-8b-fc-r is $0.176/hr.
Salesforce models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Llama-xLAM-2-8b-fc-r | 8.0B | BF16 | 17.9 GB | RTX A5000 | $0.176/hr |
| xLAM-8x22b-r | 140.6B | BF16 | 314 GB | RTX A6000 × 7 | $2.54/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.
More from Salesforce
- BLIP (3 models)
