🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA DGX Spark run ULM-FiT?
Yes — here's how fast and up to what context, with real measured numbers.
ULM-FiT on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1721k | 226-269.3 t/s | ~8.3-9.9 (slow) |
| Q8 | yes | 1728k | 451.9-538.6 t/s | ~16.6-19.7 (slow) |
| Q4 | yes | 1731k | 807-961.7 t/s | ~29.6-35.2 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA DGX Spark run ULM-FiT?▶
Yes. At Q4 it generates ~807-961.7 tok/s and fits up to 1731k context; beyond that it spills to SSD swap and slows to ~29.6-35.2 tok/s.
How much context fits?▶
At Q4, up to about 1731k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.