🔌
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 XLM-RoBERTa?
Yes — here's how fast and up to what context, with real measured numbers.
XLM-RoBERTa on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1718k | 181.2-215.9 t/s | ~6.6-7.9 (slow) |
| Q8 | yes | 1727k | 362.3-431.8 t/s | ~13.3-15.8 (slow) |
| Q4 | yes | 1730k | 647-771.1 t/s | ~23.7-28.2 (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 XLM-RoBERTa?▶
Yes. At Q4 it generates ~647-771.1 tok/s and fits up to 1730k context; beyond that it spills to SSD swap and slows to ~23.7-28.2 tok/s.
How much context fits?▶
At Q4, up to about 1730k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.