🔌
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 Transformer (Adaptive Input Embeddings) WT103?
Yes — here's how fast and up to what context, with real measured numbers.
Transformer (Adaptive Input Embeddings) WT103 on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1727k | 403.4-480.8 t/s | ~14.8-17.6 (slow) |
| Q8 | yes | 1731k | 806.8-961.6 t/s | ~29.6-35.2 (slow) |
| Q4 | yes | 1733k | 1440.8-1717.1 t/s | ~52.8-62.9 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
→ All models on the NVIDIA DGX Spark
→ Transformer (Adaptive Input Embeddings) WT103 specs
→ Other hardware
FAQ
Can the NVIDIA DGX Spark run Transformer (Adaptive Input Embeddings) WT103?▶
Yes. At Q4 it generates ~1440.8-1717.1 tok/s and fits up to 1733k context; beyond that it spills to SSD swap and slows to ~52.8-62.9 tok/s.
How much context fits?▶
At Q4, up to about 1733k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.