🔌
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 Jetson Thor T2000 run GPT-2 (355M)?
Yes — here's how fast and up to what context, with real measured numbers.
GPT-2 (355M) on the Jetson Thor T2000
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 186k | 140.9-167.9 t/s | ~10.3-12.3 (slow) |
| Q8 | yes | 191k | 281.7-335.7 t/s | ~20.6-24.5 (slow) |
| Q4 | yes | 194k | 503.1-599.5 t/s | ~36.7-43.8 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the Jetson Thor T2000 run GPT-2 (355M)?▶
Yes. At Q4 it generates ~503.1-599.5 tok/s and fits up to 194k context; beyond that it spills to SSD swap and slows to ~36.7-43.8 tok/s.
How much context fits?▶
At Q4, up to about 194k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.