Can I run clef on an RTX 5090?
Yes, with a GGUF version
The published weights are too big, but the Q6_K_L file from bartowski/Cloudflare_clef-GGUF (24.0 GB) fits, with room for ≈ 70K tokens of context.
ollama run hf.co/bartowski/Cloudflare_clef-GGUF:Q6_K_LFiles and context on RTX 5090
Each file of clef (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.
| File | Size | On RTX 5090 | Context (16-bit) | Context (8-bit) |
|---|---|---|---|---|
| As published | 54.7 GB | Doesn't fit | — | — |
| BF16 | 54.1 GB | Doesn't fit | — | — |
| Q8_0 | 28.8 GB | Doesn't fit | — | — |
| Q6_K_L | 24.0 GB | Tight | 70K | 140K |
| Q6_K | 23.3 GB | Tight | 79K | 158K |
| Q5_K_M | 20.6 GB | Fits | 119K | 239K |
| Q5_K_S | 19.6 GB | Fits | 135K | 256K |
| Q4_K_L | 18.6 GB | Fits | 150K | 256K |
| Q4_1 | 17.7 GB | Fits | 163K | 256K |
| Q4_K_M | 17.7 GB | Fits | 164K | 256K |
| Q4_K_S | 16.6 GB | Fits | 180K | 256K |
| IQ4_NL | 16.3 GB | Fits | 184K | 256K |
| Q4_0 | 16.2 GB | Fits | 186K | 256K |
| IQ4_XS | 15.6 GB | Fits | 195K | 256K |
| Q3_K_L | 15.2 GB | Fits | 201K | 256K |
| Q3_K_M | 14.5 GB | Fits | 211K | 256K |
| IQ3_M | 13.8 GB | Fits | 222K | 256K |
| Q3_K_S | 13.6 GB | Fits | 224K | 256K |
| Q2_K | 11.7 GB | Fits | 252K | 256K |
| BF16 | 54.1 GB | Doesn't fit | — | — |
| Q8_0 | 28.7 GB | Doesn't fit | — | — |
| Q4_K_M | 19.2 GB | Fits | 140K | 256K |