Validate Pure quant for 256K Ultra profile
This commit is contained in:
@@ -58,6 +58,7 @@ Zielplattform.
|
||||
| Fast | `qwen-fast` | 76.800 | IQ4-MIX, MTP2, vollständig GPU, CPU-mmproj |
|
||||
| Medium | `qwen-medium` | 94.208 | IQ4_XS Pure, ohne MTP, CPU-mmproj |
|
||||
| Long | `qwen-long` | 131.072 | IQ4-MIX, MTP2, FFN 0–11 auf CPU, CPU-mmproj |
|
||||
| Ultra | `qwen-ultra` | 262.144 | IQ4_XS Pure, MTP2, beide GPUs 80:20, text-only; 68,2 Tok/s und 220K-Fülltest bestanden |
|
||||
|
||||
## Router
|
||||
|
||||
|
||||
Reference in New Issue
Block a user