Recommend quantization level based on VRAM
(vram_gb: f64)
| 33 | |
| 34 | /// Recommend quantization level based on VRAM |
| 35 | fn recommend_quantization(vram_gb: f64) -> String { |
| 36 | match vram_gb { |
| 37 | v if v >= 8.0 => "Q5_K_M or Q8_0".to_string(), |
| 38 | v if v >= 4.0 => "Q4_K_M".to_string(), |
| 39 | _ => "Q2_K or Q3_K_S".to_string(), |
| 40 | } |
| 41 | } |
| 42 | |
| 43 | /// Calculate batch size based on GPU backend |
| 44 | fn calculate_batch_size(gpu_backend: &GpuBackend) -> i32 { |