| kv_config,vram_mb,latency_ms,quality_score,context_length | |
| fp16_baseline,2000,150,0.95,512 | |
| int8_kv,1200,130,0.94,512 | |
| int4_kv,800,125,0.89,512 | |
| int8_kv + int8_model,950,95,0.92,512 | |
| kv_config,vram_mb,latency_ms,quality_score,context_length | |
| fp16_baseline,2000,150,0.95,512 | |
| int8_kv,1200,130,0.94,512 | |
| int4_kv,800,125,0.89,512 | |
| int8_kv + int8_model,950,95,0.92,512 | |