Files
FastDeploy/benchmarks/yaml/qwen3dot6b-32k-wint8-a30-tp1.yaml
2025-06-29 23:29:37 +00:00

7 lines
132 B
YAML

max_model_len: 32768
max_num_seqs: 256
gpu_memory_utilization: 0.9
kv_cache_ratio: 0.75
quantization: wint8
tensor_parallel_size: 1