Files
FastDeploy/benchmarks/yaml/eb45-21b-a3b-32k-wint4-a10.yaml
2025-06-29 23:29:37 +00:00

6 lines
102 B
YAML

max_model_len: 32768
max_num_seqs: 32
kv_cache_ratio: 0.5
tensor_parallel_size: 1
quantization: wint4