11 lines
257 B
JSON
11 lines
257 B
JSON
{
|
|
"model": "Qwen/Qwen3-14B",
|
|
"workload": "long-doc-qa",
|
|
"kv_cache_volume": 100.0,
|
|
"tokens_per_gb_kvcache": 6553,
|
|
"ldqa_document_length": 10000,
|
|
"ldqa_query_per_document": 2,
|
|
"ldqa_shuffle_policy": "random",
|
|
"ldqa_num_inflight_requests": 3
|
|
}
|