EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

feat: update experiment_runner (#5360)

Browse Source
This commit is contained in:
Yineng Zhang
2025-04-13 15:37:05 -07:00
committed by GitHub
parent 39e411385c
commit 39d90449f3
3 changed files with 81 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
test/srt/configs/llama_405b.yaml Normal file
View File

@@ -0,0 +1,28 @@
tasks:
- name: sglang-8192-1024-concurrency1
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 1 --num-prompts 5 --output-file llama_405b_results.jsonl
- name: sglang-8192-1024-concurrency2
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 2 --num-prompts 10 --output-file llama_405b_results.jsonl
- name: sglang-8192-1024-concurrency4
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 4 --num-prompts 20 --output-file llama_405b_results.jsonl
- name: sglang-8192-1024-concurrency8
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 8 --num-prompts 32 --output-file llama_405b_results.jsonl
- name: sglang-8192-1024-concurrency16
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 16 --num-prompts 48 --output-file llama_405b_results.jsonl
- name: sglang-8192-1024-concurrency24
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 24 --num-prompts 72 --output-file llama_405b_results.jsonl
- name: sglang-8192-1024-concurrency32
server_cmd: python3 -m sglang.launch_server --model nvidia/Llama-3.1-405B-Instruct-FP8 --tp 8
client_cmd: python3 -m sglang.bench_serving --random-range-ratio 1 --random-input-len 8192 --random-output-len 1024 --max-concurrency 32 --num-prompts 96 --output-file llama_405b_results.jsonl