feat: Add vLLM inference support with a concurrent client and update …#4
feat: Add vLLM inference support with a concurrent client and update …#4shihanqu wants to merge 4 commits intoOrbiter:mainfrom
Conversation
…benchmark configuration and execution. GPT-OSS-120b benchmark results added to benchmark.json
|
thanks! unfortunately, the benchmark.js file is completely rewritten because your output of the json has different spacing. I have not enough experience with python to understand how this happens, but I don't want to overwrite all benchmarks with new values, even if those values are of the same value. Can we find out why your copy of the benchmarks writes this in such a different way? I am using the benchmark on many different computers and all of them output the same spacing. |
|
I don't think it has different spacing? I'm seeing them as character-for-character identical files, aside from the new additional entry for gpt-oss-120b at the top of the file. Perhaps since the addition was made at the top, github diff viewer couldn't recognize the additions. I'll modify my benchmark.json to have gpt-oss-120b at the end if that helps |
Results for: GPT-OSS-120B MED and HIGH. Minimax m2 REAP 50 Int4
2 participants
…benchmark configuration and execution. GPT-OSS-120b benchmark results added to benchmark.json