Comment by retinaros

quick question why put GLM 5.3 at 61 while a quick check on DeepSWE 1.1 puts it at 69?

also they forgot muse spark at 75% while claiming they were outshining all US models?