ZML releases free product to speed inference across AI chips · HackerLangs