№ 0089X post
Muse Spark 1.3 on a CVE rediscovery benchmark
Philippe Dourassov ran Muse Spark 1.3 on a cybersecurity benchmark. It rediscovered 19/32 CVEs on average at pass@1 (Grok 4.6: 23.3) and 24/32 when three runs were pooled.

pilvar (Philippe Dourassov)
@pilvar222
Philippe Dourassov ran Muse Spark 1.3 on a cybersecurity benchmark. It rediscovered 19/32 CVEs on average at pass@1 (Grok 4.6: 23.3) and 24/32 when three runs were pooled.







ChatForm
Tgmlabs