|
|
|
|
|
|
by bvisness
108 days ago
|
|
|
As someone on the SpiderMonkey team who had to evaluate some of Anthropic's bugs, I can definitely say that Anthropic's test cases were definitely far easier to assess than those generated by traditional fuzzers. Instead of extremely random and mostly superfluous gibberish, we received test cases that actually resembled a coherent program. |
|
|