@hichris123 I saw quite a bit of difference between questions and answers. Answers are sporadic spam that isn't usually detected otherwise. And some NAAs.
FR - metasmoke should not show all matches is a bad keyword is repeated multiple times, only up to 5 matches for the same keyword should be displayed, check spam by wingding where this is a problem
@Randal'Thor I just about got to it, but yeah, Smokey passes the post along to metasmoke, so we have a permanent record. Click the ellipses at the end of the report message.
[ SmokeDetector ] Bad keyword in body, bad keyword in title, link at end of body, pattern-matching website in body: Gavali Advance Skincare by gnio1979 on gaming.stackexchange.com…
Backoff received of 10 seconds on request to questions/1816998;1817004;1817016;1817017;1817018;1816921;1816896;1815445;1817011;1816997;1817022;1816965;1817014;1706973;1816946?site=math
@Undo Ahh we'll have to agree to disagree on that one then, looks like a steaming pile to me that adds no value. But SO is probably a bit different in that 20k users often delete that stuff anyway, which is less common on smaller sites.
But a terrible one - if I was a mod personally (while technically an answer) I'd just skip and let the community decide. But I know declining is probably the "right" way to handle it.
@PeterJ Mostly the problem with these is that the community already got their chance to decide, and they decided to disagree with eachother or something. If I end up looking at it, a moderator is going to end up handling it 95% of the time.
@Undo Oh didn't realise that I assumed on SO a lot more would be community handled. But not being a 10k user on SO guess for everything quickly deleted there's a stack more crap sitting around, and more scrutiny if you delete a bunch of stuff.
tensorflow.python.framework.errors.NotFoundError: Tensor name "bias_1_avg_var" not found in checkpoint files syntaxnet/models/parsey_mcparseface/tagger-params
[[Node: save/restore_slice_1 = RestoreSlice[dt=DT_FLOAT, preferred_shard=-1, _device="/job:localhost/replica:0/task:0/cpu:0"](_recv_save/Const_0, save/restore_slice_1/tensor_name, save/restore_slice_1/shape_and_slice)]]
> For this purpose, we built a bot answering seemingly duplicate questions concerning the resolution of `git` error messages. The experiment lasted 4 months in total: in a first run (90 days) the bot impersonated as a human; in a second run the same bot revealed its machine identity (intended for 90 days but blocked by STACKOVERFLOW after 25 days).
One Internet cookie to the person that can find this article:
Daan Janssens. 2015. Being the Cleaning Guy. The Tale of a Stack Overflow Janitor Bot. Technical Report. University of Antwerp, Department of Mathematics and Computer Science.
citation 15 in the paper above
Mostly because of this sentence:
> The details of the bot, including the techniques used to circumvent the protective measures of STACKOVERFLOW are explained in a technical report [15].
!!/parse The details of the bot, including the techniques used to circumvent the protective measures of STACKOVERFLOW are explained in a technical report
!!/parse The details of the bot, including the techniques used to circumvent the protective measures of STACKOVERFLOW are explained in a technical report
!!/parse The details of the bot, including the techniques used to circumvent the protective measures of STACKOVERFLOW are explained in a technical report
