The flash model was reportedly post-trained on Inkling-Small; one observer initially thought Cognition had entered pretraining.
One account says Cognition post-trained its new flash model on top of Inkling-Small. Another identifies the model as SWE-2 Flash and says its Inkling-Small base dispelled an initial impression that Cognition had entered the pretraining business.
—
5 Sources, first seen ago
Cognition says its upgraded tool prioritizes findings against a team-built threat model and opens pull requests for proposed fixes.
Alexandr Wang calls Scale alumni the 'first true mafia of the modern tech era,' listing their current roles.
Cognition says its upgraded tool prioritizes findings against a team-built threat model and opens pull requests for proposed fixes.
Alexandr Wang calls Scale alumni the 'first true mafia of the modern tech era,' listing their current roles.
@soumithchintalaCognition's new flash model is post-trained on top of Inkling-small!
@teortaxesTex> SWE-2 Flash is based on @thinkymachines Inkling-Small Phew, thought for a moment that Cognition got into pretraining business. Anyway, clever idea to display it this way. in the blog, looks a bit different.
@ADarmouniInteresting resul- based on INKLING SMALL? Forgive me what Massive props to Thinky lmao@soumithchintalaCognition's new flash model is post-trained on top of Inkling-small!@teortaxesTex> SWE-2 Flash is based on @thinkymachines Inkling-Small Phew, thought for a moment that Cognition got into pretraining business. Anyway, clever idea to display it this way. in the blog, looks a bit different.@ADarmouniInteresting resul- based on INKLING SMALL? Forgive me what Massive props to Thinky lmao@soumithchintalaCognition's new flash model is post-trained on top of Inkling-small!@teortaxesTex> SWE-2 Flash is based on @thinkymachines Inkling-Small Phew, thought for a moment that Cognition got into pretraining business. Anyway, clever idea to display it this way. in the blog, looks a bit different.@ADarmouniInteresting resul- based on INKLING SMALL? Forgive me what Massive props to Thinky lmao@soumithchintalaCognition's new flash model is post-trained on top of Inkling-small!@teortaxesTex> SWE-2 Flash is based on @thinkymachines Inkling-Small Phew, thought for a moment that Cognition got into pretraining business. Anyway, clever idea to display it this way. in the blog, looks a bit different.@ADarmouniInteresting resul- based on INKLING SMALL? Forgive me what Massive props to Thinky lmao#7
#7
