JEV: How It Works and What You Can Build

Comments · 153

• AZ

  @aziyodulla · 13 days ago · pinned

  my name is jev

  147

• JU

  @Justin-z4v · 13 days ago

  Top tier as always, Riley. Thank you. Choice: which person should I watch? Riley. Score: 5. Noul: Yes

  10

• KR

  @kraayenjon · 11 days ago

  TL;DR <br>The short version of what Jev is<br><br>- It&apos;s not a chat model. You send a state (text or JSON) plus typed<br> &nbsp;questions, and it returns typed answers:<br> &nbsp;- Choice: picks one option (up to 255)<br> &nbsp;- Score: places the input on a scale you define<br> &nbsp;- Noul: probability that a statement is true<br>- All questions in one call run in parallel. Answers in 70–500 ms.<br>- $0.042 per 1M input tokens. Output is free.<br>- Limits: 250k tokens/s, 1,200 req/min, 64k tokens per request, text only.<br><br>you will find useful the directory of jev uses cases I&apos;m building at madewithJEV<br><br>ps: adding this video as a resource there.

  6

• 4K

  @4k60hz6 · 13 days ago

  This can compliment LLM to decide quickly what kind of answer it will give, whether it should give a short answer or detailed answer, or whether it should reason deeper or lightly. So it could optimize token generation of LLM. Sometimes LLM think too hard to answer a prompt that requires only a very simple answer, such as hello, hi, thank you, etc.

  11

• AA

  @AaalanSū · 13 days ago

  I love how quick you are to try these things hands on

  14

• DI

  @digitalazorro · 12 days ago

  For a split second I honestly thought I’d finally caught onto something before you did! Turns out you were already on it. Thanks for always keeping things simple and at the forefront, and for sharing all these amazing insights—you really are fantastic!;)

• WO

  @WordsWithAndre · 12 days ago

  Yes! This was the video I was waiting for

• ST

  @stefanhapper · 12 days ago

  Great demo with the emails!

  1

• ME

  @metrodyne · 12 days ago

  Best explanation, thanks

• CA

  @carloskuhn1344 · 10 days ago

  It seems to be a classifier that take the text context to compare with the labels defined by the user.

• EP

  @Ephethios · 13 days ago

  This sound like a hole new world ❤

• IN

  @InfluintCreators · 13 days ago

  There&apos;s literally a model that is the open source version of this. That works even better <br><br>It&apos;s called Laya and made by the person who designed the foundational work that jev later used

  61