This is a great review of sparkDash! It talks about my recent (and first) PR to incorporate Cost Compare into the project so folks can see their potential savings. Iโm happy I was able to contribute to such a great project by @MiaAI_lab and that the PR has been so well received!
@ImMythicAZ To be fair, Trey McBridge is an All Pro and Michael Wilson is likely to be a Pro Bowler. Itโs not an apples to apples comparison like you make it up to be. They rightfully deserve more targets, but I do agree we didnโt appropriately utilize Marv in his time here.
@MiaAI_lab It would be nice to see a cost โsavingsโ. Like how much it would have been if it were an X frontier model for the same-token consumption.
@Gambo987 Do you think theyโll hold onto Marte or explore a trade scenario this offseason? Marte has veto power, so Iโm not sure I see them being successful in trading him even if they try.
@MiaAI_lab@NVIDIAAI Great release. One thing I have noticed with Tensorflow is that the engine really makes my sparks run hot. Constantly seeing 95C+ on my nodes when its under load (1c stream). GPU clock is capped at 2000. Never had this experience with vLLM. Anyone else experiencing this?
@JohnnyVenerable Very happy we saved our better bullpen pitchers for the last game of the season/playoffs, as TL was trying to ensure they remained โhealthyโ.