can manici

[can manici @ManiciCan]

@deepgrove_ai I love Deepgroove team in my rtx3080 i got 27000 prefill 373 decode it's stupidly fast btw no dflash or mtp pure 373 TPS thing is please deep grove can we get agentic training i wanna use it in harness system and please update llama CPP or i can publish whichever suits you best

[can manici @ManiciCan]

@DiffuSpace_Labs @near_ai Meee

[can manici @ManiciCan]

@AdinaYakup There is no int4 release yet

[can manici @ManiciCan]

@DiffuSpace_Labs @huggingface I love it please can we get gguf file too it would be best i have 16gb vram i would love to run this model

[can manici @ManiciCan]

@VeerXn @grok Turkey should be F not B i am a Turkish citizen and I can say turkey absolute thief government stole over 200 billion dollars from bonds and stocks .

[can manici @ManiciCan]

@NFT_Chen İt is cool 3080 mobile qwen 3.8 flash gets 2000 prefill and 90tps decode pretty solid for me

[can manici @ManiciCan]

@orlixx003 Game is fucked up

[can manici @ManiciCan]

@0x0SojalSec Funny 🤣 0 optimization use strata it can give noticable boost

[can manici @ManiciCan]

@MiaAI_lab @Zai_org Depends on context if you don't hit cache it will drain your account

[can manici @ManiciCan]

@missscorpiion Kız aldatmaya gidecek izin alıyor boşta kalmamak için ve kadınlar bunu haklı göstermeye çalışıyor , yok ya ol sende öyle nede olsa çok kullanılmış kadınıda alan enayi olur elbet

[can manici @ManiciCan]

@Roobar_Plays @Pirat_Nation Its actually supports but dumb IT guys doesn't understand difference between 32gb per slot and 32gb total

[can manici @ManiciCan]

@plugyawn Hell yeah 👍 go

[can manici @ManiciCan]

@geldeki Do you use llama.cpp or vllm

[can manici @ManiciCan]

@deepgrove_ai Hey, can we get llama.cpp GGUF support? Tested it on a laptop 3080 Mobile: Q4 hits 220–260 tps, though Q2 drops to ~90 tps . Prompt ingest is crazy fast an 80k token file hits 5,000 tps prompt processing speed (Q8 qcv ), and parallel execution pulls ~600 tps aggregate on Q4.

[can manici @ManiciCan]

@JonJonloretz @deepgrove_ai Check model quant page you will see if you use Nvidia its not optimized for q2

[can manici @ManiciCan]

@yakamozsun_34 @didem5654 Yoo bal gibi hak edilmiş adama sevindim bu kadar kötü birisiyle evlenmekten kurtulmuş

[can manici @ManiciCan]

@matthewmillerai Some people use god modelike we need god level desing or god level architecture that was giving higher results now they hndle god message to block those results

[can manici @ManiciCan]

@Haticecanmyaaa İşte karakter eksikliği nedir örneği olmuş

[can manici @ManiciCan]

Ben yaparım dediğimi yaparım

[can manici @ManiciCan]

@TheAhmadOsman I new to this can someone explain to me what will go wrong if I use 8xmi50 32gb with epyc 64 core 128 thread CPU on 2000$ budget