Francisco

Abogado UNC

[Francisco @FranAg_92]

@adabadaboii @CrewsMat10 Is just isolated Incel´s posting trying to get attention. There is not the real historic discrimination that may have existed or exists in another countries. No "separated" bus, bathrooms, education, and that kind of things that existed in the US for example.

[Francisco @FranAg_92]

@0xSero have you tried this model? https://t.co/2zixcuAwcE

[Francisco @FranAg_92]

@santisiri Mas allá que es una bestia el modelo, ya se equivocó en una cosa jaja

[Francisco @FranAg_92]

@0xSero @selogerkkk why better than 27b?

[Francisco @FranAg_92]

@martinyeza Excelente que lo hagan open source 👏🏻

[Francisco @FranAg_92]

@dariogallo Estuvo como vocal suplente en el Colegio de Abogados por el oficialismo en los años 2024 - 2026.

[Ismaila @Dabofcb]

Le jour que Messi a dit souvenez-vous de ce maillot jusqu'à la fin de votre vie.

[Francisco @FranAg_92]

@AlexFromAtomic @0xSero Still, qwen3.6 27b MTP and also non MTP are beasts, usefull to solve real tasks. Not worried about acceptance rate drop

[Francisco @FranAg_92]

@juliandeangeIis Hermes agent es tremendo

[Francisco @FranAg_92]

@teKa088 fine tuneado sobre Claude? jajaja cómo fine tuneas Claude si no es un modelo abierto? compró los derechos de propiedad intelectual a Anthropic? humo

[Francisco @FranAg_92]

@LottoLabs @TheAhmadOsman Update: Using Qwen3.6 27b in Qwen Code, increases coding quality (i suposse because of optimizations).
In that harness, Qwen3.6 is very near to Sonnet 4.6 quality. @LottoLabs @TheAhmadOsman

[Francisco @FranAg_92]

@LottoLabs Home tests, with gpt 5.4 as a judge, gives me qwen3.6 27b is a little better than sonnet 4.5 but below sonnet 4.6.
@TheAhmadOsman was right

[Francisco @FranAg_92]

@LottoLabs @TheAhmadOsman absolutely

[Francisco @FranAg_92]

@Alibaba_Qwen Incredible model 👏���👏🏻

[Francisco @FranAg_92]

@TheAhmadOsman Does it support ram (cpu) offload?

[Francisco @FranAg_92]

@Snixtp @ChrisZazula @LottoLabs Qwen 27b and gemma 4 31b + hermes are still the best option (i use them actively in legal workflows in my work as a lawyer, with tools created with claudes help).

[Francisco @FranAg_92]

@Snixtp @ChrisZazula @LottoLabs In my case, i think the reason i am getting low tokens/s could be pcie-e bus.
I tried the iq4_xs quant and the iq2_xss, and both gives me 4tokens/s.

The iq2_xss size is 65.4gb (offloaded 41/63 layers in gpus) it should be performing better

[Francisco @FranAg_92]

@ChrisZazula @LottoLabs