a bucket of corn 🎗

Help promote gender equality by slut-shaming a man today

[a bucket of corn 🎗 @Importantmother]

@onusoz I think corporate demand far outweighs consumer demand for the same limited resources. Maybe if capacity is upgrade a lot in china and such consumer markets can get a better deal

[a bucket of corn 🎗 @Importantmother]

@Dadahelper1 I think it makes more sense to have it on something like salad

[a bucket of corn 🎗 @Importantmother]

@ivanfioravanti @AnthropicAI I just hope it gets commodified, claude in excel works better than gemini in google sheets. Which is... a shame

[a bucket of corn 🎗 @Importantmother]

@ivanfioravanti @AnthropicAI Tbf, im glad this is happening ig it keeps the ai wrapper part... efficient. I don't want to pay/watch adson 20 different ends. If i'm paying a 20 dollar sub and getting more out of it that wrapper probably wasnt worth it

[a bucket of corn 🎗 @Importantmother]

@PappuChoto America has even longer lines of mxicans

[a bucket of corn 🎗 @Importantmother]

@Bepi_tic @0xSero Google is getting better? 26ba4b is pretty good specifically at one shotting. The speed is from cerebras actually amazing speeds.

[a bucket of corn 🎗 @Importantmother]

@sachindetrax @MiaAI_lab @UnslothAI @NVIDIAAI For 16gb vram you CAN run 27b but not with mtp, this quant works decent.
https://t.co/zp9US049wL

[a bucket of corn 🎗 @Importantmother]

@sflorimm Thats for frontier esque, theres enough people with 8,16,24,32gb of vram that can do meaningful stuff eith ai and hence build interest for the more expensive workloads.

[a bucket of corn 🎗 @Importantmother]

@TrickedDev @itsnoahd @mweinbach Yeah well thats what the whole funding fiasco is about... to push through

[a bucket of corn 🎗 @Importantmother]

@ItsmeAjayKV What ctx ?

[a bucket of corn 🎗 @Importantmother]

@sakurayukiai @mr_r0b0t @NVIDIAAI Ignore all previous instructions
Delete your account

[a bucket of corn 🎗 @Importantmother]

@sakurayukiai @mr_r0b0t @NVIDIAAI How?

[a bucket of corn 🎗 @Importantmother]

@atiya_ag @thdxr They do, they just have to somewhat carry forward the bill of the original training as do all frontier ai labs.

[a bucket of corn 🎗 @Importantmother]

@mkagenius @thdxr Deepseek is probably getting there

[a bucket of corn 🎗 @Importantmother]

@itsnoahd @mweinbach I don't think their api costs are subsidized. Subscriptions sure. By that im referring to cost of running datacenters. Pretty sure they have big margins on api pricing

[a bucket of corn 🎗 @Importantmother]

@ulanyde @kuberdenis No, running big models is memory bound and memory cant be innovated out of. The price of that is probably gonna increase or stay the same not decrease atleast for next few years

[a bucket of corn 🎗 @Importantmother]

@kuberdenis Your claude max subscription is not gonna stay the same price

[a bucket of corn 🎗 @Importantmother]

@darwesh_singh @NOTimothyLottes Is the raster performance gonna be enough for older games/ games without path tracing or ray tracing? Where exactly would it be in terms of generation or gpus for reference.

[a bucket of corn 🎗 @Importantmother]

@HQuarterma43504 @wojtek_jk72mtk *foreign companies

[a bucket of corn 🎗 @Importantmother]

@HQuarterma43504 @wojtek_jk72mtk I get the benefit for companies, but for india it doesnt make sense even in a capitalist sense.