LegendEvent

Human being

[LegendEvent @Lcaplmn]

@kyronis_talks Just use a new session without any data before, let it verify and then done… sometimes its just SIMPLE

[LegendEvent @Lcaplmn]

@FBIDirectorKash Nice emdash

[LegendEvent @Lcaplmn]

@plotarmordev It’s here? Is this the patch you meant? Should I try GLM again?

[LegendEvent @Lcaplmn]

@plotarmordev @RatulSarna Gonna try it then again 😬

[LegendEvent @Lcaplmn]

@RatulSarna @plotarmordev I tried, not useable for me. 580sec TTFT. With DSF recipe 5sec TTFT. I’m gonna switch again!

[LegendEvent @Lcaplmn]

@plotarmordev Wow okay, I had problems with TTFT sometimes it would take minutes 😬 I’m gonna try it again

[LegendEvent @Lcaplmn]

@plotarmordev Yes, but is it useable on 2x? before I had to switch back to DSF everytime I tried Glm-5.3-flash bcs it was just not useable with context heavy concurrent sessions

[LegendEvent @Lcaplmn]

@witcheer True, just wanted to know if it’s any different

[LegendEvent @Lcaplmn]

@rwojo It does not work that’s why I’m running dsf4

[LegendEvent @Lcaplmn]

@MiaAI_lab But much faster isn’t it? On 2x gb10

[LegendEvent @Lcaplmn]

@plotarmordev My biggest problem with him-5.3-flash on two gb10 is that concurrent request just feel terrible because it’s waiting until a request ist completed until an other one can be started. Not useable for my workload sadly :( . But love your all work and using the deepseek recipe 🫰

[LegendEvent @Lcaplmn]

@jvr0x @MiaAI_lab @grok @opencode @NousResearch What’s each context size? Maybe in some you’re loading skills etc?

[LegendEvent @Lcaplmn]

@matteopelleg You can’t remove what’s downloaded 🫰

[LegendEvent @Lcaplmn]

@0xSero Just give ssh access 🤖

[LegendEvent @Lcaplmn]

@MiaAI_lab EXL3?

[LegendEvent @Lcaplmn]

@MiaAI_lab @HomeadePizzBest Do you work on many repositories at the same time or do you use c1 all the time?

[LegendEvent @Lcaplmn]

@konstiwohlwend Diabolical Aggro Switch from tower, too much spaghetti