Sabitlenmiş Tweet
Alejandro
3.8K posts
Alejandro
@snupitouski
#100T Local AI enjoyer
Cambridge, Ontario Katılım Nisan 2018
794 Takip Edilen203 Takipçiler
@0xSero i usually do this(depends on the project): create 5-10 .md with the right instructions and checklist them, with that the agent will be able to remember what is already implemented. Also making an brain.md thats the lawyer of the project, if anything is not there gg
English
How to VibeMax projects:
Run your agents in your root directory.
Example how I'm organising mine.
/root
|__/projects
|__/research
|__/resources
|__/content
| AGENTS.md
- Organises context
- Provides holistic understanding
- Easier to apply fixes from 1 project to another
English
Hoy sale una versión nueva, que jugadores NO pueden faltar?
PoderValorant@PoderValorant
Valorant Washed players 🕊️
Español
@tobicriv @GapieChill @LionzbladeYT la realidad es simple, canezerra, ion y eggster yoru carreaban ese equipo, llego vct y lastimosamente es un juego muy cambiante no puedes jugar un solo agente y los yoru que hay son mejores que el
Español
@GapieChill @LionzbladeYT Rossy, literalmente no tiene sentido y menos de IGL, canezerra baneado, ion no se ni donde esta ni cuando puede jugar, Inspire esta en la banca, Poppin esta en ese equipo y siquiera es igl y no tiene firepower, un desastre. roster desarmado
Español
6 months after winning challengers, this is what remains.
Inspire → Benched from ENVY
Eggster → Worst non-sub in VCT Americas
Canezerra → Banned from Valorant for a year
P0ppin → No longer the IGL of ENVY, bottom 10 rating in Americas
Ion2x → Too young to play on ENVY
English
Alejandro retweetledi
@DevBredda @0xSero i have a friend that use more tokens than me and he doesn’t have the problems with the “peak” hours, ollama have their own cloud infrastructure
English
@snupitouski @0xSero oh really? I am considering that as well. I got this list of questions about the subscription if you dont mind me asking!
x.com/DevBredda/stat…
Dev Bredda@DevBredda
Nice so you use it daily? my questions would be the following ranked them from most important to least important. > how stable is it? do you ever get provider problems or has it always been good to you? > how stable is the latests models? such as K2.6 and DeepSeek if you have tried them? > whats your average daily token consumption? > whats the TPS on some of the models you use?
English
@DevBredda @0xSero for me, sometimes the service with the 18$ is good overall, but some days is useless and the “peak” hours are all the day, im switching to ollama cloud
English
@0xSero how are you not experiencing any of the rate limits?
I havent been able to use the model from 9am to 6pm UK time..
would love to see if you could replicate this issue and advise their team on how to resolve it..
I know people with max sub who also having the same issue..
English
Ojalá tener 512GB de vRAM para probar este monstruo 🥲
DeepSeek@deepseek_ai
🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length. 🔹 DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models. 🔹 DeepSeek-V4-Flash: 284B total / 13B active params. Your fast, efficient, and economical choice. Try it now at chat.deepseek.com via Expert Mode / Instant Mode. API is updated & available today! 📄 Tech Report: huggingface.co/deepseek-ai/De… 🤗 Open Weights: huggingface.co/collections/de… 1/n
Español
A dual 3090 rig is perfect for running Qwen 3.6 27b in FP8 with 196k of context
also, use VLLM
English
PRAGMATA has launched with #RTXON, featuring path tracing and DLSS 4!
To celebrate, we are giving away this custom wrapped GeForce RTX 5090 featuring Hugh and Diana, perfect for the adventure that awaits on the moon.
Want it? Comment "PRAGMATA RTX" to enter!
English
Alejandro retweetledi
Alejandro retweetledi
@snupitouski @parthsareen @Teknium @valenxi_r @ollama Updating is now possible directly from the UI!
English
Alejandro retweetledi
Alejandro retweetledi
@snupitouski @parthsareen @Teknium @valenxi_r @ollama Yes, I’ll have better support for fully local models shortly!
English
@nesquena @parthsareen @Teknium @valenxi_r @ollama quick question, i’m running qwen locally, but it seems like this is mainly set up for paid/API models?
is there (or will there be) proper support for fully local models?
English
@parthsareen @Teknium @valenxi_r @ollama Hermes has been great so far! If you are so inclined, I’ve recently open sourced my Hermes WebUI too github.com/nesquena/herme… lmk what you think!
English