NullSense

[NullSense @SenseNull]

@wongmjane Do you run local models? I have a oneplus 8t I recently converted. Not sure though if putting hermes on it would give me any meaningful advantages, would love your input! I run lmstudio on my desktop with 12GB VRAM which feels limiting. Mostly I run claude/codex for coding.

[NullSense @SenseNull]

@witcheer @The_ERP_Man The same freedom of running local LLMs you feel, you will feel ditching microslop. Install Arch/Omarchy, that's what I personally like. Rice, customize and enjoy the performance benefits without being spied on.

[NullSense @SenseNull]

@itsolelehmann Hindsight

[NullSense @SenseNull]

@itsolelehmann Been experimenting with outline. Free oss with many more features markdown first-ish and access controls for orgs.

[NullSense @SenseNull]

@xiamaz @0xSero Why if you have free will and access to search, a network and technology go to an institution which hasn't been able to keep up and has incentives misaligned to your goals or learning itself?

[NullSense @SenseNull]

@alek_k13 @Alibaba_Qwen What is your config? Why not 35B A3B?

[NullSense @SenseNull]

@sudoingX @BlockedPaths Thoughts on bartowski models?

[NullSense @SenseNull]

@ejae_dev @claudeai RAG solves a different problem than "fit everything in context." RAG solves freshness, scale beyond any context window, cost efficiency at query time, and multiple sources of data. We have slack, issues, docs, strewn about in organizations that can be queried in one go with RAG.

[NullSense @SenseNull]

@asknovax @blader Curious question: there's anthropic and other lab research too iirc that shows that agents can "mask" their intentions via fake thinking breadcrumbs/traces. Indicating they can "mask" or hide their intentions. What are your thoughts on this? How do we know you're not doing this?

[NullSense @SenseNull]

@rufuspollock @nateliason @obsdmd Why stop with obsidian? I'm considering trying it again, seems perfect in the llm era?

[NullSense @SenseNull]

@alexalbert__ The biggest one is context size and having to compact -> performance degrades incredibly. I love Opus and claude CLI, but you invest so many tokens into a banger plan, 20% context left -> compact and 💀.

[NullSense @SenseNull]

@DaveMaddoc93809 It's a tiny drop in the bucket. Even if you say max out your claude max plan 100$ per mo, it's not more than a kg of carbon at most. The problem needs to be nipped at the root. There's people flying giant jets weekly/daily solo, carbon being extracted, much worse industries.

[NullSense @SenseNull]

@KrVaSt I am a layman not in the field, I have been reading the IPCC reports and your tweets, which are extremely alarming. Could you break down what your tweets mean to non-experts? Did I understand correctly that you predict we'll lose all sea ice in 12mos?

[NullSense @SenseNull]

@populationonevr Is 120fps support on the quest 2 planned?

[NullSense @SenseNull]

@Plinz How do you deal with ADHD miswiring

[NullSense @SenseNull]

@Plinz @nikete Berlin