andylizf
Joined 1 year ago
Karma
135
Submissions
10
Member since
Feb 27
Recent submissions
Comment · 8 months ago
Comment · 1 year ago
Author here. We used ADRS to automatically evolve a scheduling algorithm starting from a simple greedy policy. Surprisingly, it outperformed Uniform Progress, the NSDI'24 best paper algorithm by discovering "selective waiting" strategies and adaptive pattern recognition.
Counter-intuitive finding: starting from a weaker baseline (greedy) worked better than starting from the SOTA algorithm. The AI found new insights rather than incremental improvements.
The entire process took 5 hours and cost <$20. We're open-sourcing everything and have applied this to other systems problems (MoE load balancing with 5x speedup, see https://news.ycombinator.com/item?id=45688236).
Happy to answer questions about the methodology or results!
Comment · 1 year ago
Yeah that's it. My bad lol
Comment · 1 year ago
Exactly. That's because instead of just mapping keywords, vector search stores the rich meaning of the text as massive data structures, and LEANN is our solution to that paradoxical inefficiency.
Comment · 1 year ago
Yeah, that's a fair point at first glance. 50GB might not sound like a huge burden for a modern SSD.
However, the 50GB figure was just a starting point for emails. A true "local Jarvis," would need to index everything: all your code repositories, documents, notes, and chat histories. That raw data can easily be hundreds of gigabytes.
For a 200GB text corpus, a traditional vector index can swell to >500GB. At that point, it's no longer a "meager" requirement. It becomes a heavy "tax" on your primary drive, which is often non-upgradable on modern laptops.
The goal for practical local AI shouldn't just be that it's possible, but that it's also lightweight and sustainable. That's the problem we focused on: making a comprehensive local knowledge base feasible without forcing users to dedicate half their SSD to a single index.
Comment · 1 year ago
This is fantastic work. The focus on a local, sandboxed execution layer is a huge piece of the puzzle for a private AI workspace. The `coderunner` tool looks incredibly useful.
A complementary challenge is the knowledge layer: making the AI aware of your personal data (emails, notes, files) via RAG. As soon as you try this on a large scale, storage becomes a massive bottleneck. A vector database for years of emails can easily exceed 50GB.
(Full disclosure: I'm part of the team at Berkeley that tackled this). We built LEANN, a vector index that cuts storage by ~97% by not storing the embeddings at all. It makes indexing your entire digital life locally actually feasible.
Combining a local execution engine like this with a hyper-efficient knowledge index like LEANN feels like the real path to a true "local Jarvis."
Code: https://github.com/yichuan-w/LEANN Paper: https://arxiv.org/abs/2405.08051
To clarify some questions in the comments:
This was a personal account. I normally use my university account for research. I hadn't even used up the $200 free trial, just doing some inference.
I had multiple projects for different AI Studio purposes. Google's automated system likely flagged this as an attempt to circumvent quota limits, even though I was just organizing different use cases into separate projects.
The "2 years" wasn't me ignoring it. I followed up multiple times over this period, but every attempt went nowhere - just automated replies. Eventually I got fed up and decided to post here.
I wasn't paying for support since it was just a personal account for small experiments.
Thanks everyone. The HN visibility definitely helped.