HNHacker News
TopNewBestAskShowJobs

natsucks

189 karma · joined July 8, 2023

submissionscomments
natsucks··on RAG is more than just embedding search
Author has a good point, but no mention of hybrid search?

...not that hybrid search solves everything.

natsucks··on Faster Than Llama 2: DeciLM LLM with Variable GQA
curiously, they don't mention MMLU, which is one of the most commonly reported metrics for comparison.
natsucks··on Falcon 180B
exactly. i have a hard enough time running a 70b model at fp16. no way i attempt 180 without a significant performance gain.
natsucks··on A Deep Dive into Weaviate's Fusion Algorithms for Hybrid Search
that would be great. thanks. will do.
natsucks··on A Deep Dive into Weaviate's Fusion Algorithms for Hybrid Search
Just migrated from Supabase + pgvector to Weaviate hoping to take advantage of langchain.retrievers.weaviate_hybrid_search.WeaviateHybridSearchRetriever. Unfortunately i can't get it work asynchronously :(
natsucks··on Mastering Emacs
i'll take this banter. it's fun. long live emacs.
natsucks··on Mastering Emacs
The native bindings are...not great. Personally i use xah-fly-keys, but evil (vim) keybindings are popular.
natsucks··on Ask HN: What's the biggest red flag you've encountered during a hiring process?
hilarious.
natsucks··on Face it, self-driving cars still haven’t earned their stripes
I wonder if this is a good application for recently developed multimodal/vision-language models that demonstrate "emergent" properties like an ability to classify objects unseen in their training data.
natsucks··on The ReMarkable Streaming Tool v2: Elevating Remote Work Efficiency
As a reMarkable owner, I love it. I want it. gimme gimme gimme.
natsucks··on You probably don’t need to fine-tune an LLM
- After seeing how merely quantizing a model can make it go berserk, I have very little confidence I can fine tune an LLM and expect similar performance benchmarks.

- A RAG-empowered LLM can tell you where the knowledge used to answer a question came from.

natsucks··on You probably don’t need to fine-tune an LLM
not to mention staying true to their benchmarks...
natsucks··on Why host your own LLM?
Right, especially when it's not clear if such a machine will be needed a year or two from now to run a model with the equivalent performance.
natsucks··on Important Coding Habits
I'm hoping that a potential transition from monitors to AR glasses will substantially alleviate some of these issues.
natsucks··on Container Training
One does not simply deploy...
← PreviousPage 4 of 4