fzliu

41 ms·

781.
Milvus HNSW Exploration(farfetchtechblog.com)
1 points by fzliu 4y ago|0 comments
782.
Pay Attention to MLPs(arxiv.org)
3 points by fzliu 4y ago|0 comments
783.
Implementation of the Undocumented Twitter API(github.com)
2 points by fzliu 4y ago|0 comments
784.
GPTCache: Semantic Cache for LLMs(github.com)
2 points by fzliu 4y ago|0 comments
785.
Everything I learned about accidentally running a successful tech conference(vickiboykis.com)
1 points by fzliu 4y ago|0 comments
786.
by fzliu 4y ago
Keep up the awesome work. I've run across this problem myself - I somehow used $20 just testing a small demo I made with GPT-3.5. As most ML is inherently probabilistic, it seems reasonable to make an LLM cache both semantic and _stoch
787.
by fzliu 4y ago
Computer vision seems to be gravitating heavily towards self-attention. While the results here are impressive, I'm not quite convinced that vision encoders are the right way forward. I just can't wrap my head around how discretizi
788.
What Is a 9.0 Earthquake?(opb.org)
1 points by fzliu 4y ago|0 comments
789.
by fzliu 4y ago
I think the core idea is strong, but it's missing several features that could differentiate it as a product: 1) adding an "advanced" tab which enables the user to input more information about the document, e.g. formatting, le
790.
by fzliu 4y ago
This is the path forward that I see for Langchain commercialization as well. As we begin to see more and more products built atop LLMs, these hosted services could form the backbone for a lot of those applications. Disclaimer: We - Milvus (
791.
by fzliu 4y ago
I did 18/6 for quite a while before switching back to 16/8 due to a busy work schedule. IF is a long-term play - having the occasional burger or some beers late at night on Friday or Saturday won't affect your results much. E
792.
Show HN: Enhancing GPT with domain knowledge using vector databases(osschat.io)
12 points by fzliu 4y ago|1 comments
793.
Mass Misconception(bigthink.com)
1 points by fzliu 4y ago|0 comments
794.
by fzliu 4y ago
Even though GPT-4 "fixes" many of the problems pointed out in this blog post, I question whether or not some of the underlying issues associated with transformers (and the attention mechanism in particular) can be overcome with wi
795.
by fzliu 4y ago
I was a bit skeptical about loading a _4GB_ model at first. Then I double-checked: Firefox is using about 5GB of memory for me. My current open tabs are mail, calendar, a couple Google Docs, two Arxiv papers, two blog posts, two Youtube vid
796.
by fzliu 4y ago
Vector indexes only form a component within vector databases - you'd want the database bits and bobs (scalability, replication, caching, etc) surrounding it in addition to the index itself. Milvus, for example, supports flat indexes -
797.
by fzliu 4y ago
This is typically done by taking the activations from a neural network. Activations from layer depths are generally preferred since they are more representative.
798.
by fzliu 4y ago
The design of Milvus 2.x follows this paper we published a while back: https://arxiv.org/abs/2206.13843 . In short, we used Pulsar to implement the write-ahead log, which provides coordination and a single source of tru
799.
by fzliu 4y ago
I'm curious about this as well. There are potentially many different versions of embedding models used in production and correlating different versions together could be very important.
800.
by fzliu 4y ago
I've done some quick-and-dirty testing with OpenAI's embedding API + Zilliz Cloud. The 1st gen embeddings leave something to be desired ( https://medium.com/@nils_reimers/openai-gpt-3-text-embedding... ), but t
801.
by fzliu 4y ago
Faiss is a wonderful vector search library - in particular, the ability to do hybrid indexes e.g. IVF-PQ, IVF-SQ is great. We ( https://milvus.io ) use it as one of the indexing options (along with Annoy, Nmslib, and DiskANN) to p
802.
by fzliu 4y ago
Beyond artificially increasing prompt length, vector DBs also allow you to inject domain knowledge into autoregressive LMs.
803.
Open Source Alternative to Copilot(tabbyml.com)
2 points by fzliu 4y ago|0 comments
804.
by fzliu 4y ago
Great to hear that you're considering Milvus. Feel free to reach out if you ever have any questions/comments/concerns. Just took a look at your docs and product page as well. Keep up the great work!
805.
by fzliu 4y ago
Redis does fairly well single-threaded - iirc it does better than Milvus. The problem is, once you move to larger applications and want higher performance, its architectural limitations make it impossible to scale.
806.
by fzliu 4y ago
Totally with you on that - it's annoying to need multiple pieces of infrastructure, especially for solo developers or small teams. Often times, you want to filter further based on scalar fields/metadata such as timestamps, strings
807.
by fzliu 4y ago
One key differentiator of Milvus is the ability to use a variety of different indexing algorithms. Some will consume very little memory and provide significant speedup at the expense of recall, while others are more powerful at the expensiv
808.
by fzliu 4y ago
Traditional search such as ES and Lucene rely primarily on bag-of-words retrieval and keyword matching e.g. BM25, TF-IDF, etc. Vector databases such as Milvus allow for _semantic_ search. Here's a highly simplified example: if my query
809.
by fzliu 4y ago
Yup! Zilliz Cloud (managed Milvus) provides free credits for you to get started with a full-service (no throttling or other limitations) PoC as well.
810.
by fzliu 4y ago
Milvus is completely open source ( https://github.com/milvus-io/milvus ) and supports a variety of index types ( https://milvus.io/docs/overview.md#Index-types ) and support various consistency levels

More ›