pamelafox

10 ms·

91.
by pamelafox 2y ago
Indeed! My nanny does have kids of her own. So I would either pay her overtime or I would find additional childcare for the 5th day. I'm just pointing this out because I hear "4 day workweek" talked about a lot, but it's
92.
by pamelafox 2y ago
" It separately announced another policy that will allow parents with children in grades one to three in elementary schools to trade off a bit of their salary for the option to clock out early." This is really nice, this was the o
93.
by pamelafox 2y ago
Would schools and childcare providers also have 4-day workweeks? I really appreciate that I have 5 days of childcare now, as I find it exhausting to be 24/7 parenting two lil kids for two days straight (the weekends). If the norm was 4
94.
by pamelafox 2y ago
I’ve got this RAG repo working entirely locally (Ollama/Postgres) but it doesnt RAG on documents like you want. https://github.com/Azure-Samples/rag-postgres-openai-python I’d like to make that version when I have
95.
by pamelafox 2y ago
How does your chunker actually work? The gist only shows an API call. Or is that secret?
96.
Integrating Vision into RAG Applications(blog.pamelafox.org)
26 points by pamelafox 2y ago|0 comments
97.
Making a RAG app work with Ollama models(blog.pamelafox.org)
3 points by pamelafox 2y ago|0 comments
98.
by pamelafox 2y ago
I grappled with that issue for https://github.com/pamelafox/openai-messages-token-helper as I wanted to be able to use it for a quick token check with SLMs as well, so I ended up adding a parameter "fallback_to_de
99.
by pamelafox 2y ago
I have a package here that includes calculation for images for OpenAI: https://github.com/pamelafox/openai-messages-token-helper
100.
by pamelafox 2y ago
I have a package here that includes calculation for images for OpenAI: https://github.com/pamelafox/openai-messages-token-helper
101.
by pamelafox 2y ago
Are you also accounting for costs of sending images and function calls? I didn't see that when I looked through the code. I developed this package so that I could count those sorts of calls as well: https://github.com/p
102.
by pamelafox 2y ago
I would love to see them try this with the Processing/P5Js libraries. Thats what the ASTs reminded me of. It could potentially be used to help students trying to figure out how to fix their programs. I used AST-based hints for my Proce
103.
by pamelafox 2y ago
Re 1) pgvector has an example in the repo that uses a model for re-ranking: https://github.com/pgvector/pgvector-python/blob/master/exam... I'm not using that in my own experiments since I don'
104.
by pamelafox 2y ago
I commented above with a pgvector example that does it- https://news.ycombinator.com/item?id=40527925
105.
by pamelafox 2y ago
If you're looking for an example of RRF + Hybrid Search with PostgreSQL, I've put together a FastAPI app here that uses RAG with those options: https://github.com/Azure-Samples/rag-postgres-openai-python/
106.
by pamelafox 2y ago
It looks like the supported vector DBs are Pinecone and Astra. Have you looked into Postgres with pgvector? I’ve started experimenting with building RAG flows for pgvector, works fairly well.
107.
by pamelafox 2y ago
I just tested out using GPT-4o instead of gpt-4-turbo for a RAG solution that can reason on images. It works, with some changes to our token-counting logic to account for new model/encoding (update to latest tiktoken!). I ran some spee
108.
by pamelafox 2y ago
Those are typically called “stop words” and the article removed them for the example. Many NlP algorithms do remove stop words as a first step.
109.
by pamelafox 2y ago
It actually seems to imply the opposite: “ As document length increases, the complexity of the text also increases, rendering cosine similarity less effective in capturing nuanced semantic relationships” I think its true that using TfIdf to
110.
by pamelafox 2y ago
Hm, that article says cosine similarity ranges from 0 to 1, but it can range between -1 to 1 if the vectors are allowed negative components. Also FYI to any Postgres users: the pgvector operator is for cosine distance , which is 1 - cosine
111.
by pamelafox 3y ago
I’ve also seen that GPTs struggle to admit when they dont know. I wrote up an approach for evaluating that here - http://blog.pamelafox.org/2024/03/evaluating-rag-chat-apps-c... Changing the prompt didn't hel
112.
by pamelafox 3y ago
Lol, nice truncation logic! If anyone’s looking for something slightly fancier, I made a micro-package for our tiktoken-based truncation here: https://github.com/pamelafox/llm-messages-token-helper
113.
by pamelafox 3y ago
This would actually be very helpful for improving your lighting when webcamming in your home office, since you could rotate the house until you achieved optimal lighting for that time of day. Overkill, but that's how I'd use it!
114.
by pamelafox 3y ago
Thanks! I havent had to use a syntax-enforcing framework with gpt-35, I’ll try outlines and guidance out to see if they help enforce syntax for the locally runnable models.
115.
by pamelafox 3y ago
Very interesting! Can you see those helping for RAG scenarios? Specifically: - decreasing models tendency to answer with ungrounded answers - increase models ability to respond with the correct syntax for citations- the open models like lla
116.
by pamelafox 3y ago
I like it! I wish there was a clear toggle to completely disable audio input/output. I’m trying to practice while baby naps on me, so I can’t say or hear.
117.
by pamelafox 3y ago
API wise, it looks very similar to the OpenAI python SDK but not quite the same. I was hoping I could swap out one client for another. Can anyone confirm they’re intentionally using an incompatible interface?
118.
by pamelafox 3y ago
Hm we typically talk about that phase as “data ingestion”. We also use verbs extract, chunk, index, but those are parts of ingestion. Another phrase we use is “data preparation”. The tricky thing is that your train() method is on the vanna
119.
by pamelafox 3y ago
I love that this exists but I worry how it uses the term “train”, even in quotes, as I spend a lot of time explaining how RAG works and I try to emphasize that there is no training/fine-tuning involved. Just data preparation, chunking
120.
by pamelafox 3y ago
How does this compare to llamafile? I recently started experimenting with pointing my openAI web apps at local llamafile and its pretty straightforward: http://blog.pamelafox.org/2024/01/using-llamafile-for-local-.

More ›