llmllmllm

8 ms·

1.
by llmllmllm 2y ago
The UK doesn't have exit checks.
2.
Mkcontext – Generate ChatGPT prompts from files(github.com)
1 points by llmllmllm 3y ago|1 comments
3.
by llmllmllm 3y ago
I just made this after realising that chatgpt 4 now has 32k input context so it's useful to inject multiple files into my prompts easily
4.
by llmllmllm 3y ago
"Show HN is for something you've made that other people can play with. HN users can try it out, give you feedback, and ask questions in the thread." We made it, you can play with it :)
5.
by llmllmllm 3y ago
This is an interface that has GPT-4 producing content while also producing images using DALL-E 3, we have integrated them.
6.
by llmllmllm 3y ago
Feedback is great, thank you. I linked to the shared report so that it's obvious what is actually being produced, rather than just a description of it. This is just an example summary of a single web page, you could imagine it producin
7.
by llmllmllm 3y ago
While this makes some of what my startup https://flowch.ai does a commodity (file uploads and embeddings based queries are an example, but we'll see how well they do it - chunking and querying with RAG isn't easy to do
8.
by llmllmllm 3y ago
I have found success with chunking with 100 tokens, preceeded by the last 10 tokens of the previous chunk, and the first 10 tokens of the next chunk, 120 tokens total. I generate an embedding for each, then compare that to embedding(s) deri
9.
by llmllmllm 3y ago
Things like this needing to exist are one of the reasons we are working on https://flowch.ai/?ref=hn - the ChatGPT interface is really sub-optimal for work and keeping track of interactions over time. We have been building
10.
Show HN: AI Analysis of Bryan Cantrill's Rust Talks [video](vimeo.com)
3 points by llmllmllm 3y ago|2 comments
11.
Show HN: ChatGPT Powered Code Analysis(flowch.ai)
5 points by llmllmllm 3y ago|0 comments
12.
by llmllmllm 3y ago
I'll be curious to see if many people end up subscribed to both this and ChatGPT Pro, are they different enough to both be worth it?
13.
by llmllmllm 3y ago
Very fair, we have demo videos, guides etc planned for the next week or so. As it's a tool that can do many things it's hard to describe. Still a lot to do :) In terms of what makes the report different from Bing: this could be an
14.
by llmllmllm 3y ago
Interesting that they're still centered around Chat as the interface, with https://flowch.ai (our product) we're building it much more around projects and reports, which we think is often more suitable for businesses.
15.
by llmllmllm 3y ago
Learning to leverage LLMs as part of software systems (distinct from LLMs for writing code, that'll be common soon as people catch on and it becomes acceptable to most).
16.
by llmllmllm 3y ago
As part of my testing of https://flowch.ai (which we are developing), I've been uploading multiple PDFs (including scientific papers). I then ask a question and ask it to cite its sources as part of the prompt and it'l
17.
by llmllmllm 3y ago
Its integration with public and private data along with layers of automation allowing its use way beyond chat bots. From individual tool to something that can power large parts of a business. https://flowch.ai - Currently free t
18.
by llmllmllm 3y ago
The open source solutions have mental overhead and potentially long term maintenance hassle. Even for people who could set it up, for many $20 is worth not having to deal with it.
19.
by llmllmllm 3y ago
I want to search based on semantic meaning, rather than partial or exact keyword search. I'm using pgvector within postgres, so I get all of the postgres benefits.
20.
by llmllmllm 3y ago
GPT-4 via ChatGPT, mostly for coding and GPT-4 via https://flowch.ai (our tool) for anything that involves my own data (including my own code), or where I want automation, such as running a prompt on many pieces of data. Overall
21.
by llmllmllm 3y ago
As well as heavy LLM use, we're also using vectors (with pgvector) for searching text content which is displayed in a table. Works very well.
22.
by llmllmllm 3y ago
https://flowch.ai (our project) does this and is currently free to use. It doesn't train a custom model but it gets good results. A simple way to do this is to upload your files (PDFs, Wod docs, virtually any type is suppor
23.
by llmllmllm 3y ago
A lot of LLM libraries seem incredibly overhyped. I've found that I'm able to do a lot with the underlying API calls. If you're comfortable with doing API calls in your own code, I'd suggest trying to prototype your desi
24.
by llmllmllm 3y ago
Thanks for the feedback, will iterate and expand on it.
25.
by llmllmllm 3y ago
https://flowch.ai is live and is a hit with early users. As users supply their own prompts regressions in LLMs aren't really an issue, we're currently using both GPT3.5 and GPT-4, both have their place. Rapidly moved f
26.
by llmllmllm 3y ago
I took the context from those two sources (copy and pasting the text) and it was interesting to see the difference: GPT 3.5 (fails): https://flowch.ai/shared/90274999-8eaf-4046-9115-7f623264197... GPT 4 (succeeds): ht