simonmesmith

7 ms·

31.
by simonmesmith 3y ago
Sorry for the delay. I hadn’t thought about that specifically, but I think it’s a good idea. Basically, we can add all kinds of settings and options with the JSON. But I do want to avoid unnecessary bloat, and keep this as simple as possibl
32.
by simonmesmith 3y ago
Agreed. Right now it seems like everyone’s trying to differentiate, e.g. with function calling (OpenAI) and longer context windows (Anthropic). So a universal API wouldn’t work.
33.
by simonmesmith 3y ago
I don’t know of any other models that support function calling like OpenAI’s, unfortunately.
34.
by simonmesmith 3y ago
Great question, and something I’ve thought about. The main issue right now is that I’m relying on the OpenAI API’s function-calling capabilities to enable the use of functions in workflows. If we switch to other LLMs, we’ll need to create s
35.
by simonmesmith 3y ago
Thank you for saying so! It’s very gratifying that I’m not the only one who finds this useful :)
36.
by simonmesmith 3y ago
I definitely plan on adding more functions, and hopefully having others do the same. And I think these should be versatile things like get_url(), which I added today, versus the very specialized plugins that seem to dominate something like
37.
by simonmesmith 3y ago
Yeah, I saw the Magic Loop on HN and was like, “oh no, time to dump my project!” But I think there are some key differences. I’m not sure if Magic Loop is open source, for example, so I don’t know how they’re currently building their workfl
38.
by simonmesmith 3y ago
Perhaps we could give people the option of saving workflows in different places. I went with JSON as I want to make it as easy as possible for people to create new workflows. Plus, they also get versioned easily this way.
39.
by simonmesmith 3y ago
Sure! In Agentflow, you write functions by inheriting from the BaseFunction class. You need to provide the definition in JSON that GPT-3.5/4 uses to understand how to call a function, and also the function logic itself. This just means
40.
by simonmesmith 3y ago
For sure. Let's use an example, like this: https://github.com/simonmesmith/agentflow/blob/main/agentflo... This is a workflow for coming up with a product idea and illustrating it with an image. Thi
41.
Show HN: Agentflow – Run Complex LLM Workflows from Simple JSON(github.com)
51 points by simonmesmith 3y ago|23 comments
42.
by simonmesmith 3y ago
Regarding your last comment, I think it rings true, and further, politics (or at least political communication) is to a degree citizen prompt engineering. It further strikes me that, if these are true: (1) Influencing large groups of people
43.
by simonmesmith 3y ago
It would be interesting to see this chart start much further back, such as when most humans were hunter-gatherers. Yes, working hours per week have declined since the beginning of agriculture, but there is evidence that humans worked much m
44.
by simonmesmith 3y ago
A bit of sleuthing from the company’s website to Wikipedia to Google Patents brought me to this: https://patents.google.com/patent/WO2021084159A1/en It is not peas. It is a specific bacteria. They just talk about
45.
by simonmesmith 3y ago
When I first started experimenting with LLMs, I used LangChain. It helped me get started, but over time I found that: (1) Things were often more complicated than they had to be. For example, creating prompt template objects rather than just
46.
by simonmesmith 3y ago
I agree. I also keep thinking: If ChatGPT (etc.) can write such content, and I'm interested in it, then why won't I just ask ChatGPT for the information directly rather than going to someone's bloated website?
47.
by simonmesmith 3y ago
Based on UK Biobank data and self-reports. Possible reason is: “The mineral lead in instant coffee was more abundant than that in other coffee types, and long-term consumption of instant coffee may result in excessive lead. Additional subst
48.
by simonmesmith 3y ago
TL; DR: “Compared with the omnivorous group, the plant-based diets reduced total cholesterol, low-density lipoprotein cholesterol, and apolipoprotein B levels with mean differences of −0.34 mmol/L (95% confidence interval, −0.44, −0.23
49.
by simonmesmith 4y ago
For what it’s worth, I’ll mention what works for me. I have no interest in any companies or products mentioned below other than using them and finding them useful. I’ve weight-trained for decades and switched up my routine during the pandem
50.
by simonmesmith 4y ago
This is a really simple but elegant idea, and from a few other examples I’ve seen it feels very much like there are lots of fun ideas to explore in the space of flexible GPT-based APIs. Your idea, for example, prompted an idea for a GPT-bas
51.
by simonmesmith 4y ago
I was a communications executive and had been for more than a decade. But I always had a passion for engineering and built many side projects. I decided to pursue that passion and became an entry-level engineer. I was fortunate to have my c
52.
by simonmesmith 4y ago
As a Canadian, I think it’s important to provide some additional context. In Ontario, the province where I live, the provincial government has a near monopoly on alcohol sales through its LCBO stores. These stores bring in well over $2 bill
53.
by simonmesmith 4y ago
Maybe I’m naive here, but whenever I hear about a new AI fake detector, I think about how I would use it in some kind of generative adversarial network or reinforcement learning setup whereby a neural network improves at generating non-AI-s
54.
by simonmesmith 4y ago
Wow. There were 75 articles on the page when I read it. As a former journalist, I would be curious to know what “fact-checked” actually means. I would also be curious to compare these “fact-checked” articles to direct output from ChatGPT. I
55.
by simonmesmith 4y ago
I’ve been experimenting with both ChatGPT and Copilot. I find that Copilot is much more accurate and context-aware, but I can’t use it everywhere—e.g. when working in a shared Colab notebook. Conversely, I find that ChatGPT is better if I’m
56.
by simonmesmith 4y ago
I can relate to this. For me, I think it comes from a combination of wanting meaningful conversation (which doesn’t need to happen too frequently because you run out of meaningful things to discuss), having minimal interest in networking ba
57.
by simonmesmith 4y ago
Having recently seen a paper on the benefit of chain-of-thought prompting, I wondered how this might differ if you added something like “thinking through this step by step…” to the prompts. I tried it with the mountain prompt, as follows: &
58.
by simonmesmith 4y ago
This isn’t exactly what you’re looking for, but a few months ago we had to buy a new toaster oven and I went online to find reviews. They pointed me, surprisingly, to an infrared toaster oven by Panasonic (this one: https://shop.
59.
by simonmesmith 4y ago
This looked interesting and then I read the following in the press release: “Dr Hill said the research sampled 15 participants of varying age, occupational and education backgrounds in England and the Republic of Ireland, who had been put o
60.
by simonmesmith 4y ago
What are these magical factors? “They found that those who were female, married, physically active and not obese and those who had never smoked, had higher incomes, and who did not have insomnia, heart disease or arthritis, were more likely

More ›