I just could not let this release go by without creating my own little sanctuary of prompts, something about how nano banana pro handles text and tons of detail so coherently really sparks a childlike sense of delight https://github.com/cmd8/awesome-nano-banana-pro-prompts
which UX metrics you’ve personally found the most valuable?
> Copy whatever is already good
it immediately reminded me of Steal Like an Artist. Great advice, and I always forget that sites like Dribbble exist since they’re not usually in my go-to set of tools
Can someone please explain how this works to a software engineer who used to work with heuristically observable functions and algorithms? I'm having a hard time comprehending how a mix of experts can work.
In SE, to me, it would look like (sorting example):
- Having 8 functions that do some stuff in parallel
- There's 1 function that picks the output of a function that (let's say) did the fastest sorting calculation and takes the result further
But how does that work in ML? How can you mix and match what seems like simple matrix transformations in a way that resembles if/else flowchart logic?
The feed forward layer is essentially a differentiable key-value store.
Similar to the attention layer, actually.
So it just uses an attention mechanism like pre-selector to attend to only some experts.
During inference, this cutoff is made a hard cutoff.
This is a very interesting approach. I know it may be too much to ask, but would you suggest any actual practical and hands-on workshops, playgrounds, or courses where I could practice using NN layers for stuff like that? For example, conditional/weighted selection of previous inputs, etc. It feels like I'm looking at ML programming from another angle.
Large Language Models (LLMs) have shown remarkable performance in various basic natural language tasks, which raises hopes for achieving Artificial General Intelligence. To better complete complex tasks, we need LLMs to program for the task and then follow the program to generate a specific solution for the test sample. We propose using natural language as a new programming language to describe task procedures, making them easily understandable to both humans and LLMs. LLMs are capable of directly generating natural language programs, but these programs may still contain factual errors or incomplete steps. Therefore, we further propose the Learning to Program (LP) method to ask LLMs themselves to learn natural language programs from the training dataset of complex tasks and then use the learned program to guide inference. Our experiments on the AMPS (high school math) and Math (competition mathematics problems) datasets demonstrate the effectiveness of our approach. When testing ChatGPT on 10 tasks from the AMPS dataset, our LP method's average performance outperformed the direct zero-shot test performance by 18.3%.
I'm interested as to why it does worse, and some cases much more so, on some problems.
I find myself struggling to connect all of the dots without seeing the entire log. I understand the need to editorialize to show your specific research and implementations. However I cannot fully grok what is being sent to the LLM without seeing an unedited version. It's probably very stupid, but I need to run inferences step by step on LLM prompts to see exactly what is being described.
