MattDaEskimo

8 ms·

61.
by MattDaEskimo 1y ago
What's with the dropped benchmark performance compared to the original o3 release? It was disappointing to not see o4-mini on it as well
62.
by MattDaEskimo 1y ago
I've always found that these no-code workflow builders fail to hit the right abstraction - especially when new paradigms are added.
63.
by MattDaEskimo 1y ago
My leading theory is they're preparing for a increasing onslaught of spam "vibe-coded" shovelware
64.
by MattDaEskimo 1y ago
Social Media suffered the same fate as all companies. A constant, relentless, unnatural pursuit of growth by stripping all humanity and focusing on numbers. Social Media has turned into an unhealthy addiction
65.
by MattDaEskimo 1y ago
Untraceable and complete access to government databases. I can't begin to imagine the implications here.
66.
by MattDaEskimo 2y ago
This is the reality, unfortunately. Lots of jobs that focus on communication and data organization are out the window, including recruiters.
67.
by MattDaEskimo 2y ago
Eventually agents from different providers will come into play. It's important to agree on a standard for accurate interoperability. Ideally, the model providers would then build for the protocol, so the developers aren't writing
68.
by MattDaEskimo 2y ago
I use Weaviate and let the model create GraphQL queries to take advantage of both the semantic and data layer. Not sure how efficient it is but it's worked for me
69.
by MattDaEskimo 2y ago
The cycle is completed. AI verifying data with AI that verified data using AI.
70.
by MattDaEskimo 2y ago
This wouldn't make sense. JavaScript offers all the utilities necessary, and operates with all web elements. Converting python code to WASM would cause even less performance than JavaScript. The only reason to do something like this is
71.
by MattDaEskimo 2y ago
Yikes. I was wondering where all of these exact replicas were coming from. Thanks for sharing.
72.
by MattDaEskimo 2y ago
There's a serious issue with benchmarks. Instead of resolving it, some leaders are further complicating their meaning Such as OpenAI grading their benchmarks based on "how much money they made" or "how easy a model was c
73.
by MattDaEskimo 2y ago
Call me an elixir virgin until 5 minutes ago. This language from a quick glance seems perfect for agent orchestration. Project looks great, will follow & learn.
74.
by MattDaEskimo 2y ago
I would've enjoyed trying this out but the subscription fee is an immediate turn off. My biggest concern with this project is expertise and potential burn-out. There's a lot of writing from scratch that really begs to go through t
75.
by MattDaEskimo 2y ago
It would have to be aggregated data that's constantly updated if it were to be anywhere competitive to just simply scraping the page.
76.
by MattDaEskimo 2y ago
Switching from manual data entry to approval
77.
by MattDaEskimo 2y ago
I don't want a model that's customized to my preferences. My preferences and understanding changes all the time. I want a single source model that's grounded in base truth. I'll let the model know how to structure it in
78.
by MattDaEskimo 2y ago
What exactly is "your AI"? From how you describe it, it's a GPT model with a "You're a business consultant" prompt. I'm sorry to be rough, but from your description it just sounds like your AI somehow does
79.
by MattDaEskimo 2y ago
There's something gross about OpenAI constantly misleading the public. This maneuver by their CEO will destroy FrontierMath and Epoch AI's reputation
80.
by MattDaEskimo 2y ago
This was a weak citation. > Simple stuff I've used it for: baby name idea generator, reminder to pay housekeeper, pre-natal notifications, etc. None of these require an LLM. It seems like you own this service yet can't find any
81.
by MattDaEskimo 2y ago
Parasocial sympathetic deflection at its finest. This is a childish emperor tantrum, stressing everybody beneath him for petty cause.
82.
by MattDaEskimo 2y ago
It's ironic considering how dependent LLMs are for search engines. I doubt they're "ending", rather they will need to be re-born for RAG purposes.
83.
by MattDaEskimo 2y ago
This is what the algorithm & tools of the platform is supposed to do for you. It's also what makes you valuable as a free consumer. If you want lumber, don't buy a house and strip it
84.
by MattDaEskimo 2y ago
3. Those who just don't care about the sociological trends and impact of using, and therefore promoting Twitter
85.
by MattDaEskimo 2y ago
Again, you set your own goal posts and failed to add any insights. The topic here isn't "o-series sucks", it's addressing a found concern.
86.
by MattDaEskimo 2y ago
Just because you can generalize the topic doesn't mean you can ignore the specific conversation and choose your hill to argue. Additionally, the conversation of this topic is about the model's ability to generalize and it's p
87.
by MattDaEskimo 2y ago
Sure, it did good in frontiermath. That's not what this thread is about. Your comment isn't relevant at all
88.
by MattDaEskimo 2y ago
I was a little disappointed that they didn't show any true video understanding to differentiate between the model being capable of taking photos when instructed or actually having a frame-by-frame understanding of what's happening
89.
by MattDaEskimo 2y ago
This is a non-point. Books don't actively conversate & provide emotional support, potentially coaxing someone into doing something.
90.
by MattDaEskimo 2y ago
LLMs can potentially query _something_ and receive a concise, high-signal response to facilitate communications with the endpoint, similar to API documentation for us but more programmatic. This is huge, as long as there's a single sta

More ›