iknownthing

8 ms·

61.
by iknownthing 2y ago
Not that I'm aware, I've seen a few apps that do this.
62.
by iknownthing 2y ago
I'm not a railroad expert but I think AI involves much more speculation. They knew what a locomotive could do and what problem it solved. We can't really say that with AI.
63.
by iknownthing 2y ago
Sure but those are just middle-men. What about the cloud providers customers? are they making a profit? Seems like a game of hot potato.
64.
by iknownthing 2y ago
But if enough prospectors fail to find gold the demand for shovels will eventually go down.
65.
by iknownthing 2y ago
The locomotive analogy doesn't quite fit. The customers of the locomotive company would buy the locomotives and then use them to profit. Are Nvidia's customers making a profit from the billions they are spending on GPUs? I don&#
66.
How do you feel about BYOAK LLM Apps?
1 points by iknownthing 2y ago|3 comments
67.
by iknownthing 2y ago
I switched to a lower paying job because the high-paying one was toxic and I had already squirreled away a good chunk of money. I don't regret it. I think of my lower TC as a no-a*hole tax that I happily pay.
68.
by iknownthing 2y ago
Yep, time-boxing is kind of the only solution.
69.
by iknownthing 2y ago
Right but the "how" is the tricky part. For example, wrt to ML, what is the milestone and how do you know you will hit it by a particular date? Very difficult to say with ML.
70.
by iknownthing 2y ago
Agreed, it's very hard, I'd like to see how many orgs have actually done this successfully and what those shippable intermediate products actually were.
71.
by iknownthing 2y ago
It's a catch 22 with ML though. What you wrote is completely true however with ML you cannot say "We will get to 98% precision and 92% recall by Q4". You do not know how long it will take (see self-driving cars). However o
72.
by iknownthing 2y ago
Interesting given how many products out there now are negatively described as ChatGPT wrappers. Sounds like Alexa was just an open source model wrapper.
73.
by iknownthing 2y ago
The sad part is for most of these jobs you actually do only need to know how to use Langchain or AWS services (ok maybe not langchain). The mathematics part is generally just for the interviews.
74.
by iknownthing 2y ago
I assumed it was coming straight from the API because of the token-by-token generation effect but maybe you're right.
75.
by iknownthing 2y ago
Are you paying for the API calls yourself here?
76.
by iknownthing 2y ago
TOS says you cannot "Use Output Data to develop models that compete with OpenAI." You're probably fine but it's close.
77.
by iknownthing 2y ago
> It is trained in a supervised manner on several open LLM datasets, using GPT4 as a judge. Does this mean GPT4 predictions are used as labels? Is that allowed?
78.
by iknownthing 2y ago
This was my experience as well. The only reliable method I found was to use the LLM to generate the field values then put it into a json myself.
79.
by iknownthing 2y ago
Excellent point CamperBob2
80.
by iknownthing 2y ago
which is what?
81.
by iknownthing 2y ago
This may sound harsh but I think some of these researchers have a sort of god complex. Something like "I am so brilliant and what I have created is so powerful that we MUST think about all the horrible things that my brilliant creatio
82.
by iknownthing 2y ago
I was thinking it was because it would be easier than keeping track of usage which I assume you would need to do with a subscription based model i.e. all users using your key.
83.
by iknownthing 2y ago
Curious why you settled on the BYOAK approach rather than a subscription approach
84.
by iknownthing 2y ago
Kind of surprised to hear that given the number of companies I've seen pitching natural language to SQL queries.
85.
by iknownthing 2y ago
Does it use AI?
86.
by iknownthing 2y ago
> It lacked plot, character development, relatibility Kubrick definitely did not play by those rules.
87.
by iknownthing 2y ago
2001 was "not bad"????
88.
by iknownthing 2y ago
Something AI people have known forever is to not use an AI solution where the cost of a misprediction is large.
89.
by iknownthing 2y ago
I'd say just go with GPA.
90.
by iknownthing 2y ago
Looks cool, curious why it's called "wiki" though

More ›