Reply by sightlessHorseman 2024-04-18 06:10:52

  • From: Germany
  • Registered: 2011-12-03
  • Posts: 5,532
  • User Karma: 392

Hi.
Wow thanks, this is pritty great. I had some starting troubles with the API key because I generated a project key and not a user key on the Open AI website, but figured that out and now it works pritty great.
I wonder though, I loaded up my account with 15 dollars, how much tokens does a request usually take? I know that this really depends on the request, but is there an average so I can see how much I am using and adjust my billing acordently?
Also, I might have missed that, but is there a way that I can ask further questions for a prompt, for example if I get a screenshot of a game, I could ask where a certain option is or how a table currently on screen is looking?

Greetings Moritz.

Hail the unholy church of Satan, go share it's greatness.

Quote Post 151

Thumbs up

Reply by ChipsAhoyMcCoy 2024-04-18 09:37:41

  • Registered: 2023-12-01
  • Posts: 124
  • User Karma: 11

You can not currently ask follow up questions, and it seems like the OpenAI cost is still pretty rough. There are more economical models available via the Claude API however that exceed GPT-4V's performance, so I would definitely use those models. Specifically Haiku is faster, better on paper, and signfiicantly cheaper. Like almost 10x cheaper believe it or not haha.

I will say, Haiku can give some interesting responses sometimes, so you kind of have to work with it a little.

Quote Post 152

Thumbs up

Reply by zakc93 2024-04-18 12:58:06 (edited by zakc93 2024-04-18 12:59:15)

  • zakc93
  • Altered being
  • Offline
  • From: South Africa
  • Registered: 2011-04-23
  • Posts: 921
  • User Karma: 136

It's really not that bad. I got 10 US dollars worth of credits in November last year, and I'm sitting with 7.78 now. And I've only ever used the default model. You don't go through it too quickly. Which I'm very happy about since it's a bit pricy with the exchange rate, but so far it's definitely lasting long enough to be worth it.

Quote Post 153

Thumbs up

Reply by Diego 2024-04-21 00:53:10

  • Diego
  • Swamp machine
  • Offline
  • From: Brazil
  • Registered: 2012-06-10
  • Posts: 1,026
  • User Karma: 18

Hey guys!
I'm trying to run llama.cpp on my machine.
I apparently did everything right, when I enter the address http://localhost:8080 in Chrome it opens the llama chat page, but when I try to use, for example, full-screen recognition, NVDA shows the following error:
error: [WinError 10054] An existing connection was forced to be canceled by the remote host
How can I fix it?
And how can I make it always active? If I close cmd, the connection is lost, even in chrome.
Thanks!

Quote Post 154

Thumbs up

Reply by aziz1media 2024-04-21 03:20:00

  • From: inter net
  • Registered: 2023-05-30
  • Posts: 715
  • User Karma: 28

Regarding lama CPP. How many models are there to use with the add-on and where can I get it. Some help would be appreciated

Please consider visiting my Gaming youtube channel my tech youtube channel my Tech Telegram my gaming Telegram channel

aziz1media's Website

Quote Post 155

Thumbs up

Reply by rings2006 2024-04-21 03:24:35

  • From: surrey, british columbia, cana
  • Registered: 2019-11-19
  • Posts: 2,223
  • User Karma: 36

i'm just looking forward to when we can ask followup questions

i am a system, i have headmates, and that is my life, and my discord is rings2006wilson#8609

Quote Post 156

Thumbs up

Reply by sudarshan-chakra 2024-04-25 12:36:49

  • Registered: 2021-09-10
  • Posts: 150
  • User Karma: 11

please add a feature so we can ask followup questions.

Quote Post 157

Thumbs up

Reply by Diego 2024-04-28 02:14:12

  • Diego
  • Swamp machine
  • Offline
  • From: Brazil
  • Registered: 2012-06-10
  • Posts: 1,026
  • User Karma: 18

Hi guys!
After the error that I reported in a few posts ago, I checked the cmd output and it shows the following error:
CUDA error: no kernel image is available for execution on the device
What to do?

Quote Post 158

Thumbs up

Reply by TheTrueSwampGamer 2024-04-28 06:46:15

  • Registered: 2018-06-20
  • Posts: 1,852
  • User Karma: 90

What display adapters do you have in device manager? Sounds like an error with Cuda itself, something which more than likely is some GPU driver error if it isn't an actual traceback error.

Quote Post 159

Thumbs up

Reply by Diego 2024-04-28 07:05:33

  • Diego
  • Swamp machine
  • Offline
  • From: Brazil
  • Registered: 2012-06-10
  • Posts: 1,026
  • User Karma: 18

Intel(R) UHD Graphics 620 and NVIDIA GeForce MX110

Quote Post 160

Thumbs up