Jump to content

Crono Schism

Resident
  • Posts

    7
  • Joined

  • Last visited

Reputation

3 Neutral
  1. Ahh yes forgot about that requirement. Well for those that have it. I hope they remove that limitation at some point. Otherwise the better route will be to just use a custom trained llm.
  2. I actually spent a lot of time when custom GPT's were first released researching the best way to train them. I created a custom GPT for LSL that is trained off not just public available data such as the wiki itself but updates from the blog, deprecated methods, custom user functions and other sources I myself have collected over the years. I'm not saying it's perfect and even with all the data it might still venture out online and bring back false information. But the amount of testing I've done and with the help of other SL people who have tested this with me it has done 100x better than vanilla gpt 4. Of course feedback is encouraged since all feedback would be used to train it. https://chat.openai.com/g/g-8P0lGqMIy-lsl-guru
  3. lol i completely agree, once i finish this agent I want to share it but I'm thinking I would have to get permission from Linden Labs first to see if it's ok.
  4. That's not true. I started my coding experience in SQL years ago before getting into full stack development. My biggest issue was my memory is garbage at a younger age due to always being in a hurry. I learned to put nodes all over my code as well as add checks on top of checks (and even getting away from over nesting due to thinks like resharper linting does) Like I said this agent is a wip. Currently trying to remove anything deprecated. Also trying to train it to ask for clarification instead of making guesses. (Which it shouldn't do now)
  5. AI but not chatgpt. Using a custom agent trained off LSL data that i've been fiddling with for awhile. Started with full guides for all things LSL but some of the methods were a bit deprecated so been trying to teach it newer methods from the wiki. Hasn't not found a solution yet. Problem with chatgpt is it guesses too much and is unethnical.
  6. Try this key groupKey = "555212fa-dd0a-1960-e694-1da1e6c9a798"; integer gCollapsing = FALSE; default { collision_start(integer num_detected) { // Loop through all detected avatars for (integer i = 0; i < num_detected; ++i) { key detectedKey = llDetectedKey(i); // Check if the detected avatar is an agent and not collapsing if ((llDetectedType(i) & AGENT) && !gCollapsing) { // Check if the detected avatar is in the same group as the groupKey if (llSameGroup(detectedKey) && llGetAgentGroup(detectedKey) == groupKey) { // If in the same group llSay(0, "You are in the same group!"); } else { // If not in the same group llSay(0, "Failed! You are not in the same group."); } } } } }
×
×
  • Create New...