How To Use GPT-o1 Preview (o1- Preview Tutorial) Complete Guide With Tips and Tricks

How To Use GPT-o1 Preview (o1- Preview Tutorial) Complete Guide With Tips and Tricks

Links From Todays Video:

00:00 - Introduction to OpenAI's O1 series models
00:27 - Comparison between O1-preview and O1-mini
01:00 - Unique features of O1 models, including reasoning tokens
01:24 - Limitations of O1 models in beta
01:43 - OpenAI's advice on prompting for O1 models
02:57 - Keeping prompts simple and direct
03:58 - Avoiding chain of thought prompting
05:09 - Using delimiters for clarity in prompts
06:27 - Limiting context from external sources
07:58 - Comparison between O1-preview and O1-mini
10:01 - Use cases for O1-mini
11:19 - Examples of using the models
13:26 - Coding example: Snake game in Python
14:46 - Note on model's reasoning token

Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos.

Was there anything i missed?

#LLM #Largelanguagemodel #chatgpt
#AI
#ArtificialIntelligence
#MachineLearning
#DeepLearning
#NeuralNetworks
#Robotics
#DataScience

[applebeliebes]

Just started testing out oh one preview, and I’m finding that it does really well with creative writing. My creative writing prompts are very complex with a lot of details and it handles it really really well so far.

[grim]

Dude, this was super helpful. I have been over describing and getting nothing really of what I wanted. Idk why I never thought of stuff like this got my sub.

[Ignitus]

In the first example, GPT-4o return countries with 'a' in the [3] index. While it's not what you intended, it can be easily avoided by properly understanding your tool and how to interact with it.

Same goes for the oft-cited "which is bigger: 9.9 or 9.11?" issue. The question is ambiguous and either answer can be correct, depending on context. In math 9.9 > 9.11 while in version control 9.9 < 9.11.

Problems due to ambiguity are often the fault of the user. You shouldn't expect the model to read your mind and often it's not desirable to let the model make assumptions, even when they're "obvious" ones.

[kecksbelit]

I find the o1 models to be great for creating databases and find possible errors in them

[dzagario]

what is the name of the background style you have on the video

[MasamuneX]

first use 4o then after you have some stuff done use o1 to save on prompt counts

[sunkanmibero]

Lovely breakthrough. Truly helpful. You gota sub 👍🏾

[ClassicRiki]

3:57 It’s not counterintuitive, it’s straight up stupid…if you want to specify whether to go into detail or keep it simple is clearly a user preference, it’s nothing to do with people not knowing how to write that particular prompt. 7:40 while I understand the impressive capabilities of o1, it’s tricky not to feel rather unimpressed when they have to say “don’t ask this, or that”

[GH-uofy]

But can you upload source codes, this would be awesome for debugging and optimization.

[RanHab]

guys i'm just starting out as an AI enthusiast,
would love your feedback as i make similar stuff!

[gammaraygem]

I like the counterintuitive way of prompting. It helps to remind that is a piece of tech, not something "alive". I doubt that that phase will last long and then we ´re back to our living "pet rocks".

[filbertneon]

Please don't say "It thinks" We need to get this right from the start. These things DON'T THINK! They calculate and they should always be referred to in a non-human way. Do not spread human ignorance, please.

[GursimarSinghMiglani-ymnu]

Disappointed, I was not stunned in this video

[ilyass-alami]

Is there a way to use it for free please?

[AinHab]

guys i'm starting out as an AI enthusiast,
would love your feedback as i make similar stuff!