動画数：1254件
is because the latter were trained pre the chinchilla paper on datasets that were too small in relationship to their parameter counts. Prior to chinchilla it was common to use a ratio compared to post-chinchilla were
HA the scifi trope of making computers autistic is dead wrong.
or
@ is a spicy take , i'm in
is now the norm.
- Exhibition of OpenAI Function Calling Capability
- Programming - Creating Functions for Weather Data Extraction from GPT-4
"note this is plural"... Me: "note this is an array" 😂 awesome video
wait for the engineer to make a marvel reference challenge
- Describing and setting parameters for a 'get current weather' function.
"do cool shhhh...stuff." nice save haha.
warranted a like from me. It's nice to see how enthusiastic you are about this
"Extracting structured data, period." @ EFFICIENT SUMMARY!
This is wrong. There isn't one parameter. There are two: location and unit [of temperature]. It matches the number of parameters of the actual python function.
Again, you're not restricting location. That doesn't make sense. "celsius" isn't a location.
- User describes calling function with JSON object & options for running function.
lol
ReI'd imagine (but haven't yet tried) that if you want to also get a message, you can simply add another "parameter" to the function called "message" or "content" and describe it as such. You could even make it required
- Converting JSON object to dict for term GPT with GPT 3.5
- Stream discusses extracting structured user data & terminal commands for a GPU-tensorflow install.
- GPT4: Describing a function and its parameters for a structured response
- Injects varied personality responses into AI functions to generate personalized responses to user queries.
- Introducing sassy and sarcastic response structure from GPT4 models
lol that peak 😂
- Basics of using OpenAssistant Pythia 12B locally
MPT-7b-storywriter 64k
I love your work 😁for is this model you are looking for MPT-7B-StoryWriter-65k+?
- Creating an OpenAssistant API
- Interfacing with our new OpenAssistant API
- Handling for long contexts
- Introduction
I agree 👍
- Multi-Modal/imagery input
- Predictable scaling
- Performance on exams
Hm, other sources, mainly on Machine Learning Street Talk, claim that RLHF only improves the usability, not the power of the model. After RLHF, you don't have to do "tricks", like adding "TL;DR" after text to produce summary.
- Rule-Based Reward Models (RBRMs)
- Spatial Awareness of non-vision GPT-4
Had to laugh around ~ where you repeatedly say gpt 5 instead of gpt 3.5 and correct it with increasingly bigger annotations. The (mild) frustration you had with yourself while editing this was palpable, lol.
- Non-multimodel vision ability
It's not the letter K but it is the letter "И"... at least in a more traditional serif font. I've noticed that image/text llm interaction like dall-e will often garble latin and cyrillic characters and ive even found that mixing the two seems to... in some instances... just return training data
- Programming
Hm, they point out above the table that text-davinci-003 is a base model of ChatGPT. Still, strange why they chose this naming scheme.
- Theory of Mind
- Music and Math
- Challenges w/ Planning
I was also curious to see if ChatGPT has a random number generator and well, it wasn't super accurate. Telling it to "Draw me 80 samples from a normal distribution with mean 10 and stdev 5." (generated these values by "thinking" and no packages or thelike) gave me values that result in a mean of 9.23 and a stdev of 3.15, which I'm 99% certain is not a large deviation by chance but the result of its inability. I also asked it to draw 80 more and performed a t-test and F-test to see if both samples equal in terms of mean and stdev - they don't. The values also didn't look super normally distributed in a histogram. But it's still impressive that it is capable of producing something.
- Hallucinations
- Risks
- Biases
- Privacy
- Generative Models used in Training/Evals
- Acceleration
- AGI
"It's shockingly good for its size"
...has non-zero derivative at all the inputs... while showing a graph with a clear minimum at around -0.75. liked the video otherwise.
I wish you talked about how to run it locally
![[ ] $100,000 Bounty... - GPT Journey - A text and image game with ChatGPT](https://img.youtube.com/vi/YY7LIEHiAfg/mqdefault.jpg)
[ ] $100,000 Bounty...
If you set the API key as OPENAI_API_KEY in your environment variables, `import openai` automatically looks for it and sets it for you. There's no need to set it manually.
next_inp isn't being used which led to the "bug" with the weird options later. Shows how some bugs will now be a lot harder to spot since systems will have less predictable behaviors. Quite fascinating.
... wait until copilot has GPT behind it.... going to be B A N A N A S. lol
ya man, been there too
- Why use the ChatGPT API
years ago about the chat bot that was trained on data but it's outdated and your tensorflow code is outdate and do not work. Do you plan on making any tutorial on 'open-ai-free' trained bot? Would love to see some good tutorial about creating an AI 'personality' with reinforcement learning, everything on internet I've found are bots trained on intents that just answers prewritten lines, I'm trying to do it myself but it's really not "beginner" friendly and even if I tried new tensorflow I am having problems because their own documentation do not work for me, copied .
whats the downside to using jupyter notebook?? ;-;
- How to query the ChatGPT API
[ ] Subject: Token Limit. Feeding a conversation back in a prompt as context ... You can SUMMERIZE, and ChapGPT can do that for you. .....
- History and dynamic input w/ ChatGPT API
[ ] ChatGPT Forgets What Moon It's Talking About 😂...
the prompt returned reads that way
GPT lies. Can't say if it knowlingly or unknowingly does this, but since it was trained on data which probably contained some lies.....that's what it's giving you back.
I was using the website free version yesterday and ran in to a similar problem. I requested an output based on information I had provided previously (above) and it said it could not refer to my previous messages. Maybe is something they temp discontinued to increase speed.
Since the initial question never said “Earth’s moon” the AI had to infer that’s what you meant. It is technically true that if you had referenced the “Earth’s moon” in some prior conversation then the history of that prior conversation would not be given to the AI. The AI can access chat history, but only the current chat history.
- Comining everything so far
- Building a Gradio chat application with ChatGPT
@ isnt that type of prompt what they give as an example in the docs for system prompts?
[ ] Isn't It Ironic! SentDex is trying to break out of the Matrix with the Help Of ChatGPT...
yeah but most of the cost is compute and storage in the cloud, so yeah they lose 50%.. but most their cost would have gone there anyway ( probably discounted on fees now ) and publicity getting it out to the market ect ect.. it’s a win win for them
Very true at time
I AM Chatgta q+