Hmm. Given most LLMs are trained on OAI outputs, those can all share prompts. Then we have Claude, basically can switch out things like “` for . If there's some room for latency, we can train models for this.
LLMS
-
Understanding System Prompts in AI Language Models
By
–
That assumes that it knows what a system prompt is, as opposed to knowing that when people ask about system prompts the response should look something like what it output there
-
GPT-4 Level Model Makes Unusual Python Installation Error
By
–
It did just suggest I "pip install sqlite3" in a Python prompt I ran through it though, which is a new mistake that I've not actually seen from other GPT-4 level models
-
LLMs are poor judges of ASCII art, votes should be excluded
By
–
Human preference LLM arenas are poorly suited for evaluating ASCII art because the ASCII art that most impresses a human is often verbatim regurgitation of an existing human work and this is rarely true for text. Votes on ASCII art should be detected and thrown out IMO.
-
AI Model Hallucination-Free Responses Challenge
By
–
I don't see any hallucinated details in here at all, which never happens with these kinds of ego-prompts for me
-
Training Data Contamination and Model Self-Knowledge Limitations
By
–
Loads of models do that, because their training data included examples of text that included ChatGPT output Directly asking a model questions about itself remains a surprisingly ineffective way of learning anything useful about the model
-
New GPT2 Chatbot Impresses with Better Search Results
By
–
You can try out the mysterious gpt2-chatbot at https://
chat.lmsys.org (select "Direct Chat" and pick it from the menu) Initial impressions: I'm very impressed. It gave me a better answer for an ego search ("Who is Simon Willison?") than any other model I've tried -
ChatGPT Memory Feature Now Available for Plus Users
By
–
Memory is now available to all ChatGPT Plus users. Using Memory is easy: just start a new chat and tell ChatGPT anything you’d like it to remember.
— OpenAI (@OpenAI) 29 avril 2024
Memory can be turned on or off in settings and is not currently available in Europe or Korea. Team, Enterprise, and GPTs to come. pic.twitter.com/mlt9vyYeMKMemory is now available to all ChatGPT Plus users. Using Memory is easy: just start a new chat and tell ChatGPT anything you’d like it to remember. Memory can be turned on or off in settings and is not currently available in Europe or Korea. Team, Enterprise, and GPTs to come.
-
Constrained Sampling Forces Model Function Calls
By
–
Constrained sampling to ensure that a model calls a function:
-
Emergent Abilities Plotted Against Pretraining Loss in New Paper
By
–
Enjoyed this paper that plots emergent abilities with pretraining loss on the x-axis, which is actually a suggestion that @OriolVinyalsML also made a few years back: https://
arxiv.org/abs/2403.15796 The paper uses intermediate checkpoints to plot a variety of pretraining losses. For some