sure, but due to more tokens, more parameters, new training methods, or all of the above 😛
LLMS
-
New Supervision Techniques Combined with SFT and RLHF Training
By
–
So, it's some new sort "supervision techniques" > "We trained it using new supervision techniques combined with traditional methods like supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF), similar to those used for GPT-4o."
-
GPT 4.5 criticism noise ignore today TL
By
–
You’re going to see a bunch of people shitting on GPT 4.5 in the TL today. Ignore the noise. Here’s what we wrote about that yesterday in @every
: -
Full GPT-4.5 test: how to use it and what for?
By
–
I just spent the whole day (and a bit of the night) on #GPT4_5 Here is my full test → https://youtu.be/Sf8-3JnWX7Q So what is this new AI on #ChatGPT worth? How to use it properly? What for? I show you EVERYTHING!
-
OpenAI’s ChatGPT: Monetization Strategy and Open-Source Future
By
–
Bullish on OAI! They have a generational product (chatgpt) that is sticky and will keep monetizing more and more! If they came back to the open science/open-source AI side (), they will be a trillion dollar company!
-
Closed AI APIs to disappear within 2-3 years, favoring open models
By
–
Prediction: all closed AI model providers will stop selling APIs in the next 2-3 years. Only open models will be available via APIs. Why? For an open model service, the value prop is clear…it's hard to build a scalable service to access the model and the model is commodity.
-
Run Perplexity R1-1776 Inference on Hugging Face
By
–
You can now run inference on @Perplexity R1-1776 directly from the model page on HF thanks to @FireworksAI_HQ
. Quite cool! -
Sonnet 3.7 vs 4.5: Coding and Creative Writing Performance
By
–
Haven’t done a deep dive yet due to the price tag, but I’d say Sonnet 3.7 still wins for coding tasks. From what I’ve read so far, 4.5 seems far better for creative writing.
-
EQ Impact on AI Inference Reasoning and Literacy Skills
By
–
EQ significantly influences the reasoning process of inference models.
Literacy and expressive skills are among the crucial elements for effective reasoning. -
GPT-4.5 Invents Concepts Mid-Conversation Uniquely
By
–
GPT-4.5 will just invent concepts mid-conversation. No other model does this. Honestly, it’s really cool.