Will we ever get new Opuses and Ultras of the world or is inference-time compute for the rest of our days? I want to talk with masters of language and philosophy, benchmarks be damned.

Just for kicks I looked at the newly released dataset used for Reflection 70B to see how bad it is...

Even OpenAI is excited about the Llama3 400B release

An in-depth look at locally training Stable Diffusion from scratch

Discover AstraQuasar-4B: a NEW LaMA-based arch | First training implementation of the self layer calling (Duplicate Trick)

Chatbot Arena Leaderboard Update: Qwen1.5-72B becomes #1 non-proprietary model by sizeable margin

With v62 you can no longer put a browser screen/movie on a wall 10ft away. Why have they done this?

Emad's comments regarding what they have to compete with Sora. Thoughts?

How to use?

Sora by openAI looks incredible

Unpopular Opinion: All these small open-source foundational models coming out are not moving us forward. To truly rival closed-source, we need models with 100+ billion parameters.

Any prompt experts know how to force a model to not ask a question at the end of every response in a chat conversation?

Compiling a list of all the subscription services that give a discount when you threaten to cancel

Pyrolytic graphite can levitate over a magnet (see video) due to diamagnetism so can anyone explain how LK-99 partial levitation proves anything?

New Video from Russian Academy of Sciences?

[D] Since Google buried the MMLU benchmark scores in the Appendix of the PALM 2 technical report, here it is vs GPT-4 and other LLMs

Added ChatGPT to Automatic1111

Introducing Macro Diffusion - A model fine-tuned on over 700 macro images (Link in the comments)

100 to 200 epochs comparison please let me know your thoughts - dreambooth - overtrained?

Hardware for running stable diffusion? Laptop recommendations? Generally have no idea where to start

Struggling with Training, Will Pay For Guidance

Firefighters @Dr__Macabre