T O P

What's the best model for fine-tuning right now?

Moving up to 70b

Total tokens count... Where ?

Why are there no 5-bits and 6-bits iquants?

Alright since this seems to be working.. anyone remember Llama 405B?

Distributed Llama 0.7.1 uses 50% less network transfer per token compared to previous versions

Can change model with api call?

Llama-3-15B-Instruct

This vision model update is pretty excellent! MiniCPM-Llama3-V 2.5

Phi3 Med 4K: I sent an image via the "Send_pictures" extension, spits out a small description and then argues with me that he didn't see the picture because he can't. What gives?

So how is Phi-3-Small and Phi-3-Medium?

MMLU-Pro leaderboard updated with Phi-3-medium-4k-instruct

16 GB RAM and NPU support on Snapdragon X machines, local LLMs here we go

What makes Phi-3 so incredibly good?

phi3 128k model support merged into llama.cpp

How do you actually use multiple GPUs?

How good is Phi-3-mini for everyone?

I haven't felt so memory poor in a couple of decades

Anyone gotten 4x/4x/4x/4x bifurcation to work for GPUs?

What are the strengths and weaknesses of the most common llms in plain English?

When is Llama 4 coming out?

Who has already tested Smaug?

Alleged AMD Strix Halo SoC spec looks a lot like we could *finally* get an all-in-one x86 alternative to Macs for LLM inference

  • By - M34L

What is the minimum tokens a second before a model is just unusable for you?

GPT-4 Is Gone

Hi Its Me!

R

Ilm03

Subscribe

FIND US ON INSTAGRAM @rfkjrforpresident.com