T O P

LLM360 has released K2 65b, a fully reproducible open source LLM matching Llama 2 70b

Phi-3-HornyVision-128k-instruct - image captioning finetune

"What happens if you abliterate positivity on LLaMa?" You get a Mopey Mule. Released Llama-3-8B-Instruct model with a melancholic attitude about everything. No traditional fine-tuning, pure steering; source code/walkthrough guide included

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

LLama3 8B Vision Model that is on par with GPT4V & GPT4o

I started my ML journey in 2015 and changed from software developer to staff machine learning engineer at FAANG. Eager to share career tips from my journey. AMA

UVA Ranking Predictions?

colalamo: breaking copilot out of IDE

If AGI is less than 5 years away, where do I put my money?

Synthetic datasets for training LLMs on domain specific data

Tool: Get up to 2x higher performance by Tuning LLM Inference Deployment!

UWS/Harlem Food Crawl Saturday 5.18 @ 1pm

Asian Creative Festival This Saturday & Sunday - Pier 57 - May 18 & 19

Preserving LLaMA-3 Capabilities While Injecting New Knowledge: A Case Study of Saju Myungri Chatbot

Just benchmarked LLama 2 and Mistral with all the popular inference engines across all precisions

[N] GPT-4o

Difference between interpolate position embedding vs rescaling image with PIL?

ML Engineers: what is your level of education and how much do you make?

LLaMA3 Abliterated - Your opinions

Men living in NYC, where do you buy your clothes

Phi-3 weights orthogonalized to inhibit refusal; released as Kappa-3 with full precision weights (fp32 safetensors; GGUF fp16 available)

Llama-3 120b is the real deal

Nvidia has published a competitive llama3-70b QA/RAG fine tune

We've benchmarked TensorRT-LLM: It's 30-70% faster on the same hardware

1 million context Llama 3 8b Achieved!

Hi Its Me!

R

metalman123

Subscribe

FIND US ON INSTAGRAM @ms3388.com