4 ms·
I run a 12GB model on my 3060 and use it to help answer healthcare questions. I'm currently doing a medical residency. (No I don't use it to diagnose). It helps
by rubatuga 2y ago
I run a 12GB model on my 3060 and use it to help answer healthcare questions. I'm currently doing a medical residency. (No I don't use it to diagnose). It helps comply with any HIPAA style regulations. I sometimes use it to fix up my emails. Not sure why people are longing for a 128GB card, just download a quantized model and run with LM Studio (https://lmstudio.ai/ https://lmstudio.ai/). At least two of my colleagues are using ChatGPT on a regular basis. LLMs are being used in the ER department. LLMs and speech models are being used in psychiatry visits.
- ryao 2y agoA 128GB card could run llama 3.1B 70B with FP8. It is a huge increase in quality over what the current 24GB cards can do.