Jean Louis

My card is GTX 1050 Ti with 4 GB VRAM and it is not even listed on HF. So the FLOPs shown on my profile are not true.

Qwen2.5-1.5B runs fast with 47 tokens per second, and I manage to do most of work with it. I need not run models in CPU, even though the 7B version works with CPU and GPU combined.

I am summarizing text, writing stories, agreements, correcting website pages, creating TTS speech, and all for sales and marketing. It works well.

I am contemplating buying GTX 3090 for $800 now. But need better motherboard, PSU, better CPU, 128 GB RAM< and it would cost next $800 as well. Though generating my own images for 1500 website pages is really tempting.