Soup CLI
Fine-tune an 8B LLM on a 4 GB laptop GPU
LoRA keeps the base model frozen: read, never written. So Soup keeps it in system RAM and streams it into the GPU one decoder layer at a time. Peak VRAM becomes one layer instead of the whole model. Measured on an RTX 3050 Laptop 4 GB: Llama-3.1-8B trains at 119.6 tok/s in 3.32 GB peak. One YAML, one command. SFT, DPO, GRPO, KTO, plus eval, gating and export. Apache-2.0. Every number is published, including the ones I measured and threw away.
AI Analysis
AI analysis not yet available for this product.
Similar Products
Lev8
Find, research, and reach the right people
▲ 451 votes
Auriko
Trading desk for LLM calls
▲ 332 votes
Adapt
The company brain that gets work done
▲ 124 votes
Tapfree for Chrome
Voice dictation that adapts to what’s on your screen
▲ 122 votes
React UI Kit V7
All the chat components you need. None of the complexity
▲ 115 votes
Onpilot
An AI workforce customized to your business
▲ 105 votes