cyberviser
/

sauron-v4

Model card Files Files and versions

Sauron v4 (LoRA)

Local QLoRA refresh on glasseye RTX 5070 (no cloud GPUs).

Base: mistralai/Mistral-7B-v0.3
Init: cyberviser/sauron-v3
Data: hancock_refresh_v4 (~3917 samples)
Steps: 250 | train_loss ? 1.04

Developed by: cyberviser / GLASSEYE (CyberviserAI)

Downloads last month: 17

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for cyberviser/sauron-v4

Base model

mistralai/Mistral-7B-v0.3

Adapter

(378)

this model