Modelpublic

Fabrix-AI-Inc/Argos-2B-LoRA

sourceHugging Faceapache-2.0updated 29d agoView on Hugging Face

0likes39downloads

Model Card

Argos 2B - DPO LoRA Adapter

PEFT LoRA adapter for Argos 2B, trained with Direct Preference Optimization (DPO) for Fabrix.ai / RDAF documentation citation, card-level precision, and custom widget dashboard packs.

—Base Model: unsloth/Qwen3.5-2B (or SFT unsloth_Argos-2B)
—LoRA Configuration: r=16, alpha=32, target_modules: ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"]
—Standalone Merged 16-bit Model: `Fabrix-AI-Inc/Argos-2B`