Modelpublic

rocky232/qwen3-0.6b-atlas-email-gguf

sourceHugging Faceupdated 6mo agoView on Hugging Face

0likes21downloads

Model Card

Qwen3 0.6B Atlas Email Assistant (GGUF)

Fine-tuned Qwen3 0.6B for Atlas Email Assistant — intent classification and tool routing.

Converted from rocky232/qwen3-0.6b-atlas-email (SafeTensors fp16).

File	Quant	Size	BPW	Use Case
qwen3-0.6b-atlas-email-q4km.gguf	Q4KM	462 MB	5.09	Desktop CPU default
qwen3-0.6b-atlas-email-q5km.gguf	Q5KM	526 MB	5.81	Higher quality option

—Base: Qwen/Qwen3-0.6B
—Fine-tuned with QLoRA (rank 16, alpha 32, 3 epochs)
—460 training examples (email intents + conversational)
—11 intents: searchemails, listemails, composeemail, listcalendarevents, createreminder, listreminders, createtodo, listtodos, staremail, getbriefing, generalchat

Apache 2.0 (Qwen3)