distil-labs/distil-qwen3-1.7b-customer-support-deferral-gguf

Update to iter4 distilled GGUF; results: end-to-end quality + cost framing

Replace placeholder with distilled Qwen3-1.7B BF16 GGUF; real metrics in model card

Add placeholder GGUF (base Qwen3-1.7B Q4_K_M) + model card; weights to be replaced post-training

initial commit