Card: declare base_model_relation: quantized

Thanks for the same-morning merge of my card PR #3 here.

This adds one line to the card's metadata header, base_model_relation: quantized, so the model tree on https://huggingface.co/google/efficientnet-b7 lists this repo under Quantizations instead of Finetunes. I added it so that anyone browsing the base model for a LiteRT build, a developer or a coding agent, finds it there. The int8 weight-only variant here makes quantized the closest relation value. Cards such as https://huggingface.co/litert-community/Qwen3-4B and https://huggingface.co/litert-community/Llama-3.2-1B carry the same line, and on Sep 4 I added it to the 123 litert-community repos I created.

One note on this repo's history: that Sep 4 pass also pushed the line here directly, because it selected repos with any commit of mine instead of repos I created. I reverted it 3 h 41 min later, so both Sep 4 commits here are mine and the card is back at its previous revision. Sorry for the noise. This is the same line sent as a PR, as my earlier changes here were; whether it goes in is entirely your call.

README.md

@@ -10,6 +10,7 @@ datasets:
   - imagenet-1k
 base_model:
   - google/efficientnet-b7
+base_model_relation: quantized
 model-index:
   - name: litert-community/efficientnet_b7
     results: