Retrain with flan-t5-base (248M): difficulty accuracy 93.8%, eval_loss 2.228 (upgrade from flan-t5-small 80M)

Upgrade to flan-t5-base (248M): difficulty_accuracy=94.8%, eval_loss=2.228

Upgrade to flan-t5-base (248M): difficulty_accuracy=94.8%, eval_loss=2.228

Upgrade to flan-t5-base (248M): difficulty_accuracy=94.8%, eval_loss=2.228

Update model card with final training metrics

Add special_tokens_map.json

Add SentencePiece model for T5Tokenizer compatibility

Add model card

Upload Flan-T5-small passage difficulty scorer & simplifier

initial commit