Restore tokenizer.json from google-t5/t5-large for fast tokenizer support

Fix handler: load tokenizer from google-t5/t5-large to avoid local spiece.model path issue

Fix tokenizer_config: use minimal clean T5 config (remove backend=tokenizers)

Add spiece.model (SentencePiece vocab) for T5Tokenizer

Fix: remove extra_special_tokens (incompatible list format) - extra_ids=100 handles this

Remove fast tokenizer to fix extra_special_tokens compatibility issue

Fix tokenizer_config: convert extra_special_tokens from list to dict for transformers compat

Fix handler: use T5Tokenizer with legacy=True to fix extra_special_tokens error + float16 on GPU

Add custom handler for HuggingFace Inference Endpoints

Update to v4: trained on 6.2K samples, composite 0.6047, 8/10 Human-Written verdicts

Upload tokenizer

Upload T5ForConditionalGeneration

Final trained TextSight Humanizer T5-Large model

Training in progress, step 3564, checkpoint

Training in progress, step 3564

Training in progress, step 3200, checkpoint

Training in progress, step 3200

Training in progress, step 2800, checkpoint

Training in progress, step 2800

Training in progress, step 2400, checkpoint

Training in progress, step 2400

Training in progress, step 2000, checkpoint

Training in progress, step 2000

Training in progress, step 1600, checkpoint

Training in progress, step 1600

Training in progress, step 1200, checkpoint

Training in progress, step 1200

Training in progress, step 800, checkpoint

Training in progress, step 800

Training in progress, step 400, checkpoint

Training in progress, step 400

Training in progress, step 2000, checkpoint

Training in progress, step 2000

Training in progress, step 1600, checkpoint

Training in progress, step 1600

Training in progress, step 1200, checkpoint

Training in progress, step 1200

Training in progress, step 800, checkpoint

Training in progress, step 800

Training in progress, step 400, checkpoint

Training in progress, step 400

Training in progress, step 2500, checkpoint

Training in progress, step 2000, checkpoint

Training in progress, step 1500, checkpoint

Training in progress, step 1000, checkpoint

Training in progress, step 500, checkpoint

Training in progress, step 500

initial commit