Fix postprocess: filter eos tokens, preserve casing

Fix eos_token_id to use correct <|im_end|> and <|endoftext|> tokens

Training in progress, step 14000

Training in progress, step 12000

Training in progress, step 10000

Training in progress, step 8000

Training in progress, step 6000

Training in progress, step 4000

Training in progress, step 2000

Fix base_model_name_or_path to enable pipeline loading

Training in progress, step 2000

Upload asr_pipeline.py with huggingface_hub

Upload adapter_config.json with huggingface_hub

Training in progress, step 12000

Update custom model files, README, and requirements

Training in progress, step 10000

Training in progress, step 8000

Fix base_model_name_or_path for pipeline loading

Training in progress, step 6000

Training in progress, step 4000

Training in progress, step 2000

Training in progress, step 2000

Update custom model files, README, and requirements

Model save

Training in progress, step 67132

Training in progress, step 67000

Training in progress, step 66000

Training in progress, step 65000

Training in progress, step 64000

Training in progress, step 63000

Training in progress, step 62000

Training in progress, step 61000

Training in progress, step 60000

Training in progress, step 59000

Training in progress, step 58000

Training in progress, step 57000

Training in progress, step 56000

Training in progress, step 55000

Training in progress, step 54000

Training in progress, step 53000

Training in progress, step 52000

Training in progress, step 51000

Training in progress, step 50000

Training in progress, step 49000

Training in progress, step 48000

Training in progress, step 47000

Training in progress, step 46000

Training in progress, step 45000

Training in progress, step 44000

Training in progress, step 43000