Anonymize evaluation result paths for double-blind review

Update examples to show code-switching improvements

Recalculate MER after removing preamble from transcriptions

Update evaluation results and README with new benchmarks

Update README.md

Rename eval_results/trained_static.json to eval_results/trained_emilia.json

Rename eval_results/baseline_static.json to eval_results/baseline_emilia.json

Rename eval_results/trained_onpolicy.json to eval_results/trained_csdialogue.json

Rename eval_results/baseline_onpolicy.json to eval_results/baseline_csdialogue.json

Update README.md

Update README.md

Update README.md

Update trained_seame.json with baseline transcriptions from complete dataset

Fix baseline_seame.json - use complete 9764 samples instead of 4884

Add model card metadata and detailed hyperparameters

Update README with detailed hyperparameters and key findings

Upload Qwen2-Audio DPO CodeSwitch LoRA adapter with eval results

initial commit