Instructions to use Reubencf/konkani-orpheus-tts with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Reubencf/konkani-orpheus-tts with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("unsloth/orpheus-3b-0.1-ft") model = PeftModel.from_pretrained(base_model, "Reubencf/konkani-orpheus-tts") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Unsloth Desktop
Konkani Orpheus TTS
Latest rank-64 adapter trained for one epoch from the base checkpoint on corrected Romi Konkani synthetic speech. The previous, lower-quality adapter has been removed from Hugging Face; this repository now contains the latest trained model.
Training clips: 18,300. Completed optimizer steps: 4,575. Validation and test examples were not included in training. Single synthetic voice; machine-corrected transcripts. Load alongside the base model; base weights are not duplicated here.
Voice name: Konkani. Direct inference must retain the learned fola: conditioning token before the tagged Romi transcript; this is an internal training token, not the public voice name. The apps apply it automatically. Vocal-event tags are retained in training; expression realization and pronunciation require listening review. Held-out generated samples are included; this run did not use validation-loss-based checkpoint selection.
Dataset: Konkani synthetic speech,
pinned to 72e1c29c1f47040d5b192b5c97c75f646aa45f80. Source/provider terms and synthetic-data
provenance remain documented in the dataset card. Training metrics and
generation/transcription examples are included.
- Downloads last month
- 43
Model tree for Reubencf/konkani-orpheus-tts
Base model
meta-llama/Llama-3.2-3B-Instruct