See axolotl config

axolotl version: 0.10.0.dev0

base_model: syvai/tts-v1-pretrained
# Automatically upload checkpoint and final model to HF
hub_model_id: syvai/tts-v0.2-finetuned

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
liger_fused_linear_cross_entropy: true

datasets:
  - path: syvai/zac-coral-tts
    type: 
  - path: syvai/zac-dk-voice-pro
    type:
dataset_prepared_path: last_run_prepared
val_set_size: 0.01
eval_sample_packing: False
output_dir: ./outputs/finetuned

sequence_len: 8196
sample_packing: true
pad_to_sequence_len: true

wandb_project: orph
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 16
micro_batch_size: 2
num_epochs: 3
optimizer: adamw_torch_fused
lr_scheduler: cosine
learning_rate: 2e-5

bf16: auto
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
resume_from_checkpoint:
logging_steps: 1
flash_attention: true

warmup_steps: 3
evals_per_epoch: 5
saves_per_epoch: 5
weight_decay: 0.05

special_tokens:
  pad_token: <custom_token_7>

tts-v0.2-finetuned

This model is a fine-tuned version of syvai/tts-v1-pretrained on the syvai/zac-coral-tts and the syvai/zac-dk-voice-pro datasets. It achieves the following results on the evaluation set:

• Loss: 4.4147

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 3
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
4.8887	0.0207	1	4.9883
4.6631	0.2070	10	4.6965
4.5644	0.4140	20	4.5822
4.4871	0.6210	30	4.5155
4.4669	0.8279	40	4.4730
4.4424	1.0414	50	4.4485
4.4165	1.2484	60	4.4336
4.3866	1.4554	70	4.4247
4.4057	1.6624	80	4.4194
4.3947	1.8693	90	4.4167
4.3952	2.0828	100	4.4155
4.4001	2.2898	110	4.4153
4.3948	2.4968	120	4.4150
4.3996	2.7038	130	4.4148
4.3888	2.9107	140	4.4147

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1