Upload folder using huggingface_hub

finetuned/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/best_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7535e49e8e0cb01c26dd8b3eb77f7e1b8fc04c3a4a707d0ad58af148991b6943
+size 5607856542

finetuned/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/best_model_4323.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7535e49e8e0cb01c26dd8b3eb77f7e1b8fc04c3a4a707d0ad58af148991b6943
+size 5607856542

finetuned/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/config.json

@@ -0,0 +1,206 @@
+{
+    "output_path": "/kaggle/working/coqui-ai-TTS/run/training",
+    "logger_uri": null,
+    "run_name": "GPT_XTTS_v2.0_LJSpeech_FT",
+    "project_name": "XTTS_trainer",
+    "run_description": "\n            GPT XTTS training\n            ",
+    "print_step": 50,
+    "plot_step": 100,
+    "model_param_stats": false,
+    "wandb_entity": null,
+    "dashboard_logger": "tensorboard",
+    "save_on_interrupt": true,
+    "log_model_step": 1000,
+    "save_step": 10000,
+    "save_n_checkpoints": 1,
+    "save_checkpoints": true,
+    "save_all_best": false,
+    "save_best_after": 0,
+    "target_loss": null,
+    "print_eval": false,
+    "test_delay_epochs": 0,
+    "run_eval": true,
+    "run_eval_steps": null,
+    "distributed_backend": "nccl",
+    "distributed_url": "tcp://localhost:54321",
+    "mixed_precision": false,
+    "precision": "fp16",
+    "epochs": 2,
+    "batch_size": 3,
+    "eval_batch_size": 3,
+    "grad_clip": 0.0,
+    "scheduler_after_epoch": true,
+    "lr": 5e-06,
+    "optimizer": "AdamW",
+    "optimizer_params": {
+        "betas": [
+            0.9,
+            0.96
+        ],
+        "eps": 1e-08,
+        "weight_decay": 0.01
+    },
+    "lr_scheduler": "MultiStepLR",
+    "lr_scheduler_params": {
+        "milestones": [
+            900000,
+            2700000,
+            5400000
+        ],
+        "gamma": 0.5,
+        "last_epoch": -1
+    },
+    "use_grad_scaler": false,
+    "allow_tf32": false,
+    "cudnn_enable": true,
+    "cudnn_deterministic": false,
+    "cudnn_benchmark": false,
+    "training_seed": 1,
+    "model": "xtts",
+    "num_loader_workers": 8,
+    "num_eval_loader_workers": 0,
+    "use_noise_augment": false,
+    "audio": {
+        "sample_rate": 22050,
+        "output_sample_rate": 24000,
+        "dvae_sample_rate": 22050
+    },
+    "model_args": {
+        "gpt_batch_size": 1,
+        "enable_redaction": false,
+        "kv_cache": true,
+        "gpt_checkpoint": "",
+        "clvp_checkpoint": null,
+        "decoder_checkpoint": null,
+        "num_chars": 67,
+        "tokenizer_file": "/kaggle/working/coqui-ai-TTS/run/training/XTTS_v2.0_original_model_files/vocab.json",
+        "gpt_max_audio_tokens": 605,
+        "gpt_max_text_tokens": 402,
+        "gpt_max_prompt_tokens": 70,
+        "gpt_layers": 30,
+        "gpt_n_model_channels": 1024,
+        "gpt_n_heads": 16,
+        "gpt_number_text_tokens": 6681,
+        "gpt_start_text_token": 261,
+        "gpt_stop_text_token": 0,
+        "gpt_num_audio_tokens": 1026,
+        "gpt_start_audio_token": 1024,
+        "gpt_stop_audio_token": 1025,
+        "gpt_code_stride_len": 1024,
+        "gpt_use_masking_gt_prompt_approach": true,
+        "gpt_use_perceiver_resampler": true,
+        "input_sample_rate": 22050,
+        "output_sample_rate": 24000,
+        "output_hop_length": 256,
+        "decoder_input_dim": 1024,
+        "d_vector_dim": 512,
+        "cond_d_vector_in_each_upsampling_layer": true,
+        "duration_const": 102400,
+        "min_conditioning_length": 66150,
+        "max_conditioning_length": 132300,
+        "gpt_loss_text_ce_weight": 0.01,
+        "gpt_loss_mel_ce_weight": 1.0,
+        "debug_loading_failures": false,
+        "max_wav_length": 255995,
+        "max_text_length": 200,
+        "mel_norm_file": "/kaggle/working/coqui-ai-TTS/run/training/XTTS_v2.0_original_model_files/mel_stats.pth",
+        "dvae_checkpoint": "/kaggle/working/coqui-ai-TTS/run/training/XTTS_v2.0_original_model_files/dvae.pth",
+        "xtts_checkpoint": "/kaggle/working/coqui-ai-TTS/run/training/XTTS_v2.0_original_model_files/model.pth",
+        "vocoder": ""
+    },
+    "_supports_cloning": true,
+    "languages": [
+        "en",
+        "es",
+        "fr",
+        "de",
+        "it",
+        "pt",
+        "pl",
+        "tr",
+        "ru",
+        "nl",
+        "cs",
+        "ar",
+        "zh-cn",
+        "hu",
+        "ko",
+        "ja",
+        "hi"
+    ],
+    "speakers": [],
+    "use_phonemes": false,
+    "phonemizer": null,
+    "phoneme_language": null,
+    "compute_input_seq_cache": false,
+    "text_cleaner": null,
+    "enable_eos_bos_chars": false,
+    "test_sentences_file": "",
+    "phoneme_cache_path": null,
+    "characters": null,
+    "add_blank": false,
+    "batch_group_size": 48,
+    "loss_masking": null,
+    "min_audio_len": 1,
+    "max_audio_len": Infinity,
+    "min_text_len": 1,
+    "max_text_len": Infinity,
+    "compute_f0": false,
+    "compute_energy": false,
+    "compute_linear_spec": false,
+    "precompute_num_workers": 0,
+    "start_by_longest": false,
+    "shuffle": false,
+    "drop_last": false,
+    "datasets": [
+        {
+            "formatter": "ljspeech",
+            "dataset_name": "ljspeech",
+            "path": "/kaggle/working/ljspeech_fmt/",
+            "meta_file_train": "/kaggle/working/ljspeech_fmt/metadata.csv",
+            "ignored_speakers": null,
+            "language": "en",
+            "phonemizer": "",
+            "meta_file_val": "",
+            "meta_file_attn_mask": ""
+        }
+    ],
+    "test_sentences": [
+        {
+            "text": "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent.",
+            "speaker_wav": [
+                "./tests/data/ljspeech/wavs/LJ001-0002.wav"
+            ],
+            "language": "en"
+        },
+        {
+            "text": "This cake is great. It's so delicious and moist.",
+            "speaker_wav": [
+                "./tests/data/ljspeech/wavs/LJ001-0002.wav"
+            ],
+            "language": "en"
+        }
+    ],
+    "eval_split_max_size": 256,
+    "eval_split_size": 0.01,
+    "use_speaker_weighted_sampler": false,
+    "speaker_weighted_sampler_alpha": 1.0,
+    "use_language_weighted_sampler": false,
+    "language_weighted_sampler_alpha": 1.0,
+    "use_length_weighted_sampler": false,
+    "length_weighted_sampler_alpha": 1.0,
+    "temperature": 0.85,
+    "length_penalty": 1.0,
+    "repetition_penalty": 2.0,
+    "top_k": 50,
+    "top_p": 0.85,
+    "num_gpt_outputs": 1,
+    "gpt_cond_len": 12,
+    "gpt_cond_chunk_len": 4,
+    "max_ref_len": 10,
+    "sound_norm_refs": false,
+    "optimizer_wd_only_on_weights": true,
+    "weighted_loss_attrs": {},
+    "weighted_loss_multipliers": {},
+    "github_branch": "* dev"
+}

finetuned/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/events.out.tfevents.1792547554.90c29ef008dd.122.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64033e7fa48c7b74b32f9677a57cb1846d8bea7239cd2bcbfbcb526bebb5d1ba
+size 1017758

finetuned/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/recipe_frox_finetune.py

@@ -0,0 +1,178 @@
+import os
+
+from trainer import Trainer, TrainerArgs
+
+from TTS.config.shared_configs import BaseDatasetConfig
+from TTS.tts.configs.xtts_config import XttsAudioConfig
+from TTS.tts.datasets import load_tts_samples
+from TTS.tts.layers.xtts.trainer.gpt_trainer import GPTArgs, GPTTrainer, GPTTrainerConfig
+from TTS.utils.manage import ModelManager
+
+# Logging parameters
+RUN_NAME = "GPT_XTTS_v2.0_LJSpeech_FT"
+PROJECT_NAME = "XTTS_trainer"
+DASHBOARD_LOGGER = "tensorboard"
+LOGGER_URI = None
+
+# Set here the path that the checkpoints will be saved. Default: ./run/training/
+OUT_PATH = os.path.join(os.path.dirname(os.path.abspath(__file__)), "run", "training")
+
+# Training Parameters
+OPTIMIZER_WD_ONLY_ON_WEIGHTS = True  # for multi-gpu training please make it False
+START_WITH_EVAL = True  # if True it will star with evaluation
+BATCH_SIZE = 3  # set here the batch size
+GRAD_ACUMM_STEPS = 84  # set here the grad accumulation steps
+# Note: we recommend that BATCH_SIZE * GRAD_ACUMM_STEPS need to be at least 252 for more efficient training. You can increase/decrease BATCH_SIZE but then set GRAD_ACUMM_STEPS accordingly.
+
+# Define here the dataset(s) that you want to use for the fine-tuning.
+DATASETS_CONFIG_LIST = [
+    BaseDatasetConfig(
+        formatter="ljspeech",
+        dataset_name="ljspeech",
+        path="/kaggle/working/ljspeech_fmt/",
+        meta_file_train="/kaggle/working/ljspeech_fmt/metadata.csv",
+        language="en",
+    )
+]
+
+# Define the path where XTTS v2.0.1 files will be downloaded
+CHECKPOINTS_OUT_PATH = os.path.join(OUT_PATH, "XTTS_v2.0_original_model_files/")
+os.makedirs(CHECKPOINTS_OUT_PATH, exist_ok=True)
+
+
+# DVAE files
+DVAE_CHECKPOINT_LINK = "https://huggingface.co/coqui/XTTS-v2/resolve/main/dvae.pth"
+MEL_NORM_LINK = "https://huggingface.co/coqui/XTTS-v2/resolve/main/mel_stats.pth"
+
+# Set the path to the downloaded files
+DVAE_CHECKPOINT = os.path.join(CHECKPOINTS_OUT_PATH, os.path.basename(DVAE_CHECKPOINT_LINK))
+MEL_NORM_FILE = os.path.join(CHECKPOINTS_OUT_PATH, os.path.basename(MEL_NORM_LINK))
+
+# download DVAE files if needed
+if not os.path.isfile(DVAE_CHECKPOINT) or not os.path.isfile(MEL_NORM_FILE):
+    print(" > Downloading DVAE files!")
+    ModelManager._download_model_files([MEL_NORM_LINK, DVAE_CHECKPOINT_LINK], CHECKPOINTS_OUT_PATH, progress_bar=True)
+
+
+# Download XTTS v2.0 checkpoint if needed
+TOKENIZER_FILE_LINK = "https://huggingface.co/coqui/XTTS-v2/resolve/main/vocab.json"
+XTTS_CHECKPOINT_LINK = "https://huggingface.co/coqui/XTTS-v2/resolve/main/model.pth"
+
+# XTTS transfer learning parameters: You we need to provide the paths of XTTS model checkpoint that you want to do the fine tuning.
+TOKENIZER_FILE = os.path.join(CHECKPOINTS_OUT_PATH, os.path.basename(TOKENIZER_FILE_LINK))  # vocab.json file
+XTTS_CHECKPOINT = os.path.join(CHECKPOINTS_OUT_PATH, os.path.basename(XTTS_CHECKPOINT_LINK))  # model.pth file
+
+# download XTTS v2.0 files if needed
+if not os.path.isfile(TOKENIZER_FILE) or not os.path.isfile(XTTS_CHECKPOINT):
+    print(" > Downloading XTTS v2.0 files!")
+    ModelManager._download_model_files(
+        [TOKENIZER_FILE_LINK, XTTS_CHECKPOINT_LINK], CHECKPOINTS_OUT_PATH, progress_bar=True
+    )
+
+
+# Training sentences generations
+SPEAKER_REFERENCE = [
+    "./tests/data/ljspeech/wavs/LJ001-0002.wav"  # speaker reference to be used in training test sentences
+]
+LANGUAGE = DATASETS_CONFIG_LIST[0].language
+
+
+def main():
+    # init args and config
+    model_args = GPTArgs(
+        max_conditioning_length=132300,  # 6 secs
+        min_conditioning_length=66150,  # 3 secs
+        debug_loading_failures=False,
+        max_wav_length=255995,  # ~11.6 seconds
+        max_text_length=200,
+        mel_norm_file=MEL_NORM_FILE,
+        dvae_checkpoint=DVAE_CHECKPOINT,
+        xtts_checkpoint=XTTS_CHECKPOINT,  # checkpoint path of the model that you want to fine-tune
+        tokenizer_file=TOKENIZER_FILE,
+        gpt_num_audio_tokens=1026,
+        gpt_start_audio_token=1024,
+        gpt_stop_audio_token=1025,
+        gpt_use_masking_gt_prompt_approach=True,
+        gpt_use_perceiver_resampler=True,
+    )
+    # define audio config
+    audio_config = XttsAudioConfig(sample_rate=22050, dvae_sample_rate=22050, output_sample_rate=24000)
+    # training parameters config
+    config = GPTTrainerConfig(
+        epochs=2,
+        output_path=OUT_PATH,
+        model_args=model_args,
+        run_name=RUN_NAME,
+        project_name=PROJECT_NAME,
+        run_description="""
+            GPT XTTS training
+            """,
+        datasets=DATASETS_CONFIG_LIST,
+        dashboard_logger=DASHBOARD_LOGGER,
+        logger_uri=LOGGER_URI,
+        audio=audio_config,
+        batch_size=BATCH_SIZE,
+        batch_group_size=48,
+        eval_batch_size=BATCH_SIZE,
+        num_loader_workers=8,
+        eval_split_max_size=256,
+        print_step=50,
+        plot_step=100,
+        log_model_step=1000,
+        save_step=10000,
+        save_n_checkpoints=1,
+        save_checkpoints=True,
+        # target_loss="loss",
+        print_eval=False,
+        # Optimizer values like tortoise, pytorch implementation with modifications to not apply WD to non-weight parameters.
+        optimizer="AdamW",
+        optimizer_wd_only_on_weights=OPTIMIZER_WD_ONLY_ON_WEIGHTS,
+        optimizer_params={"betas": [0.9, 0.96], "eps": 1e-8, "weight_decay": 1e-2},
+        lr=5e-06,  # learning rate
+        lr_scheduler="MultiStepLR",
+        # it was adjusted accordly for the new step scheme
+        lr_scheduler_params={"milestones": [50000 * 18, 150000 * 18, 300000 * 18], "gamma": 0.5, "last_epoch": -1},
+        test_sentences=[
+            {
+                "text": "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent.",
+                "speaker_wav": SPEAKER_REFERENCE,
+                "language": LANGUAGE,
+            },
+            {
+                "text": "This cake is great. It's so delicious and moist.",
+                "speaker_wav": SPEAKER_REFERENCE,
+                "language": LANGUAGE,
+            },
+        ],
+    )
+
+    # init the model from config
+    model = GPTTrainer(config)
+
+    # load training samples
+    train_samples, eval_samples = load_tts_samples(
+        config,
+        eval_split=True,
+        eval_split_max_size=config.eval_split_max_size,
+        eval_split_size=config.eval_split_size,
+    )
+
+    # init the trainer and 🚀
+    trainer = Trainer(
+        TrainerArgs(
+            restore_path=None,  # xtts checkpoint is restored via xtts_checkpoint key so no need of restore it using Trainer restore_path parameter
+            skip_train_epoch=False,
+            start_with_eval=START_WITH_EVAL,
+            grad_accum_steps=GRAD_ACUMM_STEPS,
+        ),
+        config,
+        output_path=OUT_PATH,
+        model=model,
+        train_samples=train_samples,
+        eval_samples=eval_samples,
+    )
+    trainer.fit()
+
+
+if __name__ == "__main__":
+    main()

finetuned/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/trainer_0_log.txt

@@ -0,0 +1,828 @@
+ > Training Environment:
+ | > Backend: Torch
+ | > Mixed precision: False
+ | > Precision: float32
+ | > Current device: 0
+ | > Num. of GPUs: 1
+ | > Num. of CPUs: 4
+ | > Num. of Torch Threads: 1
+ | > Torch seed: 1
+ | > Torch CUDNN: True
+ | > Torch CUDNN deterministic: False
+ | > Torch CUDNN benchmark: False
+ | > Torch TF32 MatMul: False
+ > Start Tensorboard: tensorboard --logdir=/kaggle/working/coqui-ai-TTS/run/training/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5
+
+ > Model has 518442047 parameters
+
+ > EPOCH: 0/1
+ --> /kaggle/working/coqui-ai-TTS/run/training/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5
+
+ > EVALUATION 
+
+
+  --> EVAL PERFORMANCE
+     | > avg_loader_time: 0.0775634299877078 (+0.0)
+     | > avg_loss_text_ce: 0.022715861107720885 (+0.0)
+     | > avg_loss_mel_ce: 3.651390269745228 (+0.0)
+     | > avg_loss: 3.6741061155186143 (+0.0)
+
+
+ > EPOCH: 1/1
+ --> /kaggle/working/coqui-ai-TTS/run/training/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5
+
+ > TRAINING (2026-10-21 01:55:49) 
+
+   --> TIME: 2026-10-21 01:55:54 -- STEP: 0/4322 -- GLOBAL_STEP: 0
+     | > current_lr: 5e-06  (5e-06)
+     | > loss_text_ce: 0.023607278242707253  (0.023607278242707253)
+     | > loss_mel_ce: 3.562453269958496  (3.562453269958496)
+     | > loss: 0.042691197246313095  (0.042691197246313095)
+     | > step_time: 1.0577  (1.0576918125152588)
+     | > loader_time: 3.7638  (3.763838768005371)
+
+
+   --> TIME: 2026-10-21 01:56:41 -- STEP: 50/4322 -- GLOBAL_STEP: 50
+     | > current_lr: 5e-06  (5.0000000000000055e-06)
+     | > loss_text_ce: 0.023330125957727432  (0.023871341645717623)
+     | > loss_mel_ce: 3.461495876312256  (3.6548802995681764)
+     | > loss: 0.04148602485656738  (0.0437946629524231)
+     | > step_time: 0.6104  (0.6138731288909911)
+     | > loader_time: 0.0148  (0.015347743034362793)
+
+
+   --> TIME: 2026-10-21 01:57:27 -- STEP: 100/4322 -- GLOBAL_STEP: 100
+     | > current_lr: 5e-06  (5.000000000000009e-06)
+     | > loss_text_ce: 0.022796062752604485  (0.023814733307808637)
+     | > loss_mel_ce: 3.5732829570770264  (3.6294152784347533)
+     | > loss: 0.04281046614050865  (0.043490834161639215)
+     | > step_time: 0.5825  (0.6164441776275634)
+     | > loader_time: 0.0171  (0.014822771549224853)
+
+
+   --> TIME: 2026-10-21 01:58:13 -- STEP: 150/4322 -- GLOBAL_STEP: 150
+     | > current_lr: 5e-06  (5.00000000000001e-06)
+     | > loss_text_ce: 0.022212712094187737  (0.02389630557348331)
+     | > loss_mel_ce: 3.4683268070220947  (3.5989029963811237)
+     | > loss: 0.0415540412068367  (0.04312856381138167)
+     | > step_time: 0.5846  (0.6149810393651326)
+     | > loader_time: 0.0157  (0.01448255221048991)
+
+
+   --> TIME: 2026-10-21 01:58:58 -- STEP: 200/4322 -- GLOBAL_STEP: 200
+     | > current_lr: 5e-06  (5.00000000000001e-06)
+     | > loss_text_ce: 0.023157035931944847  (0.02389827018603683)
+     | > loss_mel_ce: 3.4178578853607178  (3.581750749349594)
+     | > loss: 0.04096446558833122  (0.04292439378798008)
+     | > step_time: 0.6188  (0.6102964365482325)
+     | > loader_time: 0.0123  (0.014304686784744263)
+
+
+   --> TIME: 2026-10-21 01:59:41 -- STEP: 250/4322 -- GLOBAL_STEP: 250
+     | > current_lr: 5e-06  (5.0000000000000106e-06)
+     | > loss_text_ce: 0.0207450482994318  (0.023818049624562267)
+     | > loss_mel_ce: 3.610269784927368  (3.5579456262588502)
+     | > loss: 0.043226368725299835  (0.042640044435858726)
+     | > step_time: 0.4397  (0.6008358507156367)
+     | > loader_time: 0.0174  (0.014198965072631835)
+
+
+   --> TIME: 2026-10-21 02:00:26 -- STEP: 300/4322 -- GLOBAL_STEP: 300
+     | > current_lr: 5e-06  (5.0000000000000106e-06)
+     | > loss_text_ce: 0.02244749292731285  (0.02378187150383989)
+     | > loss_mel_ce: 3.2661588191986084  (3.5324360863367708)
+     | > loss: 0.03915007784962654  (0.04233592879027127)
+     | > step_time: 0.7449  (0.6004247697194415)
+     | > loader_time: 0.0105  (0.01390298843383789)
+
+
+   --> TIME: 2026-10-21 02:01:11 -- STEP: 350/4322 -- GLOBAL_STEP: 350
+     | > current_lr: 5e-06  (5.0000000000000106e-06)
+     | > loss_text_ce: 0.023772336542606354  (0.023702249436506194)
+     | > loss_mel_ce: 3.5447685718536377  (3.518683438301086)
+     | > loss: 0.04248262941837311  (0.04217125889446053)
+     | > step_time: 0.671  (0.5991947923387797)
+     | > loader_time: 0.0131  (0.013757729530334472)
+
+
+   --> TIME: 2026-10-21 02:01:55 -- STEP: 400/4322 -- GLOBAL_STEP: 400
+     | > current_lr: 5e-06  (5.0000000000000055e-06)
+     | > loss_text_ce: 0.024473151192069054  (0.023693136773072176)
+     | > loss_mel_ce: 3.1853315830230713  (3.49852819263935)
+     | > loss: 0.03821196407079697  (0.041931206975132224)
+     | > step_time: 0.6656  (0.5977507787942884)
+     | > loader_time: 0.014  (0.013656079173088073)
+
+
+   --> TIME: 2026-10-21 02:02:41 -- STEP: 450/4322 -- GLOBAL_STEP: 450
+     | > current_lr: 5e-06  (4.999999999999983e-06)
+     | > loss_text_ce: 0.023743927478790283  (0.023701659581727444)
+     | > loss_mel_ce: 3.377760171890259  (3.4801506688859725)
+     | > loss: 0.04049409553408623  (0.041712528417507805)
+     | > step_time: 0.4654  (0.5993869839774233)
+     | > loader_time: 0.0092  (0.013543645540873209)
+
+
+   --> TIME: 2026-10-21 02:03:26 -- STEP: 500/4322 -- GLOBAL_STEP: 500
+     | > current_lr: 5e-06  (4.999999999999964e-06)
+     | > loss_text_ce: 0.028117351233959198  (0.023697244830429547)
+     | > loss_mel_ce: 3.3383777141571045  (3.4620542507171628)
+     | > loss: 0.0400773249566555  (0.04149704233556986)
+     | > step_time: 0.4946  (0.5986857099533076)
+     | > loader_time: 0.0118  (0.013421703338623046)
+
+
+   --> TIME: 2026-10-21 02:04:12 -- STEP: 550/4322 -- GLOBAL_STEP: 550
+     | > current_lr: 5e-06  (4.99999999999995e-06)
+     | > loss_text_ce: 0.024159494787454605  (0.023733612895011892)
+     | > loss_mel_ce: 3.1997478008270264  (3.445531345280734)
+     | > loss: 0.03837984800338745  (0.04130077404054731)
+     | > step_time: 0.6202  (0.5997316876324739)
+     | > loader_time: 0.0123  (0.013335978768088603)
+
+
+   --> TIME: 2026-10-21 02:04:58 -- STEP: 600/4322 -- GLOBAL_STEP: 600
+     | > current_lr: 5e-06  (4.9999999999999335e-06)
+     | > loss_text_ce: 0.02490082010626793  (0.02375749908698101)
+     | > loss_mel_ce: 3.2087249755859375  (3.4295414439837137)
+     | > loss: 0.03849554806947708  (0.041110702430208555)
+     | > step_time: 0.6296  (0.6005522024631497)
+     | > loader_time: 0.0105  (0.013264158566792805)
+
+
+   --> TIME: 2026-10-21 02:05:43 -- STEP: 650/4322 -- GLOBAL_STEP: 650
+     | > current_lr: 5e-06  (4.999999999999923e-06)
+     | > loss_text_ce: 0.02134285867214203  (0.023758635317476892)
+     | > loss_mel_ce: 3.2015085220336914  (3.4134296200825616)
+     | > loss: 0.03836727887392044  (0.04091890854904286)
+     | > step_time: 0.5952  (0.6004456068919255)
+     | > loader_time: 0.011  (0.013190783720750075)
+
+
+   --> TIME: 2026-10-21 02:06:29 -- STEP: 700/4322 -- GLOBAL_STEP: 700
+     | > current_lr: 5e-06  (4.999999999999907e-06)
+     | > loss_text_ce: 0.020476728677749634  (0.023712386890713647)
+     | > loss_mel_ce: 3.1137752532958984  (3.398600721018655)
+     | > loss: 0.037312526255846024  (0.04074182349124122)
+     | > step_time: 0.5622  (0.6007283912386213)
+     | > loader_time: 0.008  (0.013160109519958496)
+
+
+   --> TIME: 2026-10-21 02:07:14 -- STEP: 750/4322 -- GLOBAL_STEP: 750
+     | > current_lr: 5e-06  (4.999999999999906e-06)
+     | > loss_text_ce: 0.025615012273192406  (0.02372802059104044)
+     | > loss_mel_ce: 3.301236152648926  (3.3874267937342326)
+     | > loss: 0.03960537165403366  (0.040608986645936915)
+     | > step_time: 0.5906  (0.6008401826222738)
+     | > loader_time: 0.0138  (0.013096263885498047)
+
+
+   --> TIME: 2026-10-21 02:07:59 -- STEP: 800/4322 -- GLOBAL_STEP: 800
+     | > current_lr: 5e-06  (4.999999999999904e-06)
+     | > loss_text_ce: 0.022030288353562355  (0.02372636462096124)
+     | > loss_mel_ce: 3.0623843669891357  (3.3740680557489395)
+     | > loss: 0.036719225347042084  (0.04044993432238694)
+     | > step_time: 0.6571  (0.5999964663386342)
+     | > loader_time: 0.0125  (0.013091897070407867)
+
+
+   --> TIME: 2026-10-21 02:08:44 -- STEP: 850/4322 -- GLOBAL_STEP: 850
+     | > current_lr: 5e-06  (4.999999999999897e-06)
+     | > loss_text_ce: 0.02976416051387787  (0.023754417444853213)
+     | > loss_mel_ce: 3.136929988861084  (3.3624984926335952)
+     | > loss: 0.037698742002248764  (0.04031253539902319)
+     | > step_time: 0.4763  (0.6000622721279365)
+     | > loader_time: 0.0117  (0.013060874097487506)
+
+
+   --> TIME: 2026-10-21 02:09:29 -- STEP: 900/4322 -- GLOBAL_STEP: 900
+     | > current_lr: 5e-06  (4.999999999999892e-06)
+     | > loss_text_ce: 0.024680709466338158  (0.023769160101397162)
+     | > loss_mel_ce: 3.0224132537841797  (3.3509844075308908)
+     | > loss: 0.036274928599596024  (0.04017563847617967)
+     | > step_time: 0.6022  (0.6000583444701298)
+     | > loader_time: 0.0069  (0.012994143433041042)
+
+
+   --> TIME: 2026-10-21 02:10:15 -- STEP: 950/4322 -- GLOBAL_STEP: 950
+     | > current_lr: 5e-06  (4.999999999999887e-06)
+     | > loss_text_ce: 0.02795293740928173  (0.02378540836470691)
+     | > loss_mel_ce: 2.8190839290618896  (3.340535431912071)
+     | > loss: 0.033893294632434845  (0.04005143933782449)
+     | > step_time: 0.5454  (0.6004340179342971)
+     | > loader_time: 0.0107  (0.01297067165374756)
+
+
+   --> TIME: 2026-10-21 02:10:59 -- STEP: 1000/4322 -- GLOBAL_STEP: 1000
+     | > current_lr: 5e-06  (4.999999999999882e-06)
+     | > loss_text_ce: 0.020246483385562897  (0.02378400870971381)
+     | > loss_mel_ce: 3.172618865966797  (3.329640386581421)
+     | > loss: 0.03801030293107033  (0.039921719752252074)
+     | > step_time: 0.5444  (0.5995927650928495)
+     | > loader_time: 0.0122  (0.012936718463897705)
+
+
+   --> TIME: 2026-10-21 02:11:45 -- STEP: 1050/4322 -- GLOBAL_STEP: 1050
+     | > current_lr: 5e-06  (4.9999999999998784e-06)
+     | > loss_text_ce: 0.023922035470604897  (0.023778165232922345)
+     | > loss_mel_ce: 3.2467410564422607  (3.32012759799049)
+     | > loss: 0.038936465978622437  (0.039808402703631465)
+     | > step_time: 0.7345  (0.599874627703712)
+     | > loader_time: 0.0125  (0.012911377180190314)
+
+
+   --> TIME: 2026-10-21 02:12:28 -- STEP: 1100/4322 -- GLOBAL_STEP: 1100
+     | > current_lr: 5e-06  (4.9999999999998725e-06)
+     | > loss_text_ce: 0.022987011820077896  (0.02377308981831778)
+     | > loss_mel_ce: 3.168280839920044  (3.3113870924169366)
+     | > loss: 0.03799128532409668  (0.03970428864048283)
+     | > step_time: 0.6301  (0.5985342764854431)
+     | > loader_time: 0.0122  (0.012879582968625155)
+
+
+   --> TIME: 2026-10-21 02:13:14 -- STEP: 1150/4322 -- GLOBAL_STEP: 1150
+     | > current_lr: 5e-06  (4.999999999999866e-06)
+     | > loss_text_ce: 0.024523338302969933  (0.023767024406596372)
+     | > loss_mel_ce: 3.2279915809631348  (3.3026936597409446)
+     | > loss: 0.03872041404247284  (0.039600723184969086)
+     | > step_time: 0.6668  (0.5987930737371031)
+     | > loader_time: 0.0128  (0.01285959927932076)
+
+
+   --> TIME: 2026-10-21 02:13:58 -- STEP: 1200/4322 -- GLOBAL_STEP: 1200
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.024073252454400063  (0.023750210852983098)
+     | > loss_mel_ce: 3.2074012756347656  (3.294540059367815)
+     | > loss: 0.038469936698675156  (0.03950345635414121)
+     | > step_time: 0.6267  (0.5983327811956408)
+     | > loader_time: 0.0086  (0.012834071715672815)
+
+
+   --> TIME: 2026-10-21 02:14:43 -- STEP: 1250/4322 -- GLOBAL_STEP: 1250
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.02361719310283661  (0.023736771930754175)
+     | > loss_mel_ce: 3.1402785778045654  (3.2860619565963747)
+     | > loss: 0.03766542673110962  (0.039402366572618454)
+     | > step_time: 0.6379  (0.5983164710998535)
+     | > loader_time: 0.0123  (0.012797017288208003)
+
+
+   --> TIME: 2026-10-21 02:15:28 -- STEP: 1300/4322 -- GLOBAL_STEP: 1300
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.02002999745309353  (0.023741446633178443)
+     | > loss_mel_ce: 2.91009783744812  (3.276316642027635)
+     | > loss: 0.034882474690675735  (0.03928640657606032)
+     | > step_time: 0.5945  (0.5977295703154349)
+     | > loader_time: 0.0122  (0.01277180983470036)
+
+
+   --> TIME: 2026-10-21 02:16:12 -- STEP: 1350/4322 -- GLOBAL_STEP: 1350
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.021801773458719254  (0.023740276019606304)
+     | > loss_mel_ce: 3.1130032539367676  (3.268503632898684)
+     | > loss: 0.037319108843803406  (0.039193380621296375)
+     | > step_time: 0.6306  (0.5971331243161812)
+     | > loader_time: 0.0121  (0.012747642022592045)
+
+
+   --> TIME: 2026-10-21 02:16:57 -- STEP: 1400/4322 -- GLOBAL_STEP: 1400
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.024811046198010445  (0.02374142150793755)
+     | > loss_mel_ce: 3.0756020545959473  (3.2625486939294)
+     | > loss: 0.03690968081355095  (0.03912250211728469)
+     | > step_time: 0.5325  (0.597321861812047)
+     | > loader_time: 0.0066  (0.012735525199345175)
+
+
+   --> TIME: 2026-10-21 02:17:43 -- STEP: 1450/4322 -- GLOBAL_STEP: 1450
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.022022727876901627  (0.023734815263285698)
+     | > loss_mel_ce: 3.0583367347717285  (3.256251094916771)
+     | > loss: 0.03667094558477402  (0.03904745205209174)
+     | > step_time: 0.6574  (0.5975845338558333)
+     | > loader_time: 0.0164  (0.012712322596845951)
+
+
+   --> TIME: 2026-10-21 02:18:28 -- STEP: 1500/4322 -- GLOBAL_STEP: 1500
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.02297208644449711  (0.02374113366380332)
+     | > loss_mel_ce: 3.050678014755249  (3.2495009508132933)
+     | > loss: 0.036591075360774994  (0.038967168403168495)
+     | > step_time: 0.6333  (0.5978817100524908)
+     | > loader_time: 0.0125  (0.01268777195612589)
+
+
+   --> TIME: 2026-10-21 02:19:13 -- STEP: 1550/4322 -- GLOBAL_STEP: 1550
+     | > current_lr: 5e-06  (4.999999999999865e-06)
+     | > loss_text_ce: 0.02353771962225437  (0.023733780615993065)
+     | > loss_mel_ce: 2.956113815307617  (3.2425041924753497)
+     | > loss: 0.035472042858600616  (0.038883786131778085)
+     | > step_time: 0.581  (0.5976957441145371)
+     | > loader_time: 0.0111  (0.012671740593448758)
+
+
+   --> TIME: 2026-10-21 02:19:57 -- STEP: 1600/4322 -- GLOBAL_STEP: 1600
+     | > current_lr: 5e-06  (4.999999999999863e-06)
+     | > loss_text_ce: 0.02201055735349655  (0.02372139338636769)
+     | > loss_mel_ce: 3.0538761615753174  (3.2370670783519744)
+     | > loss: 0.0366177000105381  (0.03881891111144796)
+     | > step_time: 0.6707  (0.5972625337541095)
+     | > loader_time: 0.0087  (0.012655767202377315)
+
+
+   --> TIME: 2026-10-21 02:20:43 -- STEP: 1650/4322 -- GLOBAL_STEP: 1650
+     | > current_lr: 5e-06  (4.9999999999998615e-06)
+     | > loss_text_ce: 0.02265886962413788  (0.023707734198507018)
+     | > loss_mel_ce: 3.0409398078918457  (3.2308994569200458)
+     | > loss: 0.0364714115858078  (0.03874532444007469)
+     | > step_time: 0.7344  (0.5974271405826904)
+     | > loader_time: 0.017  (0.012655190410036024)
+
+
+   --> TIME: 2026-10-21 02:21:28 -- STEP: 1700/4322 -- GLOBAL_STEP: 1700
+     | > current_lr: 5e-06  (4.99999999999986e-06)
+     | > loss_text_ce: 0.023584647104144096  (0.02371008093304492)
+     | > loss_mel_ce: 2.9107162952423096  (3.224411839877858)
+     | > loss: 0.034932155162096024  (0.03866811884676709)
+     | > step_time: 0.622  (0.5972898818464829)
+     | > loader_time: 0.0125  (0.012630745523116164)
+
+
+   --> TIME: 2026-10-21 02:22:12 -- STEP: 1750/4322 -- GLOBAL_STEP: 1750
+     | > current_lr: 5e-06  (4.999999999999858e-06)
+     | > loss_text_ce: 0.020321423187851906  (0.023709888081465434)
+     | > loss_mel_ce: 2.746750593185425  (3.218360703468323)
+     | > loss: 0.03294133394956589  (0.03859607922179358)
+     | > step_time: 0.5363  (0.5970047991616375)
+     | > loader_time: 0.0083  (0.012610953739711212)
+
+
+   --> TIME: 2026-10-21 02:22:57 -- STEP: 1800/4322 -- GLOBAL_STEP: 1800
+     | > current_lr: 5e-06  (4.999999999999856e-06)
+     | > loss_text_ce: 0.02185886725783348  (0.023704143172750856)
+     | > loss_mel_ce: 3.109938859939575  (3.2124653157922958)
+     | > loss: 0.03728330880403519  (0.03852582764294413)
+     | > step_time: 0.561  (0.5969216158654944)
+     | > loader_time: 0.0111  (0.012595439487033417)
+
+
+   --> TIME: 2026-10-21 02:23:42 -- STEP: 1850/4322 -- GLOBAL_STEP: 1850
+     | > current_lr: 5e-06  (4.999999999999855e-06)
+     | > loss_text_ce: 0.024742452427744865  (0.0236988801801124)
+     | > loss_mel_ce: 2.976585865020752  (3.207316894402375)
+     | > loss: 0.03573009744286537  (0.03846447426322344)
+     | > step_time: 0.6692  (0.5968153008899163)
+     | > loader_time: 0.0153  (0.012587847194156128)
+
+
+   --> TIME: 2026-10-21 02:24:25 -- STEP: 1900/4322 -- GLOBAL_STEP: 1900
+     | > current_lr: 5e-06  (4.999999999999853e-06)
+     | > loss_text_ce: 0.025433124974370003  (0.023695998475151612)
+     | > loss_mel_ce: 3.161982536315918  (3.202033802333631)
+     | > loss: 0.037945423275232315  (0.03840154600574782)
+     | > step_time: 0.579  (0.5961658215522756)
+     | > loader_time: 0.0085  (0.012579835339596394)
+
+
+   --> TIME: 2026-10-21 02:25:11 -- STEP: 1950/4322 -- GLOBAL_STEP: 1950
+     | > current_lr: 5e-06  (4.999999999999851e-06)
+     | > loss_text_ce: 0.019193362444639206  (0.023692078514932043)
+     | > loss_mel_ce: 3.1381239891052246  (3.1976696112217047)
+     | > loss: 0.03758711367845535  (0.03834954468867718)
+     | > step_time: 0.5367  (0.5963146784366696)
+     | > loader_time: 0.0108  (0.012574802789932638)
+
+
+   --> TIME: 2026-10-21 02:25:56 -- STEP: 2000/4322 -- GLOBAL_STEP: 2000
+     | > current_lr: 5e-06  (4.99999999999985e-06)
+     | > loss_text_ce: 0.020343022421002388  (0.023680220687761887)
+     | > loss_mel_ce: 3.0056772232055664  (3.192688380122185)
+     | > loss: 0.036024052649736404  (0.03829010315425694)
+     | > step_time: 0.6614  (0.5964905810356131)
+     | > loader_time: 0.0088  (0.012561839818954465)
+
+
+   --> TIME: 2026-10-21 02:26:42 -- STEP: 2050/4322 -- GLOBAL_STEP: 2050
+     | > current_lr: 5e-06  (4.999999999999849e-06)
+     | > loss_text_ce: 0.021688129752874374  (0.023675753405181358)
+     | > loss_mel_ce: 3.111064910888672  (3.1884225934889256)
+     | > loss: 0.03729468211531639  (0.0382392667979002)
+     | > step_time: 0.661  (0.5967303099283349)
+     | > loader_time: 0.0121  (0.012545026569831658)
+
+
+   --> TIME: 2026-10-21 02:27:27 -- STEP: 2100/4322 -- GLOBAL_STEP: 2100
+     | > current_lr: 5e-06  (4.999999999999846e-06)
+     | > loss_text_ce: 0.02272702194750309  (0.02366948241013145)
+     | > loss_mel_ce: 2.94507098197937  (3.183759974752154)
+     | > loss: 0.03533092886209488  (0.038183684774807526)
+     | > step_time: 0.5204  (0.5966567674137289)
+     | > loader_time: 0.0239  (0.012538471562521795)
+
+
+   --> TIME: 2026-10-21 02:28:12 -- STEP: 2150/4322 -- GLOBAL_STEP: 2150
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02649291418492794  (0.023676633604390665)
+     | > loss_mel_ce: 2.8881752490997314  (3.179549894554671)
+     | > loss: 0.03469843044877052  (0.03813364991268448)
+     | > step_time: 0.5882  (0.5967577403090712)
+     | > loader_time: 0.0124  (0.012527329311814417)
+
+
+   --> TIME: 2026-10-21 02:28:57 -- STEP: 2200/4322 -- GLOBAL_STEP: 2200
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.022103603929281235  (0.023670796513218767)
+     | > loss_mel_ce: 3.0208725929260254  (3.1755333759567956)
+     | > loss: 0.03622590750455856  (0.03808576472103598)
+     | > step_time: 0.6283  (0.5968182917074716)
+     | > loader_time: 0.0073  (0.012522149952975183)
+
+
+   --> TIME: 2026-10-21 02:29:42 -- STEP: 2250/4322 -- GLOBAL_STEP: 2250
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.024208255112171173  (0.02366303620818589)
+     | > loss_mel_ce: 2.8740413188934326  (3.1709457555347007)
+     | > loss: 0.03450297191739082  (0.0380310577998559)
+     | > step_time: 0.5892  (0.5967026240030915)
+     | > loader_time: 0.0084  (0.012517317348056365)
+
+
+   --> TIME: 2026-10-21 02:30:27 -- STEP: 2300/4322 -- GLOBAL_STEP: 2300
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.023592893034219742  (0.02365464257157368)
+     | > loss_mel_ce: 3.097675085067749  (3.166035107011379)
+     | > loss: 0.03715795278549194  (0.03797249777809437)
+     | > step_time: 0.6679  (0.596503809534984)
+     | > loader_time: 0.0145  (0.012519893127938976)
+
+
+   --> TIME: 2026-10-21 02:31:12 -- STEP: 2350/4322 -- GLOBAL_STEP: 2350
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.027811063453555107  (0.023658202513259792)
+     | > loss_mel_ce: 2.979349136352539  (3.1619461951357253)
+     | > loss: 0.035799525678157806  (0.037923862631333675)
+     | > step_time: 0.5904  (0.5964998499890579)
+     | > loader_time: 0.0116  (0.012518108854902555)
+
+
+   --> TIME: 2026-10-21 02:31:57 -- STEP: 2400/4322 -- GLOBAL_STEP: 2400
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.024188783019781113  (0.02365141263386856)
+     | > loss_mel_ce: 2.928410053253174  (3.1577433361609764)
+     | > loss: 0.035149987787008286  (0.03787374775623908)
+     | > step_time: 0.6712  (0.5967107424139966)
+     | > loader_time: 0.0071  (0.01251181562741597)
+
+
+   --> TIME: 2026-10-21 02:32:43 -- STEP: 2450/4322 -- GLOBAL_STEP: 2450
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.023117098957300186  (0.02364511824155947)
+     | > loss_mel_ce: 2.958191156387329  (3.1539475653122873)
+     | > loss: 0.035491764545440674  (0.0378284850785015)
+     | > step_time: 0.6193  (0.5967050911455725)
+     | > loader_time: 0.0085  (0.012508083654909713)
+
+
+   --> TIME: 2026-10-21 02:33:27 -- STEP: 2500/4322 -- GLOBAL_STEP: 2500
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02223431132733822  (0.02363246348947285)
+     | > loss_mel_ce: 3.126727342605591  (3.1503854570388774)
+     | > loss: 0.037487637251615524  (0.037785928382724564)
+     | > step_time: 0.6084  (0.5966647499084458)
+     | > loader_time: 0.0132  (0.012506871509552)
+
+
+   --> TIME: 2026-10-21 02:34:13 -- STEP: 2550/4322 -- GLOBAL_STEP: 2550
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02278938516974449  (0.02361871360198537)
+     | > loss_mel_ce: 2.8958399295806885  (3.1467342854481077)
+     | > loss: 0.03474558889865875  (0.03774229835864963)
+     | > step_time: 0.6697  (0.5966710258932658)
+     | > loader_time: 0.0125  (0.012501213503818885)
+
+
+   --> TIME: 2026-10-21 02:34:57 -- STEP: 2600/4322 -- GLOBAL_STEP: 2600
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02256777510046959  (0.02361473225057121)
+     | > loss_mel_ce: 2.9899072647094727  (3.1431978268806695)
+     | > loss: 0.03586279973387718  (0.03770015026299427)
+     | > step_time: 0.6698  (0.5966223816688224)
+     | > loader_time: 0.0102  (0.012498493744776799)
+
+
+   --> TIME: 2026-10-21 02:35:43 -- STEP: 2650/4322 -- GLOBAL_STEP: 2650
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02460002340376377  (0.02361912247426101)
+     | > loss_mel_ce: 3.0308732986450195  (3.1390183373217257)
+     | > loss: 0.03637468442320824  (0.03765044669708551)
+     | > step_time: 0.6612  (0.5969012598721465)
+     | > loader_time: 0.0138  (0.0124975423093112)
+
+
+   --> TIME: 2026-10-21 02:36:28 -- STEP: 2700/4322 -- GLOBAL_STEP: 2700
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.026347188279032707  (0.023611818909369087)
+     | > loss_mel_ce: 2.926762819290161  (3.1353130488925456)
+     | > loss: 0.035156071186065674  (0.037606249166721534)
+     | > step_time: 0.5838  (0.5968237059204643)
+     | > loader_time: 0.0137  (0.012485156500781023)
+
+
+   --> TIME: 2026-10-21 02:37:14 -- STEP: 2750/4322 -- GLOBAL_STEP: 2750
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02484717033803463  (0.023605261645533775)
+     | > loss_mel_ce: 2.905120372772217  (3.131761779004872)
+     | > loss: 0.03488056734204292  (0.03756389408423141)
+     | > step_time: 0.6069  (0.5968708774393233)
+     | > loader_time: 0.0082  (0.012482620152560147)
+
+
+   --> TIME: 2026-10-21 02:37:58 -- STEP: 2800/4322 -- GLOBAL_STEP: 2800
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.023038901388645172  (0.0235968307126313)
+     | > loss_mel_ce: 3.145516872406006  (3.1285898637771554)
+     | > loss: 0.037720903754234314  (0.037526032818880464)
+     | > step_time: 0.6692  (0.59682412947927)
+     | > loader_time: 0.0069  (0.012476146221160889)
+
+
+   --> TIME: 2026-10-21 02:38:43 -- STEP: 2850/4322 -- GLOBAL_STEP: 2850
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.028272511437535286  (0.023586630681366207)
+     | > loss_mel_ce: 2.8701202869415283  (3.1255907545591604)
+     | > loss: 0.03450467810034752  (0.03749020771172485)
+     | > step_time: 0.5  (0.5968013382794545)
+     | > loader_time: 0.0101  (0.012469556457117984)
+
+
+   --> TIME: 2026-10-21 02:39:27 -- STEP: 2900/4322 -- GLOBAL_STEP: 2900
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.023247918114066124  (0.023578409356675226)
+     | > loss_mel_ce: 3.136256456375122  (3.122102714735881)
+     | > loss: 0.0376131497323513  (0.0374485855530305)
+     | > step_time: 0.5847  (0.5964668889703402)
+     | > loader_time: 0.0082  (0.012465520151730242)
+
+
+   --> TIME: 2026-10-21 02:40:12 -- STEP: 2950/4322 -- GLOBAL_STEP: 2950
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.022790543735027313  (0.023573600706028694)
+     | > loss_mel_ce: 3.000525951385498  (3.118728409217571)
+     | > loss: 0.035991862416267395  (0.03740835800451244)
+     | > step_time: 0.5957  (0.596490475767748)
+     | > loader_time: 0.0086  (0.012460524106429794)
+
+
+   --> TIME: 2026-10-21 02:40:57 -- STEP: 3000/4322 -- GLOBAL_STEP: 3000
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.018387440592050552  (0.023560796816522834)
+     | > loss_mel_ce: 2.9166338443756104  (3.1155400846004437)
+     | > loss: 0.03494073078036308  (0.037370249329134846)
+     | > step_time: 0.6231  (0.5963328799406666)
+     | > loader_time: 0.0142  (0.01245480227470398)
+
+
+   --> TIME: 2026-10-21 02:41:43 -- STEP: 3050/4322 -- GLOBAL_STEP: 3050
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.02384212426841259  (0.023554867990925623)
+     | > loss_mel_ce: 2.994652032852173  (3.112538642883296)
+     | > loss: 0.0359344556927681  (0.03733444729484006)
+     | > step_time: 0.5883  (0.5966500426120424)
+     | > loader_time: 0.0144  (0.012449125383721024)
+
+
+   --> TIME: 2026-10-21 02:42:28 -- STEP: 3100/4322 -- GLOBAL_STEP: 3100
+     | > current_lr: 5e-06  (4.999999999999845e-06)
+     | > loss_text_ce: 0.023726437240839005  (0.02355175362659554)
+     | > loss_mel_ce: 3.072147846221924  (3.109775276722442)
+     | > loss: 0.03685564547777176  (0.03730151300168325)
+     | > step_time: 0.6109  (0.5965689244577942)
+     | > loader_time: 0.0235  (0.012446198078893847)
+
+
+   --> TIME: 2026-10-21 02:43:12 -- STEP: 3150/4322 -- GLOBAL_STEP: 3150
+     | > current_lr: 5e-06  (4.999999999999859e-06)
+     | > loss_text_ce: 0.023431098088622093  (0.02354969015256276)
+     | > loss_mel_ce: 3.0573296546936035  (3.106731427889017)
+     | > loss: 0.036675725132226944  (0.03726525213924194)
+     | > step_time: 0.4902  (0.5964068209178846)
+     | > loader_time: 0.0075  (0.012434173160129123)
+
+
+   --> TIME: 2026-10-21 02:43:58 -- STEP: 3200/4322 -- GLOBAL_STEP: 3200
+     | > current_lr: 5e-06  (4.999999999999885e-06)
+     | > loss_text_ce: 0.022733112797141075  (0.0235456599690951)
+     | > loss_mel_ce: 3.0735044479370117  (3.1041155190020753)
+     | > loss: 0.03685997426509857  (0.03723406239121686)
+     | > step_time: 0.7319  (0.5965762753784638)
+     | > loader_time: 0.0084  (0.01243378609418869)
+
+
+   --> TIME: 2026-10-21 02:44:43 -- STEP: 3250/4322 -- GLOBAL_STEP: 3250
+     | > current_lr: 5e-06  (4.999999999999911e-06)
+     | > loss_text_ce: 0.02168012410402298  (0.023540749792296114)
+     | > loss_mel_ce: 3.084882974624634  (3.1013121072328964)
+     | > loss: 0.03698289394378662  (0.037200629989115094)
+     | > step_time: 0.6666  (0.5965495873231133)
+     | > loader_time: 0.0068  (0.01242878349010761)
+
+
+   --> TIME: 2026-10-21 02:45:29 -- STEP: 3300/4322 -- GLOBAL_STEP: 3300
+     | > current_lr: 5e-06  (4.999999999999935e-06)
+     | > loss_text_ce: 0.021574407815933228  (0.02353459311987862)
+     | > loss_mel_ce: 2.7126429080963135  (3.0985339783899666)
+     | > loss: 0.032550208270549774  (0.037167483733571864)
+     | > step_time: 0.5793  (0.5967610989917383)
+     | > loader_time: 0.014  (0.012422258709416245)
+
+
+   --> TIME: 2026-10-21 02:46:14 -- STEP: 3350/4322 -- GLOBAL_STEP: 3350
+     | > current_lr: 5e-06  (4.999999999999959e-06)
+     | > loss_text_ce: 0.022662699222564697  (0.023536021248506928)
+     | > loss_mel_ce: 2.932328224182129  (3.0958801448167215)
+     | > loss: 0.03517846390604973  (0.0371359074755169)
+     | > step_time: 0.6539  (0.596746546190174)
+     | > loader_time: 0.008  (0.01241784928449944)
+
+
+   --> TIME: 2026-10-21 02:46:59 -- STEP: 3400/4322 -- GLOBAL_STEP: 3400
+     | > current_lr: 5e-06  (4.999999999999982e-06)
+     | > loss_text_ce: 0.02598716877400875  (0.023529283185215518)
+     | > loss_mel_ce: 2.8259429931640625  (3.0933126112993983)
+     | > loss: 0.03395155072212219  (0.037105261384893014)
+     | > step_time: 0.6138  (0.5967671413982597)
+     | > loader_time: 0.0154  (0.012412546662723317)
+
+
+   --> TIME: 2026-10-21 02:47:44 -- STEP: 3450/4322 -- GLOBAL_STEP: 3450
+     | > current_lr: 5e-06  (5.000000000000005e-06)
+     | > loss_text_ce: 0.024678675457835197  (0.023523903610589675)
+     | > loss_mel_ce: 2.7041141986846924  (3.0908136884716932)
+     | > loss: 0.03248563036322594  (0.03707544826241076)
+     | > step_time: 0.4569  (0.596860606013863)
+     | > loader_time: 0.0084  (0.01241255193516828)
+
+
+   --> TIME: 2026-10-21 02:48:30 -- STEP: 3500/4322 -- GLOBAL_STEP: 3500
+     | > current_lr: 5e-06  (5.000000000000027e-06)
+     | > loss_text_ce: 0.023256434127688408  (0.023517030794705625)
+     | > loss_mel_ce: 2.9353296756744385  (3.088597696099958)
+     | > loss: 0.03522126376628876  (0.03704898557758757)
+     | > step_time: 0.6356  (0.5969392660004735)
+     | > loader_time: 0.0078  (0.012413677215576171)
+
+
+   --> TIME: 2026-10-21 02:49:15 -- STEP: 3550/4322 -- GLOBAL_STEP: 3550
+     | > current_lr: 5e-06  (5.000000000000048e-06)
+     | > loss_text_ce: 0.0238086748868227  (0.023514846461024887)
+     | > loss_mel_ce: 2.7836387157440186  (3.0861410168526837)
+     | > loss: 0.033421993255615234  (0.0370197133914056)
+     | > step_time: 0.6238  (0.5969673211809599)
+     | > loader_time: 0.0079  (0.012413687101552185)
+
+
+   --> TIME: 2026-10-21 02:50:00 -- STEP: 3600/4322 -- GLOBAL_STEP: 3600
+     | > current_lr: 5e-06  (5.000000000000068e-06)
+     | > loss_text_ce: 0.02341790869832039  (0.023513922610630556)
+     | > loss_mel_ce: 2.905672550201416  (3.0837201721138383)
+     | > loss: 0.034870125353336334  (0.036990882810722624)
+     | > step_time: 0.5812  (0.5969302547640254)
+     | > loader_time: 0.0187  (0.0124112421936459)
+
+
+   --> TIME: 2026-10-21 02:50:46 -- STEP: 3650/4322 -- GLOBAL_STEP: 3650
+     | > current_lr: 5e-06  (5.0000000000000885e-06)
+     | > loss_text_ce: 0.022184323519468307  (0.023513602474066458)
+     | > loss_mel_ce: 2.865701675415039  (3.081556038987143)
+     | > loss: 0.03437959775328636  (0.03696511551074377)
+     | > step_time: 0.6591  (0.5972059504626532)
+     | > loader_time: 0.0087  (0.012405855165768975)
+
+
+   --> TIME: 2026-10-21 02:51:31 -- STEP: 3700/4322 -- GLOBAL_STEP: 3700
+     | > current_lr: 5e-06  (5.000000000000108e-06)
+     | > loss_text_ce: 0.023852912709116936  (0.023514605919150854)
+     | > loss_mel_ce: 2.9706547260284424  (3.07943305814588)
+     | > loss: 0.035648901015520096  (0.036939853875318895)
+     | > step_time: 0.6115  (0.5970250767630484)
+     | > loader_time: 0.0126  (0.012402766330822095)
+
+
+   --> TIME: 2026-10-21 02:52:15 -- STEP: 3750/4322 -- GLOBAL_STEP: 3750
+     | > current_lr: 5e-06  (5.000000000000127e-06)
+     | > loss_text_ce: 0.021222300827503204  (0.023512201261023647)
+     | > loss_mel_ce: 3.0658493041992188  (3.0769362913131677)
+     | > loss: 0.036750853061676025  (0.03691010183642308)
+     | > step_time: 0.6207  (0.5969499096552515)
+     | > loader_time: 0.0065  (0.012400638135274252)
+
+
+   --> TIME: 2026-10-21 02:53:01 -- STEP: 3800/4322 -- GLOBAL_STEP: 3800
+     | > current_lr: 5e-06  (5.000000000000145e-06)
+     | > loss_text_ce: 0.025848623365163803  (0.02351087316331499)
+     | > loss_mel_ce: 2.920400619506836  (3.074521592290774)
+     | > loss: 0.035074397921562195  (0.03688133960442715)
+     | > step_time: 0.54  (0.5970745817611076)
+     | > loader_time: 0.0123  (0.012395272129460386)
+
+
+   --> TIME: 2026-10-21 02:53:46 -- STEP: 3850/4322 -- GLOBAL_STEP: 3850
+     | > current_lr: 5e-06  (5.000000000000163e-06)
+     | > loss_text_ce: 0.02371077984571457  (0.023509672148177344)
+     | > loss_mel_ce: 2.9974076747894287  (3.072430862079963)
+     | > loss: 0.035965695977211  (0.03685643566361108)
+     | > step_time: 0.5916  (0.5971751257661085)
+     | > loader_time: 0.0102  (0.01239291153944932)
+
+
+   --> TIME: 2026-10-21 02:54:32 -- STEP: 3900/4322 -- GLOBAL_STEP: 3900
+     | > current_lr: 5e-06  (5.000000000000181e-06)
+     | > loss_text_ce: 0.023781029507517815  (0.023504801939408716)
+     | > loss_mel_ce: 2.8465991020202637  (3.0702009644263795)
+     | > loss: 0.03417119383811951  (0.03682983128640514)
+     | > step_time: 0.6621  (0.5972338971724862)
+     | > loader_time: 0.0108  (0.01238854414377457)
+
+
+   --> TIME: 2026-10-21 02:55:18 -- STEP: 3950/4322 -- GLOBAL_STEP: 3950
+     | > current_lr: 5e-06  (5.000000000000198e-06)
+     | > loss_text_ce: 0.02363649569451809  (0.023503537125225276)
+     | > loss_mel_ce: 2.942638635635376  (3.068133991277667)
+     | > loss: 0.035312801599502563  (0.036805209405720234)
+     | > step_time: 0.6292  (0.597473596259007)
+     | > loader_time: 0.0184  (0.012386093381084973)
+
+
+   --> TIME: 2026-10-21 02:56:03 -- STEP: 4000/4322 -- GLOBAL_STEP: 4000
+     | > current_lr: 5e-06  (5.000000000000215e-06)
+     | > loss_text_ce: 0.02388708107173443  (0.02349735677009445)
+     | > loss_mel_ce: 2.860335111618042  (3.0659711222052537)
+     | > loss: 0.03433597832918167  (0.03677938738977537)
+     | > step_time: 0.6243  (0.5974928858876214)
+     | > loader_time: 0.0123  (0.012379054605960846)
+
+
+   --> TIME: 2026-10-21 02:56:48 -- STEP: 4050/4322 -- GLOBAL_STEP: 4050
+     | > current_lr: 5e-06  (5.000000000000231e-06)
+     | > loss_text_ce: 0.018802084028720856  (0.02349870981303627)
+     | > loss_mel_ce: 3.0977272987365723  (3.0638770363065895)
+     | > loss: 0.037101540714502335  (0.03675447390081338)
+     | > step_time: 0.6147  (0.59731181203583)
+     | > loader_time: 0.0138  (0.012381412482556002)
+
+
+   --> TIME: 2026-10-21 02:57:33 -- STEP: 4100/4322 -- GLOBAL_STEP: 4100
+     | > current_lr: 5e-06  (5.000000000000246e-06)
+     | > loss_text_ce: 0.022759081795811653  (0.023493108649897107)
+     | > loss_mel_ce: 2.878800392150879  (3.0615965936823555)
+     | > loss: 0.034542378038167953  (0.036727259095031314)
+     | > step_time: 0.7327  (0.5974419159423999)
+     | > loader_time: 0.0186  (0.012378487994031209)
+
+
+   --> TIME: 2026-10-21 02:58:18 -- STEP: 4150/4322 -- GLOBAL_STEP: 4150
+     | > current_lr: 5e-06  (5.000000000000261e-06)
+     | > loss_text_ce: 0.02283826284110546  (0.023487839882811837)
+     | > loss_mel_ce: 2.988551616668701  (3.0595337802243483)
+     | > loss: 0.035849880427122116  (0.03670263906693962)
+     | > step_time: 0.485  (0.597323453512535)
+     | > loader_time: 0.008  (0.012371215820312502)
+
+
+   --> TIME: 2026-10-21 02:59:02 -- STEP: 4200/4322 -- GLOBAL_STEP: 4200
+     | > current_lr: 5e-06  (5.000000000000276e-06)
+     | > loss_text_ce: 0.027317557483911514  (0.023485163285707406)
+     | > loss_mel_ce: 2.6367311477661133  (3.0571146906557503)
+     | > loss: 0.03171486407518387  (0.03667380851249965)
+     | > step_time: 0.4114  (0.5972230173860267)
+     | > loader_time: 0.0125  (0.012365819215774537)
+
+
+   --> TIME: 2026-10-21 02:59:47 -- STEP: 4250/4322 -- GLOBAL_STEP: 4250
+     | > current_lr: 5e-06  (5.000000000000293e-06)
+     | > loss_text_ce: 0.02426888607442379  (0.023483077582629262)
+     | > loss_mel_ce: 2.925359010696411  (3.0548168312521504)
+     | > loss: 0.03511461988091469  (0.036646428215591366)
+     | > step_time: 0.655  (0.5971050168205698)
+     | > loader_time: 0.0173  (0.012364748113295613)
+
+
+   --> TIME: 2026-10-21 03:00:32 -- STEP: 4300/4322 -- GLOBAL_STEP: 4300
+     | > current_lr: 5e-06  (5.000000000000306e-06)
+     | > loss_text_ce: 0.02220184914767742  (0.02347449325891426)
+     | > loss_mel_ce: 2.9004642963409424  (3.053137297519415)
+     | > loss: 0.03479364514350891  (0.036626331571836125)
+     | > step_time: 0.6792  (0.5971289326423813)
+     | > loader_time: 0.0115  (0.012370509824087461)
+
+
+ > EVALUATION 
+
+
+  --> EVAL PERFORMANCE
+     | > avg_loader_time: 0.07410516849783964 (-0.00345826148986815)
+     | > avg_loss_text_ce: 0.021929892559730733 (-0.0007859685479901513)
+     | > avg_loss_mel_ce: 2.8193006404610568 (-0.8320896292841713)
+     | > avg_loss: 2.8412305366161257 (-0.8328755789024886)
+
+ > BEST MODEL : /kaggle/working/coqui-ai-TTS/run/training/GPT_XTTS_v2.0_LJSpeech_FT-October-21-2026_01+52AM-03339e5/best_model_4323.pth