Whisper

🦥 Unsloth: Will patch your computer to enable 2x faster free finetuning.
🦥 Unsloth Zoo will now patch everything to make training faster!
==((====))==  Unsloth 2025.8.10: Fast Whisper patching. Transformers: 4.51.3.
   \\   /|    Tesla T4. Num GPUs = 1. Max memory: 14.741 GB. Platform: Linux.
O^O/ \_/ \    Torch: 2.8.0+cu126. CUDA: 7.5. CUDA Toolkit: 12.6. Triton: 3.4.0
\        /    Bfloat16 = FALSE. FA [Xformers = 0.0.32.post2. FA2 = False]
 "-____-"     Free license: http://github.com/unslothai/unsloth
Unsloth: Fast downloading is enabled - ignore downloading bars which are red colored!
Unsloth: QLoRA and full finetuning all not selected. Switching to 16bit LoRA.

model.safetensors:   0%|          | 0.00/3.09G [00:00<?, ?B/s]

generation_config.json: 0.00B [00:00, ?B/s]

preprocessor_config.json:   0%|          | 0.00/357 [00:00<?, ?B/s]

tokenizer_config.json: 0.00B [00:00, ?B/s]

vocab.json: 0.00B [00:00, ?B/s]

merges.txt: 0.00B [00:00, ?B/s]

normalizer.json: 0.00B [00:00, ?B/s]

added_tokens.json: 0.00B [00:00, ?B/s]

special_tokens_map.json: 0.00B [00:00, ?B/s]

unsloth/whisper-large-v3 does not have a padding token! Will use pad_token = <|endoftext|>.

Unsloth: Making `model.base_model.model.model.encoder` require gradients

README.md: 0.00B [00:00, ?B/s]

data/train-00000-of-00001.parquet:   0%|          | 0.00/328M [00:00<?, ?B/s]

Generating train split:   0%|          | 0/1195 [00:00<?, ? examples/s]

Train split: 100%|██████████| 1123/1123 [00:47<00:00, 23.69it/s]
Test split: 100%|██████████| 72/72 [00:01<00:00, 50.11it/s]

Downloading builder script: 0.00B [00:00, ?B/s]

/tmp/ipython-input-715647589.py:3: FutureWarning: `tokenizer` is deprecated and will be removed in version 5.0.0 for `Seq2SeqTrainer.__init__`. Use `processing_class` instead.
  trainer = Seq2SeqTrainer(

GPU = Tesla T4. Max memory = 14.741 GB.
3.012 GB of memory reserved.

==((====))==  Unsloth - 2x faster free finetuning | Num GPUs used = 1
   \\   /|    Num examples = 1,123 | Num Epochs = 1 | Total steps = 60
O^O/ \_/ \    Batch size per device = 1 | Gradient accumulation steps = 4
\        /    Data Parallel GPUs = 1 | Total batch size (1 x 4 x 1) = 4
 "-____-"     Trainable parameters = 31,457,280 of 1,574,947,840 (2.00% trained)
Unsloth: Not an error, but WhisperForConditionalGeneration does not accept `num_items_in_batch`.
Using gradient accumulation will be very slightly less accurate.
Read more on gradient accumulation issues here: https://unsloth.ai/blog/gradient
Passing a tuple of `past_key_values` is deprecated and will be removed in Transformers v4.43.0. You should pass an instance of `EncoderDecoderCache` instead, e.g. `past_key_values=EncoderDecoderCache.from_legacy_cache(past_key_values)`.
`use_cache = True` is incompatible with gradient checkpointing. Setting `use_cache = False`...

Unsloth: Will smartly offload gradients to save VRAM!

422.5309 seconds used for training.
7.04 minutes used for training.
Peak reserved memory = 10.348 GB.
Peak reserved memory for training = 7.336 GB.
Peak reserved memory % of max memory = 70.199 %.
Peak reserved memory for training % of max memory = 49.766 %.

--2025-08-30 12:08:09--  https://upload.wikimedia.org/wikipedia/commons/5/5b/Speech_12dB_s16.flac
Resolving upload.wikimedia.org (upload.wikimedia.org)... 198.35.26.112, 2620:0:863:ed1a::2:b
Connecting to upload.wikimedia.org (upload.wikimedia.org)|198.35.26.112|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 495373 (484K) [audio/x-flac]
Saving to: ‘Speech_12dB_s16.flac’

Speech_12dB_s16.fla 100%[===================>] 483.76K  --.-KB/s    in 0.09s   

2025-08-30 12:08:09 (5.32 MB/s) - ‘Speech_12dB_s16.flac’ saved [495373/495373]

Device set to use cuda:0
/usr/local/lib/python3.12/dist-packages/transformers/models/whisper/generation_whisper.py:573: FutureWarning: The input name `inputs` is deprecated. Please make sure to use `input_features` instead.
  warnings.warn(
`generation_config` default values have been modified to match model-specific defaults: {'suppress_tokens': [1, 2, 7, 8, 9, 10, 14, 25, 26, 27, 28, 29, 31, 58, 59, 60, 61, 62, 63, 90, 91, 92, 93, 359, 503, 522, 542, 873, 893, 902, 918, 922, 931, 1350, 1853, 1982, 2460, 2627, 3246, 3253, 3268, 3536, 3846, 3961, 4183, 4667, 6585, 6647, 7273, 9061, 9383, 10428, 10929, 11938, 12033, 12331, 12562, 13793, 14157, 14635, 15265, 15618, 16553, 16604, 18362, 18956, 20075, 21675, 22520, 26130, 26161, 26435, 28279, 29464, 31650, 32302, 32470, 36865, 42863, 47425, 49870, 50254, 50258, 50359, 50360, 50361, 50362, 50363], 'begin_suppress_tokens': [220, 50257]}. If this is not desired, please set these values explicitly.
A custom logits processor of type <class 'transformers.generation.logits_process.SuppressTokensLogitsProcessor'> has been passed to `.generate()`, but it was also created in `.generate()`, given its parameterization. The custom <class 'transformers.generation.logits_process.SuppressTokensLogitsProcessor'> will take precedence. Please check the docstring of <class 'transformers.generation.logits_process.SuppressTokensLogitsProcessor'> to see related `.generate()` flags.
A custom logits processor of type <class 'transformers.generation.logits_process.SuppressTokensAtBeginLogitsProcessor'> has been passed to `.generate()`, but it was also created in `.generate()`, given its parameterization. The custom <class 'transformers.generation.logits_process.SuppressTokensAtBeginLogitsProcessor'> will take precedence. Please check the docstring of <class 'transformers.generation.logits_process.SuppressTokensAtBeginLogitsProcessor'> to see related `.generate()` flags.

 The birch canoe slid on the smooth planks. Glued the sheet to the dark blue background. It's easy to tell the depth of a well. Four hours of steady work faced us.

[]