tiny | whisper_finetuning Overview – Weights & Biases

Skip to main content

Parambharat's group workspace

Group: tiny

33

1-1

of 1

Tags

Notes

Author

State

Finished

Start time

December 8th, 2022 7:13:35 PM

Runtime

23m 34s

Tracked hours

17m 47s

Run path

parambharat/whisper_finetuning/wof84bz2

OS

Linux-5.15.0-56-generic-x86_64-with-glibc2.17

Python version

3.8.13

Command

whisper_base_ta

System Hardware

CPU count	10
Logical CPU count	20
GPU count	1
GPU type	NVIDIA GeForce RTX 3060

W&B CLI Version

0.13.5

Group

Job Type

fine-tuning

Config parameters are your model's inputs. Learn more

▶
Config parameters:{} 186 keys
- _name_or_path:
  "whisper-tiny-ta/"
- activation_dropout:
  0
- activation_function:
  "gelu"
- adafactor:
  false
- adam_beta1:
  0.9
- adam_beta2:
  0.999
- adam_epsilon:
  0.00000001
- add_cross_attention:
  false
- ▶
  architectures:[] 1 item
  - 0:
    "WhisperForConditionalGeneration"
- attention_dropout:
  0
- auto_find_batch_size:
  false
- bad_words_ids:
  null
- ▶
  begin_suppress_tokens:[] 2 items
  - 0:
    220
  - 1:
    50,257
- bf16:
  false
- bf16_full_eval:
  false
- bos_token_id:
  50,257
- chunk_size_feed_forward:
  0
- cross_attention_hidden_size:
  null
- d_model:
  384
- data_seed:
  "None"
- dataloader_drop_last:
  false
- dataloader_num_workers:
  0
- dataloader_pin_memory:
  true
- ddp_bucket_cap_mb:
  "None"
- ddp_find_unused_parameters:
  "None"
- ddp_timeout:
  1,800
- debug:
  "[]"
- decoder_attention_heads:
  6
- decoder_ffn_dim:
  1,536
- decoder_layerdrop:
  0
- decoder_layers:
  4
- decoder_start_token_id:
  50,258
- deepspeed:
  "None"
- disable_tqdm:
  false
- diversity_penalty:
  0
- do_eval:
  true
- do_predict:
  false
- do_sample:
  false
- do_train:
  false
- dropout:
  0
- early_stopping:
  false
- encoder_attention_heads:
  6
- encoder_ffn_dim:
  1,536
- encoder_layerdrop:
  0
- encoder_layers:
  4
- encoder_no_repeat_ngram_size:
  0
- warmup_ratio:
  0
- warmup_steps:
  1,000
- weight_decay:
  0
- xpu_backend:
  "None"

Summary metrics are your model's outputs. Learn more

▶
Summary metrics:{} 10 keys
- eval/loss:
  0.3562886714935303
- eval/runtime:
  4.0936
- eval/samples_per_second:
  2.443
- eval/steps_per_second:
  0.244
- eval/wer:
  35.82887700534759
- sample_predictions:
  "table-file"
- train/epoch:
  0
- train/global_step:
  20
- train/learning_rate:
  0.0000002
- train/loss:
  0.3549

This run produced these artifacts as outputs. Total: 10. Learn more