sin2piusc/whisper-medium-5k-jp

whisper-medium-5k

This model is a fine-tuned version of openai/whisper-medium on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1389

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

--Original sentence:

集団内のすべての個体が特定の表現形質に関して同一である場合 それらは単形性と呼ばれます。

When all individuals in a population are identical with respect to a particular phenotypic trait, they are called monomorphic.

--sin2piusc/whisper-medium-5ksteps:

集団内のすべての個体が特定の表現形質に関して同一である場合 それらは単形性と呼ばれます

When all individuals in a population are identical with respect to a particular phenotypic trait, they are called monomorphic.

--openai/whisper-medium:

集団内のすべての個体が特定の表現形式に関して同一である場合、それらは単形性と呼ばれます。

If all individuals in a population are identical with respect to a particular form of expression, they are called monomorphic.

---

--sin2piusc/whisper-medium-5ksteps:

When I drink alcohol, I can become quite unsightly, so I ordered a glass of water and stopped drinking.

--openai/whisper-medium:

I don't like drinking alcohol, so I asked for water and avoided it.

--Original sentence:

I can be quite unsightly when I'm drunk, so I abstained from alcohol and mainly drank water.

Training procedure

On a laptop running windows.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 8
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.3
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.3706	0.3697	200	1.1454
0.7963	0.7394	400	0.5219
0.2503	1.1091	600	0.2178
0.2062	1.4787	800	0.2005
0.1867	1.8484	1000	0.1869
0.1644	2.2181	1200	0.1738
0.1501	2.5878	1400	0.1630
0.1386	2.9575	1600	0.1524
0.1186	3.3272	1800	0.1458
0.1086	3.6969	2000	0.1424
0.1019	4.0665	2200	0.1364
0.0871	4.4362	2400	0.1347
0.085	4.8059	2600	0.1326
0.0746	5.1756	2800	0.1336
0.0729	5.5453	3000	0.1312
0.0688	5.9150	3200	0.1316
0.0598	6.2847	3400	0.1328
0.0574	6.6543	3600	0.1340
0.0598	7.0240	3800	0.1336
0.0481	7.3937	4000	0.1356
0.0514	7.7634	4200	0.1366
0.0465	8.1331	4400	0.1382
0.0428	8.5028	4600	0.1378
0.043	8.8725	4800	0.1384
0.0425	9.2421	5000	0.1389

Framework versions

• PEFT 0.10.0
• Transformers 4.41.0.dev0
• Pytorch 2.2.2+cu118
• Datasets 2.19.0
• Tokenizers 0.19.1