metadata

license: apache-2.0
base_model: openai/whisper-small
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: >-
      Teamim-small_WeightDecay-0.05_Augmented_Combined-Data_date-11-07-2024_12-42
    results: []

Teamim-small_WeightDecay-0.05_Augmented_Combined-Data_date-11-07-2024_12-42

This model is a fine-tuned version of openai/whisper-small on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1474
Wer: 9.8084
Avg Precision Exact: 0.9128
Avg Recall Exact: 0.9142
Avg F1 Exact: 0.9131
Avg Precision Letter Shift: 0.9270
Avg Recall Letter Shift: 0.9286
Avg F1 Letter Shift: 0.9274
Avg Precision Word Level: 0.9292
Avg Recall Word Level: 0.9309
Avg F1 Word Level: 0.9297
Avg Precision Word Shift: 0.9711
Avg Recall Word Shift: 0.9735
Avg F1 Word Shift: 0.9719
Precision Median Exact: 1.0
Recall Median Exact: 1.0
F1 Median Exact: 1.0
Precision Max Exact: 1.0
Recall Max Exact: 1.0
F1 Max Exact: 1.0
Precision Min Exact: 0.0
Recall Min Exact: 0.0
F1 Min Exact: 0.0
Precision Min Letter Shift: 0.0
Recall Min Letter Shift: 0.0
F1 Min Letter Shift: 0.0
Precision Min Word Level: 0.0
Recall Min Word Level: 0.0
F1 Min Word Level: 0.0
Precision Min Word Shift: 0.1429
Recall Min Word Shift: 0.125
F1 Min Word Shift: 0.1333

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: 1e-05
train_batch_size: 8
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 200000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Avg Precision Exact	Avg Recall Exact	Avg F1 Exact	Avg Precision Letter Shift	Avg Recall Letter Shift	Avg F1 Letter Shift	Avg Precision Word Level	Avg Recall Word Level	Avg F1 Word Level	Avg Precision Word Shift	Avg Recall Word Shift	Avg F1 Word Shift	Precision Median Exact	Recall Median Exact	F1 Median Exact	Precision Max Exact	Recall Max Exact	F1 Max Exact	Precision Min Word Shift	Recall Min Word Shift	F1 Min Word Shift
No log	0.0001	1	9.9314	148.2331	0.0007	0.0023	0.0010	0.0225	0.0232	0.0218	0.0100	0.0586	0.0155	0.1303	0.1348	0.1270	0.0	0.0	0.0	0.1	0.5	0.1538	0.0	0.0	0.0
0.048	0.5167	10000	0.1206	17.0868	0.8509	0.8591	0.8544	0.8713	0.8797	0.8749	0.8748	0.8830	0.8783	0.9382	0.9482	0.9423	0.9231	0.9231	0.9286	1.0	1.0	1.0	0.0	0.0	0.0
0.0182	1.0334	20000	0.1214	14.3208	0.8671	0.8696	0.8679	0.8853	0.8879	0.8861	0.8885	0.8911	0.8893	0.9504	0.9534	0.9513	0.9286	0.9333	0.9474	1.0	1.0	1.0	0.0	0.0	0.0
0.0136	1.5501	30000	0.1261	13.5939	0.8761	0.8797	0.8775	0.8943	0.8981	0.8957	0.8971	0.9008	0.8984	0.9531	0.9572	0.9546	0.9412	1.0	0.9600	1.0	1.0	1.0	0.0	0.0	0.0
0.0039	2.0668	40000	0.1274	12.8292	0.8843	0.8873	0.8853	0.9000	0.9033	0.9012	0.9024	0.9057	0.9036	0.9551	0.9597	0.9569	1.0	1.0	0.9630	1.0	1.0	1.0	0.0	0.0	0.0
0.0034	2.5834	50000	0.1294	12.4579	0.8877	0.8890	0.8879	0.9032	0.9046	0.9035	0.9059	0.9078	0.9064	0.9593	0.9618	0.9600	1.0	1.0	0.9655	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0067	3.1001	60000	0.1309	11.7593	0.8955	0.8991	0.8969	0.9115	0.9152	0.9129	0.9139	0.9177	0.9154	0.9626	0.9675	0.9646	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.1	0.1176
0.0048	3.6168	70000	0.1306	11.5926	0.8960	0.8986	0.8969	0.9125	0.9152	0.9134	0.9152	0.9178	0.9161	0.9637	0.9670	0.9648	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.1111	0.125
0.0022	4.1335	80000	0.1356	11.8884	0.8918	0.8928	0.8919	0.9073	0.9084	0.9074	0.9095	0.9106	0.9096	0.9605	0.9632	0.9613	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.002	4.6502	90000	0.1348	11.0356	0.9012	0.9002	0.9004	0.9155	0.9145	0.9146	0.9181	0.9170	0.9172	0.9666	0.9664	0.9661	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.1111	0.125
0.001	5.1669	100000	0.1398	11.1929	0.8983	0.9002	0.8989	0.9134	0.9154	0.9140	0.9156	0.9176	0.9162	0.9648	0.9678	0.9658	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0016	5.6836	110000	0.1426	10.9821	0.9019	0.9032	0.9022	0.9168	0.9182	0.9171	0.9191	0.9207	0.9196	0.9644	0.9670	0.9652	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0002	6.2003	120000	0.1417	10.5384	0.9032	0.9045	0.9035	0.9179	0.9193	0.9182	0.9203	0.9214	0.9205	0.9680	0.9701	0.9686	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0004	6.7170	130000	0.1439	10.5604	0.9022	0.9048	0.9031	0.9167	0.9195	0.9177	0.9190	0.9218	0.9200	0.9656	0.9699	0.9673	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0005	7.2336	140000	0.1460	10.4188	0.9048	0.9056	0.9048	0.9193	0.9202	0.9194	0.9215	0.9224	0.9216	0.9673	0.9696	0.9681	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0002	7.7503	150000	0.1396	10.1860	0.9068	0.9072	0.9066	0.9203	0.9209	0.9203	0.9228	0.9233	0.9227	0.9687	0.9705	0.9691	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0001	8.2670	160000	0.1442	10.1136	0.9074	0.9068	0.9068	0.9214	0.9209	0.9208	0.9237	0.9234	0.9232	0.9697	0.9701	0.9695	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0005	8.7837	170000	0.1432	9.9688	0.9098	0.9107	0.9099	0.9243	0.9253	0.9244	0.9266	0.9277	0.9268	0.9702	0.9717	0.9705	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0	9.3004	180000	0.1467	10.0538	0.9093	0.9098	0.9092	0.9239	0.9246	0.9239	0.9261	0.9268	0.9261	0.9703	0.9716	0.9705	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0	9.8171	190000	0.1449	9.9342	0.9119	0.9125	0.9118	0.9262	0.9269	0.9261	0.9283	0.9288	0.9282	0.9702	0.9715	0.9704	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333
0.0001	10.3338	200000	0.1474	9.8084	0.9128	0.9142	0.9131	0.9270	0.9286	0.9274	0.9292	0.9309	0.9297	0.9711	0.9735	0.9719	1.0	1.0	1.0	1.0	1.0	1.0	0.1429	0.125	0.1333

Framework versions

Transformers 4.41.2
Pytorch 2.2.1
Datasets 2.20.0
Tokenizers 0.19.1