whisper-small-multilingual-spoken-ner-pipeline-step-2

qmeeus/whisper-small-multilingual-spoken-ner-pipeline-step-2

whisper-small-multilingual-spoken-ner-pipeline-step-2 is a machine learning model from qmeeus. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as apache-2.0.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

21% of all-time downloads

All-time downloads

Public

Parameters

253M

58.3 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.pt33 GB · 56%

At a glance

Library: transformers
License: apache-2.0
Model type: whisper_for_slu
Access: Public
Created: Jan 9, 2024
Updated: Jan 9, 2024
SHA: 6cec6e26

Base models

openai/whisper-small

Datasets

facebook/voxpopuli

Model card

From the Hugging Face model README

WhisperForSpokenNER

This model is a fine-tuned version of openai/whisper-small on the facebook/voxpopuli de+es+fr+nl dataset. It achieves the following results on the evaluation set:

Loss: 0.3166
F1 Score: 0.7276
Label F1: 0.8546
Wer: 0.0888

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: 0.0001
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 16
total_train_batch_size: 128
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 500
training_steps: 5000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Score	Label F1	Wer
0.2754	0.36	200	0.2577	0.4922	0.6581	0.0988
0.2461	0.71	400	0.2499	0.6282	0.7808	0.1028
0.2196	1.07	600	0.2557	0.6825	0.8146	0.1107
0.1824	1.43	800	0.2517	0.6783	0.8189	0.1037
0.1852	1.79	1000	0.2455	0.6880	0.8274	0.1018
0.1152	2.14	1200	0.2439	0.7038	0.8434	0.1012
0.1012	2.5	1400	0.2441	0.7165	0.8428	0.0969
0.1076	2.86	1600	0.2430	0.7052	0.8484	0.0989
0.0487	3.22	1800	0.2527	0.7069	0.8418	0.0924
0.0504	3.57	2000	0.2532	0.7041	0.8481	0.0935
0.0527	3.93	2200	0.2567	0.7073	0.8450	0.0953
0.0191	4.29	2400	0.2702	0.7273	0.8596	0.0915
0.0192	4.65	2600	0.2691	0.7162	0.8535	0.0920
0.0196	5.0	2800	0.2727	0.7175	0.8539	0.0910
0.0072	5.36	3000	0.2854	0.7333	0.8550	0.0899
0.0068	5.72	3200	0.2888	0.7247	0.8507	0.0902
0.0053	6.08	3400	0.2980	0.7281	0.8559	0.0884
0.0035	6.43	3600	0.3029	0.7201	0.8589	0.0886
0.0034	6.79	3800	0.3061	0.724	0.8544	0.0893
0.0026	7.15	4000	0.3111	0.7239	0.8534	0.0885
0.0023	7.51	4200	0.3137	0.7269	0.8522	0.0887
0.0023	7.86	4400	0.3145	0.7255	0.8542	0.0889
0.002	8.22	4600	0.3159	0.7268	0.8534	0.0889
0.002	8.58	4800	0.3166	0.7257	0.8559	0.0888
0.002	8.94	5000	0.3166	0.7276	0.8546	0.0888

Framework versions

Transformers 4.37.0.dev0
Pytorch 2.1.0
Datasets 2.14.6
Tokenizers 0.14.1

Library: transformers
Type: whisper_for_slu
License: apache-2.0
Created: Jan 9, 2024
Updated: Jan 9, 2024