wav2vec2-base-finetuned-sentiment-mesd-v9

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3500
• Accuracy: 0.9154

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: 64
• eval_batch_size: 40
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.01
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.86	3	1.7825	0.1846
1.9553	1.86	6	1.7212	0.4308
1.9553	2.86	9	1.6164	0.3769
2.002	3.86	12	1.4904	0.3769
1.6191	4.86	15	1.4426	0.4385
1.6191	5.86	18	1.3516	0.5231
1.6209	6.86	21	1.2176	0.5538
1.6209	7.86	24	1.1683	0.5692
1.371	8.86	27	1.0885	0.5923
1.1568	9.86	30	1.0152	0.6385
1.1568	10.86	33	0.9289	0.6385
1.1023	11.86	36	0.9141	0.6308
1.1023	12.86	39	0.8526	0.6462
0.9448	13.86	42	0.8420	0.6769
0.7972	14.86	45	0.7976	0.6692
0.7972	15.86	48	0.8192	0.7308
0.7793	16.86	51	0.7108	0.7615
0.7793	17.86	54	0.6712	0.7769
0.6468	18.86	57	0.6684	0.7923
0.5083	19.86	60	0.6922	0.7385
0.5083	20.86	63	0.6148	0.7923
0.4988	21.86	66	0.5846	0.7923
0.4988	22.86	69	0.6050	0.8154
0.4123	23.86	72	0.5506	0.7846
0.3511	24.86	75	0.6095	0.7846
0.3511	25.86	78	0.5916	0.8154
0.3268	26.86	81	0.5912	0.8077
0.3268	27.86	84	0.5142	0.8538
0.3036	28.86	87	0.5492	0.8077
0.3066	29.86	90	0.6007	0.8231
0.3066	30.86	93	0.5748	0.8231
0.2538	31.86	96	0.6027	0.7692
0.2538	32.86	99	0.6979	0.7462
0.2281	33.86	102	0.7002	0.7615
0.2183	34.86	105	0.6650	0.7769
0.2183	35.86	108	0.5192	0.8462
0.2202	36.86	111	0.5389	0.8308
0.2202	37.86	114	0.5050	0.8385
0.1906	38.86	117	0.5722	0.7769
0.154	39.86	120	0.5239	0.8308
0.154	40.86	123	0.4448	0.8615
0.1474	41.86	126	0.4623	0.8615
0.1474	42.86	129	0.4282	0.8615
0.1345	43.86	132	0.5087	0.8615
0.1567	44.86	135	0.4859	0.8385
0.1567	45.86	138	0.6603	0.8077
0.1731	46.86	141	0.5379	0.8385
0.1731	47.86	144	0.8666	0.7538
0.1606	48.86	147	0.7518	0.8
0.1484	49.86	150	0.5986	0.8385
0.1484	50.86	153	0.6368	0.8231
0.2256	51.86	156	0.4639	0.8692
0.2256	52.86	159	0.5533	0.8462
0.1178	53.86	162	0.5038	0.8615
0.0815	54.86	165	0.5052	0.8692
0.0815	55.86	168	0.4337	0.8846
0.0998	56.86	171	0.4422	0.8769
0.0998	57.86	174	0.4317	0.8692
0.0855	58.86	177	0.4025	0.8923
0.0962	59.86	180	0.4605	0.8769
0.0962	60.86	183	0.4356	0.8769
0.0763	61.86	186	0.4614	0.8769
0.0763	62.86	189	0.4382	0.8846
0.0902	63.86	192	0.4701	0.8692
0.0654	64.86	195	0.4922	0.8692
0.0654	65.86	198	0.5413	0.8538
0.0651	66.86	201	0.5759	0.8615
0.0651	67.86	204	0.4238	0.9
0.0822	68.86	207	0.3500	0.9154
0.0625	69.86	210	0.3878	0.8923
0.0625	70.86	213	0.4952	0.8615
0.0548	71.86	216	0.4544	0.8615
0.0548	72.86	219	0.5497	0.8769
0.054	73.86	222	0.4434	0.8846
0.0543	74.86	225	0.4732	0.8769
0.0543	75.86	228	0.4425	0.8923
0.0881	76.86	231	0.4788	0.8769
0.0881	77.86	234	0.5448	0.8769
0.061	78.86	237	0.4221	0.9077
0.0567	79.86	240	0.4404	0.8769
0.0567	80.86	243	0.4099	0.9
0.052	81.86	246	0.5259	0.8769
0.052	82.86	249	0.5874	0.8692
0.0444	83.86	252	0.5555	0.8846
0.0332	84.86	255	0.5156	0.8615
0.0332	85.86	258	0.4564	0.8615
0.0449	86.86	261	0.4826	0.8692
0.0449	87.86	264	0.4726	0.8615
0.0385	88.86	267	0.4206	0.8846
0.0356	89.86	270	0.4050	0.8769
0.0356	90.86	273	0.4161	0.8923
0.0391	91.86	276	0.4100	0.9077
0.0391	92.86	279	0.4047	0.9
0.0249	93.86	282	0.4044	0.9
0.0399	94.86	285	0.3968	0.8846
0.0399	95.86	288	0.3802	0.9
0.031	96.86	291	0.3689	0.9
0.031	97.86	294	0.3616	0.9077
0.036	98.86	297	0.3584	0.9077
0.0386	99.86	300	0.3574	0.9077

Framework versions

• Transformers 4.17.0
• Pytorch 1.10.0+cu111
• Datasets 2.0.0
• Tokenizers 0.11.6