FFZG-cleopatra/M2SA-text-only

cardiffnlp-twitter-xlmr-finetuned-txtnly-all-42

This model is a fine-tuned version of cardiffnlp/twitter-xlm-roberta-base-sentiment-multilingual on the all dataset. It achieves the following results on the evaluation set:

• Loss: 0.6972
• Precision: 0.6687
• Recall: 0.6729
• F1: 0.6703

Model description

More information needed

Usage

To use the model use the following script. Kindly set the device based on availability of the GPU.

from transformers import (pipeline)

analyzer = pipeline(
    "sentiment-analysis", model="FFZG-cleopatra/M2SA-text-only"
)

input_text = "I feel amazing today."
print(analyzer(input_text)[0]["label"])

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1
0.6122	0.06	500	0.8542	0.6559	0.4905	0.4841
0.5497	0.12	1000	0.8037	0.7044	0.6070	0.6209
0.5404	0.18	1500	0.9700	0.5591	0.4176	0.3652
0.5165	0.24	2000	0.7449	0.7349	0.5297	0.5369
0.5136	0.3	2500	0.7885	0.6766	0.5025	0.5001
0.5072	0.36	3000	0.8124	0.6076	0.6132	0.5917
0.5011	0.42	3500	0.8767	0.6427	0.5987	0.5784
0.5021	0.48	4000	0.7958	0.6848	0.6362	0.6503
0.4946	0.54	4500	0.8045	0.7220	0.4968	0.4983
0.4928	0.6	5000	0.7803	0.7582	0.5381	0.5503
0.5008	0.66	5500	0.7507	0.4407	0.4798	0.4594
0.4966	0.72	6000	0.8239	0.6140	0.6767	0.6311
0.4791	0.78	6500	0.7028	0.6568	0.5206	0.5413
0.494	0.84	7000	0.8034	0.6660	0.5189	0.5227
0.4861	0.9	7500	0.9003	0.5781	0.4785	0.4541
0.4804	0.96	8000	0.7740	0.6239	0.5775	0.5792
0.4614	1.02	8500	0.7397	0.6848	0.6312	0.6471
0.4315	1.08	9000	0.7889	0.6642	0.6035	0.6149
0.4506	1.14	9500	0.8784	0.6387	0.5017	0.4968
0.4489	1.2	10000	0.7994	0.5340	0.4964	0.4949
0.4466	1.26	10500	0.8110	0.5776	0.4735	0.4464
0.4319	1.32	11000	0.8069	0.6612	0.5399	0.5481
0.4243	1.38	11500	0.7942	0.5948	0.5705	0.5797
0.4398	1.44	12000	0.9738	0.5370	0.6070	0.5247
0.4526	1.5	12500	0.7196	0.7046	0.5478	0.5590
0.4529	1.56	13000	0.8050	0.6419	0.5731	0.5863
0.446	1.62	13500	0.7564	0.6521	0.5912	0.6107
0.4315	1.68	14000	0.7515	0.6475	0.6069	0.6212
0.4464	1.74	14500	0.8308	0.6276	0.5513	0.5599
0.4423	1.8	15000	0.7982	0.6176	0.5937	0.5992
0.4551	1.86	15500	0.8223	0.6356	0.5934	0.6020
0.4408	1.92	16000	0.7691	0.6088	0.5147	0.5131
0.4389	1.98	16500	0.6972	0.6687	0.6729	0.6703
0.3886	2.04	17000	0.7798	0.6126	0.5437	0.5543
0.3883	2.1	17500	0.8385	0.5948	0.6225	0.5978
0.4011	2.16	18000	0.7755	0.6551	0.5787	0.5915
0.3992	2.22	18500	0.7886	0.5582	0.5519	0.5472
0.393	2.28	19000	0.7660	0.5901	0.5923	0.5889
0.3891	2.34	19500	0.7702	0.5792	0.5331	0.5354
0.4119	2.41	20000	0.8545	0.5406	0.5243	0.5111
0.3981	2.47	20500	0.8641	0.5695	0.5536	0.5364
0.4	2.53	21000	0.8045	0.5988	0.5845	0.5822
0.4059	2.59	21500	0.8023	0.6301	0.5549	0.5696
0.3805	2.65	22000	0.8242	0.5633	0.5363	0.5387
0.4126	2.71	22500	0.8866	0.5630	0.5244	0.5253
0.3959	2.77	23000	0.9228	0.6486	0.5570	0.5716
0.3972	2.83	23500	0.8297	0.6415	0.6336	0.6330
0.3779	2.89	24000	0.8683	0.6023	0.5920	0.5897
0.3951	2.95	24500	0.8628	0.5892	0.5116	0.5125
0.3916	3.01	25000	0.9203	0.6305	0.5026	0.5024
0.3524	3.07	25500	0.9825	0.6089	0.5039	0.5011
0.3332	3.13	26000	0.8755	0.5980	0.5712	0.5814
0.3517	3.19	26500	0.9922	0.6701	0.5941	0.6181
0.3534	3.25	27000	0.9573	0.5653	0.5175	0.5243
0.3544	3.31	27500	0.9827	0.5739	0.5531	0.5551
0.3526	3.37	28000	0.9517	0.6019	0.4737	0.4657
0.3448	3.43	28500	0.9559	0.5744	0.5138	0.5232
0.3662	3.49	29000	0.8470	0.6417	0.6176	0.6173
0.3502	3.55	29500	0.8524	0.6606	0.5776	0.5912
0.3733	3.61	30000	0.9210	0.5578	0.5555	0.5466
0.3424	3.67	30500	0.9295	0.5863	0.6100	0.5809
0.3591	3.73	31000	0.9707	0.5828	0.4769	0.4588
0.3634	3.79	31500	0.8524	0.6136	0.5681	0.5752

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2