bertweet-base-sentiment-tuned

This model is a fine-tuned version of vinai/bertweet-base on the EPFL CS-433 Text Classification dataset. It achieves the following results on the evaluation set:

• Loss: 0.2120
• Accuracy: 0.9126
• F1: 0.9126
• Precision: 0.9127
• Recall: 0.9126

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.5687	0.0400	707	0.3330	0.8668	0.8668	0.8672	0.8668
0.3066	0.0801	1414	0.2736	0.8852	0.8852	0.8855	0.8852
0.2733	0.1201	2121	0.2515	0.895	0.8950	0.8950	0.895
0.26	0.1601	2828	0.2419	0.8993	0.8993	0.8994	0.8993
0.254	0.2002	3535	0.2340	0.9035	0.9035	0.9035	0.9035
0.243	0.2402	4242	0.2333	0.9023	0.9023	0.9024	0.9023
0.2412	0.2802	4949	0.2306	0.901	0.9010	0.9015	0.901
0.2405	0.3203	5656	0.2281	0.9048	0.9048	0.9049	0.9048
0.233	0.3603	6363	0.2253	0.9071	0.9071	0.9073	0.9071
0.2357	0.4003	7070	0.2250	0.9073	0.9073	0.9079	0.9073
0.2321	0.4403	7777	0.2245	0.9051	0.9051	0.9051	0.9051
0.2335	0.4804	8484	0.2325	0.9029	0.9028	0.9045	0.9029
0.2341	0.5204	9191	0.2229	0.9082	0.9082	0.9083	0.9082
0.2295	0.5604	9898	0.2187	0.9087	0.9087	0.9088	0.9087
0.2281	0.6005	10605	0.2228	0.9055	0.9055	0.9058	0.9055
0.2293	0.6405	11312	0.2188	0.9087	0.9087	0.9087	0.9087
0.2286	0.6805	12019	0.2188	0.9087	0.9087	0.9087	0.9087
0.2262	0.7206	12726	0.2183	0.9105	0.9105	0.9105	0.9105
0.2255	0.7606	13433	0.2176	0.9082	0.9082	0.9084	0.9082
0.2204	0.8006	14140	0.2189	0.911	0.9110	0.9111	0.911
0.2256	0.8407	14847	0.2176	0.9083	0.9083	0.9086	0.9083
0.222	0.8807	15554	0.2145	0.9116	0.9116	0.9116	0.9116
0.2198	0.9207	16261	0.2155	0.9113	0.9113	0.9116	0.9113
0.2223	0.9608	16968	0.2177	0.9075	0.9075	0.9079	0.9075
0.2223	1.0008	17675	0.2147	0.9112	0.9112	0.9112	0.9112
0.2064	1.0408	18382	0.2157	0.9105	0.9105	0.9105	0.9105
0.2053	1.0809	19089	0.2153	0.9102	0.9102	0.9102	0.9102
0.2071	1.1209	19796	0.2133	0.9113	0.9113	0.9113	0.9113
0.2035	1.1609	20503	0.2165	0.913	0.9130	0.9130	0.913
0.2033	1.2010	21210	0.2153	0.9119	0.9119	0.9119	0.9119
0.2071	1.2410	21917	0.2144	0.9124	0.9124	0.9124	0.9124
0.2025	1.2810	22624	0.2132	0.913	0.9130	0.9131	0.913
0.2056	1.3210	23331	0.2158	0.9111	0.9111	0.9113	0.9111
0.2058	1.3611	24038	0.2127	0.9117	0.9117	0.9117	0.9117
0.2026	1.4011	24745	0.2150	0.9124	0.9124	0.9124	0.9124
0.2053	1.4411	25452	0.2155	0.9123	0.9123	0.9125	0.9123
0.2006	1.4812	26159	0.2143	0.9135	0.9135	0.9136	0.9135
0.2054	1.5212	26866	0.2123	0.9142	0.9142	0.9142	0.9142
0.2017	1.5612	27573	0.2154	0.9123	0.9123	0.9127	0.9123
0.2027	1.6013	28280	0.2117	0.9137	0.9137	0.9137	0.9137
0.2029	1.6413	28987	0.2136	0.9132	0.9132	0.9133	0.9132
0.2025	1.6813	29694	0.2136	0.9123	0.9123	0.9124	0.9123
0.2037	1.7214	30401	0.2121	0.9125	0.9125	0.9125	0.9125
0.2015	1.7614	31108	0.2123	0.9131	0.9131	0.9131	0.9131
0.201	1.8014	31815	0.2127	0.9127	0.9127	0.9127	0.9127
0.2017	1.8415	32522	0.2109	0.913	0.9130	0.9130	0.913
0.2003	1.8815	33229	0.2114	0.9132	0.9132	0.9132	0.9132
0.2012	1.9215	33936	0.2123	0.9131	0.9131	0.9132	0.9131
0.199	1.9616	34643	0.2120	0.9126	0.9126	0.9127	0.9126

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0