lr5e-05_bs16_0509_1701

This model is a fine-tuned version of nvidia/mit-b0 on the greenkwd/upwellingdetection_SST dataset. It achieves the following results on the evaluation set:

• Loss: 0.2247
• Mean Iou: 0.4779
• Mean Accuracy: 0.8016
• Overall Accuracy: 0.8593
• Accuracy Land: nan
• Accuracy Upwelling: 0.9306
• Accuracy Not Upwelling: 0.6727
• Iou Land: 0.0
• Iou Upwelling: 0.8576
• Iou Not Upwelling: 0.5761
• Dice Macro: 0.8589
• Dice Micro: 0.9055

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: 5e-05
• train_batch_size: 32
• 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
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Land	Accuracy Upwelling	Accuracy Not Upwelling	Iou Land	Iou Upwelling	Iou Not Upwelling	Dice Macro	Dice Micro
1.1187	0.8	20	1.1005	0.1569	0.4351	0.3122	nan	0.1597	0.7104	0.0	0.1507	0.3199	0.4029	0.4345
0.9878	1.6	40	1.0030	0.2851	0.6584	0.5931	nan	0.5106	0.8061	0.0	0.4830	0.3723	0.6490	0.6790
0.8057	2.4	60	0.7942	0.3603	0.7501	0.7138	nan	0.6698	0.8303	0.0	0.6371	0.4437	0.7415	0.7798
0.6353	3.2	80	0.5933	0.4045	0.7321	0.7904	nan	0.8641	0.6002	0.0	0.7669	0.4465	0.7828	0.8433
0.5465	4.0	100	0.4706	0.4092	0.7118	0.8090	nan	0.9289	0.4946	0.0	0.8069	0.4207	0.7918	0.8672
0.4525	4.8	120	0.4175	0.4265	0.7415	0.8151	nan	0.9048	0.5782	0.0	0.8110	0.4685	0.8062	0.8696
0.4248	5.6	140	0.3652	0.4370	0.7546	0.8247	nan	0.9115	0.5978	0.0	0.8200	0.4908	0.8175	0.8769
0.3756	6.4	160	0.3417	0.4512	0.7735	0.8372	nan	0.9152	0.6317	0.0	0.8320	0.5216	0.8305	0.8859
0.375	7.2	180	0.3401	0.4514	0.7772	0.8313	nan	0.8993	0.6551	0.0	0.8273	0.5268	0.8318	0.8834
0.327	8.0	200	0.3223	0.4467	0.7630	0.8310	nan	0.9158	0.6102	0.0	0.8318	0.5084	0.8303	0.8869
0.3417	8.8	220	0.3140	0.4526	0.7669	0.8369	nan	0.9241	0.6098	0.0	0.8406	0.5173	0.8376	0.8925
0.315	9.6	240	0.3118	0.4501	0.7662	0.8352	nan	0.9192	0.6131	0.0	0.8381	0.5121	0.8341	0.8906
0.3303	10.4	260	0.3035	0.4664	0.7992	0.8509	nan	0.9136	0.6848	0.0	0.8391	0.5602	0.8408	0.8902
0.3201	11.2	280	0.3016	0.4492	0.7595	0.8368	nan	0.9333	0.5858	0.0	0.8402	0.5074	0.8339	0.8918
0.3107	12.0	300	0.3018	0.4397	0.7439	0.8170	nan	0.9071	0.5807	0.0	0.8376	0.4814	0.8299	0.8892
0.3042	12.8	320	0.2965	0.4601	0.8012	0.8349	nan	0.8777	0.7247	0.0	0.8227	0.5576	0.8437	0.8881
0.3126	13.6	340	0.2858	0.4494	0.7647	0.8254	nan	0.9001	0.6293	0.0	0.8384	0.5099	0.8393	0.8936
0.3021	14.4	360	0.2797	0.4566	0.7728	0.8369	nan	0.9160	0.6296	0.0	0.8445	0.5253	0.8430	0.8960
0.2897	15.2	380	0.2830	0.4481	0.7575	0.8300	nan	0.9215	0.5935	0.0	0.8393	0.5049	0.8359	0.8924
0.3312	16.0	400	0.2742	0.4590	0.7791	0.8420	nan	0.9194	0.6387	0.0	0.8426	0.5343	0.8433	0.8960
0.2575	16.8	420	0.2835	0.4472	0.7644	0.8223	nan	0.8947	0.6340	0.0	0.8344	0.5072	0.8370	0.8895
0.2851	17.6	440	0.2684	0.4634	0.7884	0.8421	nan	0.9078	0.6689	0.0	0.8441	0.5460	0.8479	0.8968
0.2996	18.4	460	0.2802	0.4467	0.7631	0.8199	nan	0.8893	0.6369	0.0	0.8343	0.5057	0.8366	0.8901
0.3527	19.2	480	0.2615	0.4557	0.7652	0.8475	nan	0.9472	0.5833	0.0	0.8484	0.5186	0.8401	0.8985
0.2549	20.0	500	0.2846	0.4686	0.8241	0.8413	nan	0.8622	0.7861	0.0	0.8235	0.5823	0.8507	0.8915
0.2657	20.8	520	0.2594	0.4513	0.7557	0.8460	nan	0.9560	0.5553	0.0	0.8493	0.5044	0.8368	0.8983
0.2576	21.6	540	0.2623	0.4761	0.8106	0.8587	nan	0.9189	0.7023	0.0	0.8446	0.5838	0.8511	0.8963
0.2866	22.4	560	0.2647	0.4505	0.7624	0.8289	nan	0.9091	0.6157	0.0	0.8444	0.5073	0.8398	0.8950
0.2638	23.2	580	0.2590	0.4619	0.7791	0.8464	nan	0.9306	0.6276	0.0	0.8461	0.5395	0.8442	0.8966
0.2602	24.0	600	0.2584	0.4750	0.8019	0.8601	nan	0.9314	0.6724	0.0	0.8506	0.5744	0.8490	0.8975
0.2775	24.8	620	0.2589	0.4663	0.8020	0.8399	nan	0.8863	0.7177	0.0	0.8381	0.5607	0.8523	0.8972
0.256	25.6	640	0.2511	0.4592	0.7730	0.8385	nan	0.9203	0.6256	0.0	0.8489	0.5287	0.8477	0.8994
0.2645	26.4	660	0.2587	0.4572	0.7677	0.8542	nan	0.9599	0.5756	0.0	0.8474	0.5241	0.8358	0.8951
0.2838	27.2	680	0.2488	0.4599	0.7738	0.8467	nan	0.9356	0.6120	0.0	0.8498	0.5300	0.8440	0.8991
0.2507	28.0	700	0.2493	0.4673	0.7938	0.8439	nan	0.9071	0.6806	0.0	0.8452	0.5566	0.8527	0.8993
0.2496	28.8	720	0.2604	0.4832	0.8234	0.8662	nan	0.9178	0.7289	0.0	0.8484	0.6014	0.8533	0.8969
0.2578	29.6	740	0.2395	0.4724	0.7942	0.8551	nan	0.9286	0.6599	0.0	0.8557	0.5615	0.8550	0.9041
0.2593	30.4	760	0.2436	0.4675	0.7819	0.8574	nan	0.9509	0.6129	0.0	0.8538	0.5486	0.8496	0.9029
0.3011	31.2	780	0.2583	0.4765	0.8021	0.8647	nan	0.9428	0.6614	0.0	0.8495	0.5800	0.8461	0.8953
0.2386	32.0	800	0.2348	0.4747	0.7964	0.8575	nan	0.9325	0.6602	0.0	0.8570	0.5670	0.8562	0.9046
0.2213	32.8	820	0.2473	0.4714	0.7955	0.8530	nan	0.9243	0.6668	0.0	0.8486	0.5657	0.8510	0.8993
0.2443	33.6	840	0.2470	0.4675	0.7873	0.8553	nan	0.9385	0.6362	0.0	0.8499	0.5527	0.8471	0.8991
0.2499	34.4	860	0.2407	0.4792	0.8101	0.8568	nan	0.9155	0.7047	0.0	0.8526	0.5850	0.8591	0.9031
0.2641	35.2	880	0.2390	0.4800	0.8067	0.8639	nan	0.9345	0.6789	0.0	0.8563	0.5837	0.8555	0.9022
0.2627	36.0	900	0.2356	0.4801	0.8146	0.8577	nan	0.9107	0.7185	0.0	0.8521	0.5881	0.8605	0.9037
0.269	36.8	920	0.2437	0.4714	0.7950	0.8561	nan	0.9313	0.6587	0.0	0.8504	0.5639	0.8498	0.8993
0.3014	37.6	940	0.2415	0.4648	0.7926	0.8422	nan	0.9029	0.6823	0.0	0.8433	0.5512	0.8511	0.8993
0.2557	38.4	960	0.2463	0.4671	0.7922	0.8473	nan	0.9154	0.6691	0.0	0.8462	0.5552	0.8493	0.8974
0.2927	39.2	980	0.2401	0.4736	0.8034	0.8524	nan	0.9139	0.6930	0.0	0.8470	0.5738	0.8553	0.9009
0.2371	40.0	1000	0.2486	0.4547	0.7724	0.8324	nan	0.9048	0.6400	0.0	0.8427	0.5214	0.8433	0.8959
0.2279	40.8	1020	0.2371	0.4733	0.7908	0.8587	nan	0.9447	0.6369	0.0	0.8559	0.5640	0.8537	0.9036
0.2328	41.6	1040	0.2365	0.4740	0.7968	0.8551	nan	0.9282	0.6653	0.0	0.8543	0.5677	0.8552	0.9023
0.2371	42.4	1060	0.2383	0.4732	0.7916	0.8653	nan	0.9518	0.6314	0.0	0.8575	0.5620	0.8486	0.9018
0.2423	43.2	1080	0.2411	0.4636	0.7765	0.8505	nan	0.9426	0.6104	0.0	0.8519	0.5388	0.8456	0.8999
0.2685	44.0	1100	0.2312	0.4755	0.7989	0.8539	nan	0.9221	0.6757	0.0	0.8572	0.5692	0.8584	0.9046
0.2897	44.8	1120	0.2547	0.4628	0.7860	0.8537	nan	0.9375	0.6344	0.0	0.8400	0.5485	0.8404	0.8931
0.24	45.6	1140	0.2457	0.4531	0.7601	0.8346	nan	0.9267	0.5936	0.0	0.8495	0.5096	0.8420	0.8978
0.2539	46.4	1160	0.2325	0.4737	0.7896	0.8618	nan	0.9518	0.6273	0.0	0.8583	0.5627	0.8527	0.9040
0.2477	47.2	1180	0.2308	0.4712	0.7954	0.8474	nan	0.9117	0.6792	0.0	0.8521	0.5616	0.8571	0.9041
0.2915	48.0	1200	0.2351	0.4747	0.7985	0.8593	nan	0.9341	0.6630	0.0	0.8539	0.5702	0.8543	0.9027
0.2313	48.8	1220	0.2352	0.4756	0.8097	0.8555	nan	0.9099	0.7095	0.0	0.8495	0.5773	0.8570	0.9025
0.2479	49.6	1240	0.2328	0.4770	0.8081	0.8547	nan	0.9123	0.7039	0.0	0.8520	0.5789	0.8591	0.9032
0.2587	50.4	1260	0.2250	0.4801	0.8071	0.8605	nan	0.9247	0.6894	0.0	0.8597	0.5805	0.8607	0.9069
0.259	51.2	1280	0.2409	0.4674	0.7848	0.8571	nan	0.9457	0.6239	0.0	0.8509	0.5513	0.8458	0.8985
0.248	52.0	1300	0.2370	0.4738	0.7958	0.8596	nan	0.9384	0.6531	0.0	0.8541	0.5672	0.8524	0.9014
0.2612	52.8	1320	0.2362	0.4693	0.7912	0.8521	nan	0.9264	0.6560	0.0	0.8516	0.5563	0.8528	0.9019
0.2421	53.6	1340	0.2412	0.4714	0.7866	0.8627	nan	0.9584	0.6148	0.0	0.8555	0.5588	0.8483	0.9006
0.2371	54.4	1360	0.2444	0.4721	0.7939	0.8586	nan	0.9391	0.6488	0.0	0.8515	0.5647	0.8486	0.8983
0.2412	55.2	1380	0.2286	0.4753	0.7944	0.8601	nan	0.9404	0.6484	0.0	0.8594	0.5663	0.8568	0.9059
0.235	56.0	1400	0.2268	0.4764	0.7970	0.8617	nan	0.9411	0.6529	0.0	0.8580	0.5713	0.8560	0.9052
0.2265	56.8	1420	0.2310	0.4824	0.8132	0.8651	nan	0.9283	0.6981	0.0	0.8550	0.5921	0.8576	0.9029
0.2233	57.6	1440	0.2382	0.4784	0.8090	0.8604	nan	0.9237	0.6943	0.0	0.8514	0.5839	0.8539	0.8999
0.2657	58.4	1460	0.2290	0.4716	0.7926	0.8547	nan	0.9307	0.6545	0.0	0.8556	0.5590	0.8549	0.9044
0.2576	59.2	1480	0.2355	0.4749	0.7987	0.8588	nan	0.9334	0.6639	0.0	0.8535	0.5711	0.8531	0.9011
0.1939	60.0	1500	0.2213	0.4825	0.8091	0.8612	nan	0.9255	0.6927	0.0	0.8599	0.5876	0.8632	0.9074
0.2252	60.8	1520	0.2244	0.4751	0.7978	0.8571	nan	0.9292	0.6665	0.0	0.8584	0.5670	0.8577	0.9056
0.2052	61.6	1540	0.2268	0.4775	0.8007	0.8607	nan	0.9348	0.6667	0.0	0.8570	0.5754	0.8571	0.9045
0.2403	62.4	1560	0.2270	0.4750	0.7935	0.8618	nan	0.9449	0.6422	0.0	0.8588	0.5661	0.8554	0.9054
0.2355	63.2	1580	0.2410	0.4658	0.7834	0.8570	nan	0.9449	0.6218	0.0	0.8509	0.5466	0.8442	0.8984
0.2487	64.0	1600	0.2336	0.4755	0.7991	0.8636	nan	0.9419	0.6564	0.0	0.8546	0.5720	0.8515	0.9011
0.2557	64.8	1620	0.2303	0.4782	0.8103	0.8564	nan	0.9137	0.7069	0.0	0.8517	0.5828	0.8585	0.9025
0.2468	65.6	1640	0.2333	0.4685	0.7892	0.8464	nan	0.9171	0.6613	0.0	0.8522	0.5532	0.8532	0.9018
0.2203	66.4	1660	0.2348	0.4643	0.7805	0.8478	nan	0.9297	0.6313	0.0	0.8515	0.5415	0.8489	0.9009
0.2529	67.2	1680	0.2332	0.4724	0.7948	0.8574	nan	0.9349	0.6546	0.0	0.8520	0.5651	0.8517	0.9014
0.219	68.0	1700	0.2318	0.4757	0.8077	0.8510	nan	0.9048	0.7107	0.0	0.8505	0.5767	0.8589	0.9022
0.2225	68.8	1720	0.2376	0.4661	0.7898	0.8427	nan	0.9092	0.6704	0.0	0.8456	0.5526	0.8516	0.8993
0.2713	69.6	1740	0.2239	0.4777	0.8036	0.8581	nan	0.9252	0.6821	0.0	0.8566	0.5767	0.8601	0.9062
0.2666	70.4	1760	0.2375	0.4736	0.8021	0.8563	nan	0.9232	0.6811	0.0	0.8488	0.5720	0.8530	0.8999
0.2714	71.2	1780	0.2268	0.4685	0.7870	0.8521	nan	0.9303	0.6437	0.0	0.8556	0.5500	0.8525	0.9035
0.2815	72.0	1800	0.2224	0.4791	0.8051	0.8585	nan	0.9237	0.6865	0.0	0.8583	0.5789	0.8619	0.9071
0.2372	72.8	1820	0.2198	0.4831	0.8079	0.8643	nan	0.9349	0.6809	0.0	0.8611	0.5882	0.8625	0.9078
0.2311	73.6	1840	0.2298	0.4764	0.8007	0.8621	nan	0.9381	0.6633	0.0	0.8543	0.5749	0.8541	0.9021
0.2248	74.4	1860	0.2323	0.4695	0.7907	0.8509	nan	0.9262	0.6552	0.0	0.8511	0.5573	0.8524	0.9015
0.2777	75.2	1880	0.2294	0.4802	0.8098	0.8602	nan	0.9219	0.6977	0.0	0.8547	0.5858	0.8592	0.9042
0.2348	76.0	1900	0.2215	0.4855	0.8109	0.8676	nan	0.9386	0.6831	0.0	0.8623	0.5943	0.8629	0.9074
0.2268	76.8	1920	0.2273	0.4766	0.7990	0.8608	nan	0.9363	0.6617	0.0	0.8568	0.5729	0.8564	0.9047
0.2597	77.6	1940	0.2252	0.4794	0.8061	0.8603	nan	0.9272	0.6849	0.0	0.8576	0.5807	0.8602	0.9058
0.2644	78.4	1960	0.2301	0.4693	0.7892	0.8499	nan	0.9250	0.6535	0.0	0.8532	0.5548	0.8528	0.9023
0.2315	79.2	1980	0.2223	0.4849	0.8125	0.8676	nan	0.9358	0.6892	0.0	0.8607	0.5938	0.8612	0.9065
0.2225	80.0	2000	0.2264	0.4782	0.8027	0.8585	nan	0.9276	0.6779	0.0	0.8582	0.5763	0.8588	0.9050
0.2068	80.8	2020	0.2193	0.4829	0.8072	0.8627	nan	0.9320	0.6824	0.0	0.8622	0.5864	0.8642	0.9091
0.2449	81.6	2040	0.2242	0.4811	0.8083	0.8607	nan	0.9266	0.6900	0.0	0.8583	0.5851	0.8617	0.9060
0.2462	82.4	2060	0.2319	0.4745	0.7985	0.8611	nan	0.9365	0.6605	0.0	0.8537	0.5697	0.8522	0.9014
0.228	83.2	2080	0.2276	0.4756	0.7974	0.8593	nan	0.9363	0.6584	0.0	0.8556	0.5711	0.8545	0.9034
0.2686	84.0	2100	0.2202	0.4774	0.8017	0.8597	nan	0.9299	0.6734	0.0	0.8584	0.5738	0.8589	0.9065
0.2389	84.8	2120	0.2292	0.4674	0.7870	0.8471	nan	0.9198	0.6543	0.0	0.8533	0.5488	0.8530	0.9026
0.2377	85.6	2140	0.2194	0.4804	0.8022	0.8608	nan	0.9347	0.6698	0.0	0.8612	0.5799	0.8621	0.9078
0.226	86.4	2160	0.2254	0.4768	0.8013	0.8579	nan	0.9272	0.6753	0.0	0.8565	0.5739	0.8578	0.9047
0.2504	87.2	2180	0.2328	0.4671	0.7855	0.8496	nan	0.9279	0.6430	0.0	0.8524	0.5489	0.8511	0.9017
0.2687	88.0	2200	0.2265	0.4783	0.8054	0.8591	nan	0.9251	0.6857	0.0	0.8568	0.5780	0.8585	0.9044
0.2483	88.8	2220	0.2253	0.4773	0.7991	0.8598	nan	0.9358	0.6625	0.0	0.8570	0.5749	0.8583	0.9055
0.2322	89.6	2240	0.2232	0.4764	0.8001	0.8583	nan	0.9280	0.6722	0.0	0.8593	0.5699	0.8579	0.9054
0.2253	90.4	2260	0.2225	0.4802	0.8073	0.8618	nan	0.9282	0.6863	0.0	0.8577	0.5829	0.8601	0.9058
0.2344	91.2	2280	0.2240	0.4777	0.8025	0.8594	nan	0.9286	0.6764	0.0	0.8573	0.5758	0.8583	0.9052
0.2241	92.0	2300	0.2175	0.4823	0.8070	0.8649	nan	0.9356	0.6784	0.0	0.8620	0.5848	0.8624	0.9083
0.2408	92.8	2320	0.2278	0.4729	0.7934	0.8584	nan	0.9382	0.6485	0.0	0.8559	0.5628	0.8535	0.9033
0.2196	93.6	2340	0.2193	0.4815	0.8076	0.8622	nan	0.9294	0.6858	0.0	0.8596	0.5850	0.8621	0.9072
0.2362	94.4	2360	0.2169	0.4802	0.8021	0.8641	nan	0.9414	0.6628	0.0	0.8608	0.5798	0.8604	0.9079
0.2374	95.2	2380	0.2235	0.4780	0.8049	0.8598	nan	0.9272	0.6826	0.0	0.8560	0.5780	0.8593	0.9055
0.2591	96.0	2400	0.2369	0.4689	0.7909	0.8494	nan	0.9218	0.6600	0.0	0.8506	0.5561	0.8512	0.9000
0.2254	96.8	2420	0.2148	0.4846	0.8101	0.8672	nan	0.9365	0.6837	0.0	0.8635	0.5902	0.8643	0.9096
0.2433	97.6	2440	0.2322	0.4689	0.7879	0.8517	nan	0.9294	0.6464	0.0	0.8553	0.5515	0.8508	0.9009
0.2353	98.4	2460	0.2309	0.4721	0.7939	0.8557	nan	0.9312	0.6566	0.0	0.8531	0.5631	0.8528	0.9020
0.2301	99.2	2480	0.2252	0.4774	0.8014	0.8584	nan	0.9287	0.6740	0.0	0.8571	0.5751	0.8574	0.9044
0.2322	100.0	2500	0.2247	0.4779	0.8016	0.8593	nan	0.9306	0.6727	0.0	0.8576	0.5761	0.8589	0.9055

Framework versions

• Transformers 4.41.2
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.19.1