glpn-nyu-finetuned-diode-230131-041708

This model is a fine-tuned version of vinvino02/glpn-nyu on the diode-subset dataset. It achieves the following results on the evaluation set:

Loss: 0.4425
Mae: 0.4270
Rmse: 0.6196
Abs Rel: 0.4543
Log Mae: 0.1732
Log Rmse: 0.2288
Delta1: 0.3787
Delta2: 0.6298
Delta3: 0.8083

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: 0.0003
train_batch_size: 24
eval_batch_size: 48
seed: 2022
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Mae	Rmse	Abs Rel	Log Mae	Log Rmse	Delta1	Delta2	Delta3
0.5276	1.0	72	0.4701	0.4590	0.6348	0.4983	0.1903	0.2393	0.3169	0.5544	0.7661
0.4595	2.0	144	0.4867	0.4690	0.6369	0.5588	0.1956	0.2483	0.3090	0.5269	0.7532
0.4802	3.0	216	0.4854	0.4648	0.6344	0.5581	0.1935	0.2475	0.3135	0.5355	0.7531
0.4566	4.0	288	0.4709	0.4559	0.6756	0.4223	0.1890	0.2516	0.3668	0.6329	0.7696
0.4916	5.0	360	0.4835	0.4555	0.6343	0.5302	0.1881	0.2447	0.3435	0.5716	0.7437
0.4822	6.0	432	0.4756	0.4585	0.6301	0.5264	0.1894	0.2414	0.3238	0.5628	0.7435
0.4588	7.0	504	0.4655	0.4481	0.6509	0.4425	0.1843	0.2413	0.3498	0.6157	0.7809
0.4214	8.0	576	0.4869	0.4706	0.6391	0.5669	0.1961	0.2500	0.3033	0.5388	0.7371
0.426	9.0	648	0.4835	0.4679	0.6472	0.5117	0.1951	0.2486	0.3216	0.5474	0.7399
0.4135	10.0	720	0.4621	0.4439	0.6287	0.4803	0.1825	0.2365	0.3451	0.5887	0.7878
0.3778	11.0	792	0.4756	0.4566	0.6337	0.5174	0.1892	0.2431	0.3297	0.5560	0.7690
0.426	12.0	864	0.4542	0.4362	0.6219	0.4621	0.1779	0.2303	0.3572	0.6083	0.7835
0.4282	13.0	936	0.4514	0.4306	0.6195	0.4678	0.1754	0.2307	0.3661	0.6228	0.8083
0.4045	14.0	1008	0.4575	0.4390	0.6315	0.4530	0.1794	0.2343	0.3641	0.6128	0.7787
0.4351	15.0	1080	0.4669	0.4373	0.6423	0.4322	0.1796	0.2423	0.3917	0.6233	0.7850
0.4001	16.0	1152	0.4540	0.4356	0.6331	0.4336	0.1767	0.2320	0.3919	0.6132	0.7732
0.3741	17.0	1224	0.4890	0.4645	0.6361	0.5707	0.1926	0.2494	0.3253	0.5469	0.7386
0.4128	18.0	1296	0.4815	0.4593	0.6328	0.5511	0.1899	0.2457	0.3302	0.5571	0.7471
0.3809	19.0	1368	0.5002	0.4768	0.6425	0.6061	0.1991	0.2560	0.3105	0.5222	0.7118
0.4089	20.0	1440	0.4503	0.4311	0.6449	0.4081	0.1752	0.2370	0.4147	0.6445	0.7823
0.3612	21.0	1512	0.4541	0.4280	0.6215	0.4543	0.1735	0.2302	0.3823	0.6291	0.7968
0.3664	22.0	1584	0.4425	0.4251	0.6347	0.3970	0.1717	0.2300	0.4181	0.6374	0.7860
0.3787	23.0	1656	0.4722	0.4477	0.6378	0.4868	0.1846	0.2432	0.3541	0.6041	0.7733
0.4184	24.0	1728	0.4749	0.4506	0.6303	0.5329	0.1857	0.2434	0.3465	0.5752	0.7698
0.3928	25.0	1800	0.4646	0.4485	0.6395	0.4744	0.1847	0.2407	0.3528	0.5946	0.7816
0.3704	26.0	1872	0.4492	0.4340	0.6331	0.4344	0.1765	0.2326	0.3778	0.6314	0.7916
0.3462	27.0	1944	0.4467	0.4307	0.6314	0.4296	0.1751	0.2317	0.3840	0.6359	0.7983
0.3808	28.0	2016	0.4758	0.4622	0.6331	0.5236	0.1913	0.2425	0.3230	0.5439	0.7438
0.3641	29.0	2088	0.4609	0.4452	0.6315	0.4545	0.1824	0.2339	0.3484	0.5934	0.7716
0.3602	30.0	2160	0.4546	0.4413	0.6230	0.4729	0.1804	0.2318	0.3515	0.5944	0.7778
0.3638	31.0	2232	0.4498	0.4340	0.6245	0.4449	0.1764	0.2296	0.3725	0.6079	0.7923
0.3699	32.0	2304	0.4472	0.4305	0.6228	0.4568	0.1750	0.2307	0.3757	0.6239	0.8000
0.3805	33.0	2376	0.4647	0.4439	0.6325	0.4875	0.1823	0.2392	0.3609	0.5921	0.7833
0.3454	34.0	2448	0.4640	0.4442	0.6276	0.5008	0.1820	0.2376	0.3573	0.5865	0.7866
0.3452	35.0	2520	0.4646	0.4454	0.6276	0.4966	0.1827	0.2374	0.3489	0.5913	0.7726
0.3509	36.0	2592	0.4522	0.4394	0.6259	0.4605	0.1799	0.2321	0.3534	0.6001	0.7944
0.3432	37.0	2664	0.4656	0.4484	0.6290	0.5067	0.1841	0.2390	0.3487	0.5802	0.7687
0.381	38.0	2736	0.4630	0.4405	0.6287	0.4970	0.1807	0.2387	0.3565	0.6067	0.7907
0.3591	39.0	2808	0.4637	0.4452	0.6269	0.4995	0.1825	0.2374	0.3487	0.5966	0.7654
0.3826	40.0	2880	0.4723	0.4527	0.6307	0.5279	0.1867	0.2421	0.3338	0.5745	0.7713
0.3585	41.0	2952	0.4485	0.4306	0.6238	0.4470	0.1749	0.2297	0.3736	0.6251	0.7995
0.3518	42.0	3024	0.4369	0.4229	0.6293	0.4111	0.1701	0.2277	0.4004	0.6563	0.8009
0.359	43.0	3096	0.4545	0.4348	0.6274	0.4607	0.1777	0.2338	0.3592	0.6237	0.8000
0.3274	44.0	3168	0.4595	0.4359	0.6278	0.4781	0.1779	0.2357	0.3729	0.6093	0.7980
0.3368	45.0	3240	0.4617	0.4434	0.6253	0.5001	0.1819	0.2368	0.3400	0.5966	0.7953
0.3638	46.0	3312	0.4634	0.4380	0.6264	0.4925	0.1794	0.2371	0.3576	0.6158	0.7907
0.3698	47.0	3384	0.4559	0.4343	0.6223	0.4890	0.1776	0.2346	0.3579	0.6103	0.8110
0.3392	48.0	3456	0.4646	0.4477	0.6267	0.5029	0.1837	0.2374	0.3451	0.5798	0.7665
0.3548	49.0	3528	0.4598	0.4394	0.6245	0.4885	0.1793	0.2351	0.3647	0.6016	0.7815
0.3375	50.0	3600	0.4441	0.4271	0.6226	0.4487	0.1729	0.2293	0.3808	0.6354	0.8075
0.3315	51.0	3672	0.4613	0.4403	0.6292	0.4868	0.1805	0.2373	0.3630	0.6016	0.7905
0.3313	52.0	3744	0.4445	0.4307	0.6442	0.4108	0.1746	0.2342	0.3942	0.6577	0.7932
0.3372	53.0	3816	0.4456	0.4258	0.6269	0.4404	0.1720	0.2308	0.3924	0.6489	0.8027
0.3285	54.0	3888	0.4526	0.4348	0.6241	0.4723	0.1772	0.2328	0.3615	0.6160	0.8027
0.3474	55.0	3960	0.4498	0.4369	0.6258	0.4595	0.1782	0.2315	0.3617	0.6070	0.7978
0.3349	56.0	4032	0.4613	0.4428	0.6307	0.4858	0.1819	0.2376	0.3523	0.6012	0.7875
0.3207	57.0	4104	0.4476	0.4342	0.6230	0.4500	0.1765	0.2289	0.3658	0.6151	0.7910
0.3399	58.0	4176	0.4600	0.4413	0.6248	0.4940	0.1812	0.2360	0.3531	0.5954	0.7814
0.3327	59.0	4248	0.4463	0.4339	0.6215	0.4570	0.1770	0.2294	0.3590	0.6069	0.8063
0.3215	60.0	4320	0.4482	0.4317	0.6203	0.4595	0.1756	0.2295	0.3698	0.6154	0.8034
0.3276	61.0	4392	0.4406	0.4218	0.6192	0.4370	0.1705	0.2268	0.3878	0.6425	0.8111
0.3179	62.0	4464	0.4530	0.4331	0.6217	0.4765	0.1764	0.2327	0.3660	0.6121	0.8068
0.3129	63.0	4536	0.4614	0.4398	0.6263	0.5002	0.1803	0.2378	0.3529	0.6023	0.7974
0.3354	64.0	4608	0.4538	0.4374	0.6234	0.4777	0.1788	0.2333	0.3565	0.6013	0.7995
0.3261	65.0	4680	0.4367	0.4258	0.6283	0.4249	0.1724	0.2291	0.3861	0.6484	0.8026
0.3114	66.0	4752	0.4565	0.4366	0.6225	0.4852	0.1780	0.2334	0.3647	0.6073	0.7939
0.3377	67.0	4824	0.4519	0.4308	0.6185	0.4771	0.1755	0.2314	0.3681	0.6175	0.8079
0.3266	68.0	4896	0.4372	0.4216	0.6167	0.4345	0.1702	0.2245	0.3850	0.6392	0.8163
0.3347	69.0	4968	0.4343	0.4193	0.6179	0.4318	0.1690	0.2252	0.3890	0.6557	0.8114
0.3207	70.0	5040	0.4426	0.4269	0.6180	0.4465	0.1728	0.2266	0.3810	0.6296	0.8038
0.3313	71.0	5112	0.4362	0.4234	0.6177	0.4360	0.1712	0.2252	0.3777	0.6386	0.8133
0.326	72.0	5184	0.4392	0.4251	0.6182	0.4431	0.1723	0.2265	0.3783	0.6356	0.8088
0.3141	73.0	5256	0.4532	0.4385	0.6214	0.4818	0.1796	0.2327	0.3513	0.5981	0.8044
0.3301	74.0	5328	0.4536	0.4361	0.6230	0.4808	0.1783	0.2333	0.3585	0.6097	0.8037
0.3194	75.0	5400	0.4501	0.4335	0.6216	0.4698	0.1765	0.2312	0.3623	0.6164	0.8033
0.3071	76.0	5472	0.4455	0.4310	0.6201	0.4598	0.1751	0.2292	0.3625	0.6231	0.8087
0.3174	77.0	5544	0.4472	0.4316	0.6219	0.4625	0.1756	0.2307	0.3654	0.6256	0.8022
0.3171	78.0	5616	0.4461	0.4305	0.6204	0.4614	0.1750	0.2298	0.3663	0.6263	0.8052
0.3244	79.0	5688	0.4501	0.4328	0.6226	0.4725	0.1765	0.2324	0.3611	0.6233	0.8083
0.3188	80.0	5760	0.4427	0.4280	0.6199	0.4507	0.1735	0.2281	0.3757	0.6307	0.8054
0.3212	81.0	5832	0.4383	0.4222	0.6196	0.4365	0.1702	0.2266	0.3875	0.6476	0.8093
0.3234	82.0	5904	0.4434	0.4278	0.6216	0.4479	0.1735	0.2288	0.3728	0.6337	0.8064
0.3024	83.0	5976	0.4502	0.4331	0.6214	0.4728	0.1764	0.2317	0.3645	0.6192	0.8070
0.3145	84.0	6048	0.4409	0.4258	0.6199	0.4475	0.1726	0.2280	0.3778	0.6357	0.8075
0.329	85.0	6120	0.4491	0.4302	0.6221	0.4710	0.1749	0.2322	0.3755	0.6246	0.8065
0.3034	86.0	6192	0.4504	0.4321	0.6241	0.4699	0.1757	0.2325	0.3767	0.6217	0.8035
0.3074	87.0	6264	0.4373	0.4224	0.6188	0.4396	0.1706	0.2267	0.3878	0.6439	0.8107
0.3089	88.0	6336	0.4379	0.4235	0.6191	0.4402	0.1709	0.2266	0.3893	0.6410	0.8089
0.2995	89.0	6408	0.4448	0.4292	0.6193	0.4597	0.1744	0.2292	0.3740	0.6225	0.8065
0.3248	90.0	6480	0.4413	0.4279	0.6227	0.4494	0.1732	0.2283	0.3766	0.6305	0.8100
0.3203	91.0	6552	0.4445	0.4290	0.6213	0.4568	0.1740	0.2295	0.3754	0.6290	0.8085
0.3109	92.0	6624	0.4452	0.4295	0.6203	0.4597	0.1744	0.2297	0.3749	0.6245	0.8035
0.3241	93.0	6696	0.4419	0.4258	0.6190	0.4533	0.1725	0.2285	0.3833	0.6313	0.8092
0.3078	94.0	6768	0.4446	0.4278	0.6201	0.4597	0.1736	0.2297	0.3778	0.6284	0.8097
0.3141	95.0	6840	0.4466	0.4305	0.6208	0.4660	0.1749	0.2306	0.3720	0.6233	0.8058
0.3198	96.0	6912	0.4440	0.4275	0.6194	0.4584	0.1736	0.2293	0.3774	0.6279	0.8088
0.3	97.0	6984	0.4426	0.4269	0.6192	0.4545	0.1731	0.2287	0.3770	0.6302	0.8091
0.3096	98.0	7056	0.4433	0.4274	0.6197	0.4568	0.1735	0.2292	0.3765	0.6296	0.8088
0.3317	99.0	7128	0.4394	0.4244	0.6196	0.4448	0.1716	0.2276	0.3844	0.6374	0.8096
0.3132	100.0	7200	0.4425	0.4270	0.6196	0.4543	0.1732	0.2288	0.3787	0.6298	0.8083

Framework versions

Transformers 4.24.0
Pytorch 1.12.1+cu116
Datasets 2.8.0
Tokenizers 0.13.2

作者:

Sayak Paul

数据集大小:

233.87 MB