train_copa_1745950332

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the copa dataset. It achieves the following results on the evaluation set:

• Loss: 0.5806
• Num Input Tokens Seen: 11206480

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: 2
• eval_batch_size: 2
• seed: 123
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• training_steps: 40000

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.5247	2.2222	200	0.6453	56064
0.3059	4.4444	400	0.6131	112064
0.4658	6.6667	600	0.6006	168096
0.3239	8.8889	800	0.5966	224048
0.699	11.1111	1000	0.5949	280048
0.4743	13.3333	1200	0.5968	336032
0.6378	15.5556	1400	0.5886	392032
0.5638	17.7778	1600	0.6024	448128
0.6903	20.0	1800	0.5930	503904
0.8633	22.2222	2000	0.5898	559936
0.2962	24.4444	2200	0.5938	615968
0.4546	26.6667	2400	0.5912	672064
0.5245	28.8889	2600	0.5971	728128
0.6674	31.1111	2800	0.5930	784032
0.5847	33.3333	3000	0.6022	839984
0.5704	35.5556	3200	0.5930	896288
0.6593	37.7778	3400	0.5853	952128
0.5875	40.0	3600	0.5806	1008096
0.4087	42.2222	3800	0.5932	1063984
0.5909	44.4444	4000	0.5908	1120080
0.6478	46.6667	4200	0.5956	1176240
0.5998	48.8889	4400	0.5960	1232160
0.4967	51.1111	4600	0.5897	1288160
0.7146	53.3333	4800	0.6051	1344160
0.7315	55.5556	5000	0.5910	1400368
0.3447	57.7778	5200	0.5956	1456368
0.6198	60.0	5400	0.5946	1512336
0.715	62.2222	5600	0.5861	1568192
0.769	64.4444	5800	0.5964	1624288
0.7341	66.6667	6000	0.5992	1680352
0.7653	68.8889	6200	0.5860	1736384
0.6777	71.1111	6400	0.5905	1792480
0.4726	73.3333	6600	0.5953	1848416
0.2734	75.5556	6800	0.5887	1904480
0.4698	77.7778	7000	0.5914	1960496
0.4363	80.0	7200	0.5871	2016368
0.417	82.2222	7400	0.5934	2072400
0.7079	84.4444	7600	0.6031	2128384
0.3862	86.6667	7800	0.5933	2184416
0.5057	88.8889	8000	0.6000	2240512
0.7429	91.1111	8200	0.5899	2296496
0.3976	93.3333	8400	0.5937	2352560
0.5796	95.5556	8600	0.5922	2408640
0.8014	97.7778	8800	0.5963	2464672
0.4672	100.0	9000	0.5936	2520688
0.4186	102.2222	9200	0.5940	2576656
0.8893	104.4444	9400	0.5933	2632720
0.4723	106.6667	9600	0.5885	2688704
0.5552	108.8889	9800	0.5985	2744768
0.5825	111.1111	10000	0.5858	2800768
0.664	113.3333	10200	0.5939	2856768
0.6141	115.5556	10400	0.5816	2912640
0.7501	117.7778	10600	0.5926	2968832
0.5082	120.0	10800	0.5948	3024896
0.4786	122.2222	11000	0.5913	3081056
0.6338	124.4444	11200	0.5895	3136944
0.5686	126.6667	11400	0.5883	3192960
0.492	128.8889	11600	0.5977	3248976
0.6217	131.1111	11800	0.6016	3305024
0.449	133.3333	12000	0.5961	3361008
0.5559	135.5556	12200	0.5869	3417152
0.5477	137.7778	12400	0.5960	3472832
0.5997	140.0	12600	0.5940	3529008
1.0409	142.2222	12800	0.5897	3585200
0.6995	144.4444	13000	0.5909	3641200
0.5804	146.6667	13200	0.5989	3697232
0.5644	148.8889	13400	0.5850	3753168
0.6163	151.1111	13600	0.5982	3809136
0.654	153.3333	13800	0.5920	3865216
0.6615	155.5556	14000	0.5916	3921216
0.6268	157.7778	14200	0.5823	3977312
0.5235	160.0	14400	0.5897	4033488
0.7357	162.2222	14600	0.5942	4089504
0.577	164.4444	14800	0.5987	4145504
0.5209	166.6667	15000	0.5963	4201440
0.5282	168.8889	15200	0.5873	4257504
0.7211	171.1111	15400	0.6066	4313408
0.4555	173.3333	15600	0.5993	4369488
0.3674	175.5556	15800	0.5935	4425536
0.6888	177.7778	16000	0.5898	4481568
0.3667	180.0	16200	0.6009	4537616
0.5047	182.2222	16400	0.5901	4593600
0.6513	184.4444	16600	0.5957	4649664
0.6596	186.6667	16800	0.5932	4705600
0.6953	188.8889	17000	0.5922	4761760
0.7941	191.1111	17200	0.5954	4817728
0.7163	193.3333	17400	0.5937	4873856
0.5062	195.5556	17600	0.5925	4929936
0.5253	197.7778	17800	0.5895	4985840
0.3207	200.0	18000	0.5997	5041920
0.4597	202.2222	18200	0.5909	5097872
0.5831	204.4444	18400	0.5981	5154064
0.5745	206.6667	18600	0.5881	5210112
0.4919	208.8889	18800	0.6006	5266064
0.5265	211.1111	19000	0.5922	5322160
0.4583	213.3333	19200	0.5896	5378224
0.5041	215.5556	19400	0.5905	5434432
0.5953	217.7778	19600	0.5943	5490352
0.4611	220.0	19800	0.5948	5546432
0.4757	222.2222	20000	0.5919	5602400
0.425	224.4444	20200	0.5954	5658464
0.6132	226.6667	20400	0.5904	5714352
0.4604	228.8889	20600	0.5916	5770416
0.6042	231.1111	20800	0.5881	5826496
0.7861	233.3333	21000	0.5867	5882496
0.45	235.5556	21200	0.5952	5938432
0.7427	237.7778	21400	0.5948	5994480
0.3559	240.0	21600	0.5922	6050656
0.5895	242.2222	21800	0.5895	6106736
0.4452	244.4444	22000	0.5900	6162896
0.6951	246.6667	22200	0.5837	6218976
0.5729	248.8889	22400	0.5936	6274960
0.6379	251.1111	22600	0.5899	6331008
0.6795	253.3333	22800	0.5971	6387152
0.553	255.5556	23000	0.5916	6443200
0.8381	257.7778	23200	0.5951	6499088
0.5589	260.0	23400	0.5881	6555184
0.4607	262.2222	23600	0.5931	6611312
0.5773	264.4444	23800	0.5904	6667104
0.7634	266.6667	24000	0.5937	6723024
0.5353	268.8889	24200	0.5957	6779376
0.6405	271.1111	24400	0.6005	6835232
0.4808	273.3333	24600	0.5897	6891104
0.6208	275.5556	24800	0.5926	6947456
0.4931	277.7778	25000	0.5843	7003408
0.4467	280.0	25200	0.5923	7059536
0.8506	282.2222	25400	0.5912	7115504
0.4577	284.4444	25600	0.5812	7171744
0.546	286.6667	25800	0.5934	7227712
0.8128	288.8889	26000	0.5878	7283856
0.547	291.1111	26200	0.5882	7339872
0.4865	293.3333	26400	0.5897	7395808
0.3535	295.5556	26600	0.5931	7451904
0.5505	297.7778	26800	0.5893	7507792
0.5664	300.0	27000	0.6017	7563888
0.6761	302.2222	27200	0.5857	7619872
0.3909	304.4444	27400	0.5936	7676016
0.4994	306.6667	27600	0.5878	7731872
0.5033	308.8889	27800	0.5835	7787920
0.5191	311.1111	28000	0.5952	7844080
0.7039	313.3333	28200	0.6011	7900064
0.4878	315.5556	28400	0.5907	7956016
0.6062	317.7778	28600	0.5900	8012160
0.4951	320.0	28800	0.5903	8068256
0.5753	322.2222	29000	0.5905	8124112
0.444	324.4444	29200	0.5929	8180192
0.6223	326.6667	29400	0.5876	8236304
0.6414	328.8889	29600	0.5926	8292272
0.4893	331.1111	29800	0.5910	8348416
0.3826	333.3333	30000	0.5867	8404432
0.5794	335.5556	30200	0.5903	8460384
0.7639	337.7778	30400	0.5897	8516432
0.6105	340.0	30600	0.5926	8572496
0.3847	342.2222	30800	0.5966	8628448
0.5284	344.4444	31000	0.5888	8684672
0.5963	346.6667	31200	0.5919	8740800
0.5242	348.8889	31400	0.5947	8796784
0.5337	351.1111	31600	0.5908	8852784
0.6405	353.3333	31800	0.5884	8909040
1.0247	355.5556	32000	0.5904	8965104
0.5278	357.7778	32200	0.5875	9021344
0.8911	360.0	32400	0.5860	9077456
0.888	362.2222	32600	0.5862	9133648
0.5925	364.4444	32800	0.5855	9189616
0.7401	366.6667	33000	0.5874	9245504
0.5229	368.8889	33200	0.5993	9301520
0.5437	371.1111	33400	0.5892	9357712
0.7065	373.3333	33600	0.5834	9413712
0.4855	375.5556	33800	0.5877	9469696
0.6214	377.7778	34000	0.5909	9525760
0.5044	380.0	34200	0.5862	9581648
0.4892	382.2222	34400	0.5952	9637632
0.3498	384.4444	34600	0.5953	9693568
0.5319	386.6667	34800	0.5976	9749792
0.5776	388.8889	35000	0.5917	9805840
0.8169	391.1111	35200	0.5974	9861856
0.8543	393.3333	35400	0.5967	9917904
0.8396	395.5556	35600	0.5940	9973888
0.5002	397.7778	35800	0.5911	10030096
0.5788	400.0	36000	0.5914	10086192
0.3754	402.2222	36200	0.5914	10142304
0.5558	404.4444	36400	0.5915	10198320
0.6992	406.6667	36600	0.5915	10254256
0.6034	408.8889	36800	0.5915	10310096
0.6889	411.1111	37000	0.5915	10366160
0.4812	413.3333	37200	0.5915	10422192
0.5055	415.5556	37400	0.5915	10478368
0.4931	417.7778	37600	0.5915	10534240
0.6183	420.0	37800	0.5915	10590208
0.392	422.2222	38000	0.5915	10646384
0.5446	424.4444	38200	0.5915	10702336
0.5018	426.6667	38400	0.5915	10758400
0.5608	428.8889	38600	0.5915	10814480
0.5076	431.1111	38800	0.5915	10870400
0.4828	433.3333	39000	0.5915	10926320
0.6352	435.5556	39200	0.5915	10982240
0.4988	437.7778	39400	0.5915	11038352
0.5288	440.0	39600	0.5915	11094352
0.4205	442.2222	39800	0.5915	11150400
0.8414	444.4444	40000	0.5915	11206480

Framework versions

• PEFT 0.15.2.dev0
• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1