exceptions_exp2_swap_0.7_cost_to_hit_5039

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.5627
• Accuracy: 0.3686

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: 0.0006
• train_batch_size: 16
• eval_batch_size: 16
• seed: 5039
• gradient_accumulation_steps: 5
• total_train_batch_size: 80
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.98) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 50.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
4.8254	0.2917	1000	4.7497	0.2555
4.3473	0.5834	2000	4.2906	0.2989
4.1662	0.8750	3000	4.1110	0.3140
4.0089	1.1665	4000	3.9986	0.3239
3.9468	1.4582	5000	3.9244	0.3306
3.8954	1.7499	6000	3.8653	0.3356
3.7599	2.0414	7000	3.8213	0.3404
3.7614	2.3331	8000	3.7913	0.3433
3.7333	2.6248	9000	3.7630	0.3461
3.7353	2.9165	10000	3.7343	0.3486
3.6353	3.2080	11000	3.7209	0.3502
3.6464	3.4996	12000	3.7050	0.3523
3.651	3.7913	13000	3.6839	0.3541
3.5514	4.0828	14000	3.6798	0.3547
3.5977	4.3745	15000	3.6690	0.3558
3.5826	4.6662	16000	3.6543	0.3573
3.5786	4.9579	17000	3.6397	0.3586
3.5123	5.2494	18000	3.6422	0.3593
3.5323	5.5411	19000	3.6319	0.3597
3.5372	5.8327	20000	3.6213	0.3607
3.4518	6.1243	21000	3.6249	0.3611
3.4819	6.4159	22000	3.6158	0.3619
3.4945	6.7076	23000	3.6058	0.3627
3.4983	6.9993	24000	3.5985	0.3632
3.4383	7.2908	25000	3.6055	0.3634
3.4662	7.5825	26000	3.5980	0.3639
3.471	7.8742	27000	3.5888	0.3650
3.39	8.1657	28000	3.5970	0.3646
3.4261	8.4574	29000	3.5891	0.3653
3.4344	8.7490	30000	3.5795	0.3659
3.3343	9.0405	31000	3.5856	0.3660
3.3842	9.3322	32000	3.5849	0.3661
3.4045	9.6239	33000	3.5780	0.3664
3.4279	9.9156	34000	3.5681	0.3676
3.3386	10.2071	35000	3.5805	0.3669
3.3786	10.4988	36000	3.5743	0.3676
3.3956	10.7905	37000	3.5654	0.3679
3.3017	11.0820	38000	3.5721	0.3680
3.3311	11.3736	39000	3.5750	0.3679
3.368	11.6653	40000	3.5627	0.3686
3.3692	11.9570	41000	3.5585	0.3690
3.3175	12.2485	42000	3.5673	0.3687
3.3491	12.5402	43000	3.5662	0.3690
3.3595	12.8319	44000	3.5551	0.3695
3.2751	13.1234	45000	3.5682	0.3691
3.3066	13.4151	46000	3.5649	0.3694
3.3459	13.7067	47000	3.5551	0.3698
3.3443	13.9984	48000	3.5495	0.3700
3.2903	14.2899	49000	3.5655	0.3693
3.3171	14.5816	50000	3.5562	0.3699
3.3282	14.8733	51000	3.5500	0.3704
3.2508	15.1648	52000	3.5646	0.3697
3.284	15.4565	53000	3.5588	0.3703
3.311	15.7482	54000	3.5497	0.3707
3.2153	16.0397	55000	3.5651	0.3703
3.2621	16.3313	56000	3.5589	0.3707
3.2777	16.6230	57000	3.5537	0.3711
3.3053	16.9147	58000	3.5446	0.3713
3.2309	17.2062	59000	3.5624	0.3703
3.2671	17.4979	60000	3.5530	0.3711
3.2881	17.7896	61000	3.5441	0.3717
3.1982	18.0811	62000	3.5611	0.3712
3.2554	18.3728	63000	3.5544	0.3714
3.2631	18.6644	64000	3.5450	0.3716
3.2762	18.9561	65000	3.5401	0.3721
3.2213	19.2476	66000	3.5625	0.3712
3.2506	19.5393	67000	3.5498	0.3718
3.2798	19.8310	68000	3.5441	0.3723
3.1919	20.1225	69000	3.5620	0.3714
3.2248	20.4142	70000	3.5535	0.3718
3.2525	20.7059	71000	3.5443	0.3723
3.2589	20.9975	72000	3.5392	0.3725
3.2056	21.2891	73000	3.5601	0.3717
3.2353	21.5807	74000	3.5484	0.3721
3.2478	21.8724	75000	3.5437	0.3726
3.1735	22.1639	76000	3.5617	0.3717
3.2156	22.4556	77000	3.5527	0.3724
3.2259	22.7473	78000	3.5409	0.3727
3.1469	23.0388	79000	3.5563	0.3724
3.1909	23.3305	80000	3.5521	0.3726
3.2152	23.6222	81000	3.5484	0.3726
3.2269	23.9138	82000	3.5400	0.3734
3.1555	24.2053	83000	3.5576	0.3722
3.2004	24.4970	84000	3.5537	0.3724
3.2151	24.7887	85000	3.5455	0.3732
3.1358	25.0802	86000	3.5608	0.3724
3.1878	25.3719	87000	3.5566	0.3726
3.1937	25.6636	88000	3.5462	0.3730
3.2214	25.9553	89000	3.5413	0.3735
3.1504	26.2468	90000	3.5591	0.3725
3.1829	26.5384	91000	3.5540	0.3729
3.2094	26.8301	92000	3.5458	0.3735

Framework versions

• Transformers 4.55.2
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4