lemexp-task1-v2-lemma_object_full_nodefs-deepseek-coder-1.3b-base-ddp-8lr-v2

This model is a fine-tuned version of deepseek-ai/deepseek-coder-1.3b-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2426

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: 0.0008
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 16
• total_eval_batch_size: 16
• 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
• num_epochs: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.5096	0.2	3094	0.5142
0.4699	0.4	6188	0.4815
0.4503	0.6	9282	0.4479
0.4359	0.8	12376	0.4406
0.4266	1.0	15470	0.4249
0.4181	1.2	18564	0.4146
0.4126	1.4	21658	0.4122
0.4076	1.6	24752	0.4043
0.4022	1.8	27846	0.4012
0.3969	2.0	30940	0.3975
0.3874	2.2	34034	0.3964
0.3865	2.4	37128	0.3813
0.379	2.6	40222	0.3783
0.3772	2.8	43316	0.3750
0.3735	3.0	46410	0.3765
0.3637	3.2	49504	0.3659
0.3669	3.4	52598	0.3610
0.3577	3.6	55692	0.3615
0.3578	3.8	58786	0.3567
0.3563	4.0	61880	0.3510
0.3442	4.2	64974	0.3461
0.3403	4.4	68068	0.3428
0.3385	4.6	71162	0.3442
0.3309	4.8	74256	0.3399
0.3271	5.0	77350	0.3290
0.3225	5.2	80444	0.3299
0.3241	5.4	83538	0.3253
0.321	5.6	86632	0.3258
0.3168	5.8	89726	0.3225
0.3117	6.0	92820	0.3182
0.2992	6.2	95914	0.3187
0.2985	6.4	99008	0.3104
0.2975	6.6	102102	0.3072
0.3021	6.8	105196	0.3018
0.2921	7.0	108290	0.3012
0.2807	7.2	111384	0.2967
0.2758	7.4	114478	0.2962
0.2807	7.6	117572	0.2932
0.2786	7.8	120666	0.2901
0.2778	8.0	123760	0.2846
0.2632	8.2	126854	0.2863
0.262	8.4	129948	0.2809
0.2611	8.6	133042	0.2828
0.2648	8.8	136136	0.2762
0.2632	9.0	139230	0.2730
0.2461	9.2	142324	0.2676
0.2443	9.4	145418	0.2669
0.2435	9.6	148512	0.2655
0.2431	9.8	151606	0.2631
0.2379	10.0	154700	0.2599
0.2275	10.2	157794	0.2583
0.2281	10.4	160888	0.2570
0.2243	10.6	163982	0.2530
0.2222	10.8	167076	0.2541
0.2219	11.0	170170	0.2494
0.2112	11.2	173264	0.2495
0.2077	11.4	176358	0.2471
0.2065	11.6	179452	0.2451
0.2029	11.8	182546	0.2432
0.2073	12.0	185640	0.2426

Framework versions

• PEFT 0.14.0
• Transformers 4.47.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0