lemexp-task1-v2-lemma_object_full-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: 0.2261

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
0.4519	0.2	3094	0.4528
0.4192	0.4	6188	0.4162
0.4051	0.6	9282	0.4043
0.3928	0.8	12376	0.3904
0.3846	1.0	15470	0.3827
0.3797	1.2	18564	0.3772
0.3744	1.4	21658	0.3697
0.3697	1.6	24752	0.3640
0.3643	1.8	27846	0.3624
0.3614	2.0	30940	0.3526
0.3546	2.2	34034	0.3512
0.3503	2.4	37128	0.3487
0.345	2.6	40222	0.3421
0.3449	2.8	43316	0.3431
0.3421	3.0	46410	0.3432
0.335	3.2	49504	0.3359
0.3351	3.4	52598	0.3336
0.33	3.6	55692	0.3340
0.3283	3.8	58786	0.3282
0.3266	4.0	61880	0.3166
0.317	4.2	64974	0.3149
0.3122	4.4	68068	0.3149
0.313	4.6	71162	0.3147
0.3043	4.8	74256	0.3130
0.3019	5.0	77350	0.3036
0.2952	5.2	80444	0.3000
0.2996	5.4	83538	0.3003
0.2957	5.6	86632	0.2993
0.2935	5.8	89726	0.3047
0.2885	6.0	92820	0.2928
0.2755	6.2	95914	0.2915
0.2763	6.4	99008	0.2875
0.2755	6.6	102102	0.2855
0.2811	6.8	105196	0.2812
0.2704	7.0	108290	0.2796
0.26	7.2	111384	0.2776
0.2564	7.4	114478	0.2691
0.2613	7.6	117572	0.2702
0.2568	7.8	120666	0.2684
0.2579	8.0	123760	0.2643
0.2422	8.2	126854	0.2624
0.243	8.4	129948	0.2619
0.2421	8.6	133042	0.2583
0.2455	8.8	136136	0.2575
0.2428	9.0	139230	0.2511
0.2286	9.2	142324	0.2478
0.227	9.4	145418	0.2507
0.2246	9.6	148512	0.2474
0.2273	9.8	151606	0.2452
0.2211	10.0	154700	0.2432
0.2117	10.2	157794	0.2434
0.2098	10.4	160888	0.2377
0.2092	10.6	163982	0.2376
0.2073	10.8	167076	0.2355
0.2051	11.0	170170	0.2303
0.1966	11.2	173264	0.2321
0.1923	11.4	176358	0.2294
0.1913	11.6	179452	0.2275
0.189	11.8	182546	0.2267
0.1924	12.0	185640	0.2261

Framework versions

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