lemexp-task1-v2-lemma_object_full_notypes-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: 0.2454

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
0.4873	0.2	3094	0.4845
0.4521	0.4	6188	0.4463
0.4379	0.6	9282	0.4339
0.4256	0.8	12376	0.4230
0.4164	1.0	15470	0.4166
0.4099	1.2	18564	0.4037
0.4011	1.4	21658	0.4006
0.3985	1.6	24752	0.3908
0.3935	1.8	27846	0.3863
0.388	2.0	30940	0.3882
0.3787	2.2	34034	0.3869
0.3788	2.4	37128	0.3700
0.3724	2.6	40222	0.3694
0.3718	2.8	43316	0.3681
0.368	3.0	46410	0.3615
0.3601	3.2	49504	0.3575
0.3592	3.4	52598	0.3544
0.3541	3.6	55692	0.3536
0.3552	3.8	58786	0.3493
0.3517	4.0	61880	0.3444
0.3401	4.2	64974	0.3432
0.3357	4.4	68068	0.3358
0.3368	4.6	71162	0.3340
0.3274	4.8	74256	0.3342
0.3264	5.0	77350	0.3287
0.3193	5.2	80444	0.3240
0.3209	5.4	83538	0.3155
0.3186	5.6	86632	0.3242
0.316	5.8	89726	0.3222
0.3097	6.0	92820	0.3105
0.2982	6.2	95914	0.3095
0.2996	6.4	99008	0.3094
0.2983	6.6	102102	0.3070
0.3019	6.8	105196	0.3034
0.2914	7.0	108290	0.2960
0.2805	7.2	111384	0.2988
0.2776	7.4	114478	0.2953
0.2821	7.6	117572	0.2906
0.2792	7.8	120666	0.2876
0.2786	8.0	123760	0.2857
0.2617	8.2	126854	0.2825
0.2643	8.4	129948	0.2771
0.2618	8.6	133042	0.2797
0.2641	8.8	136136	0.2778
0.262	9.0	139230	0.2736
0.2492	9.2	142324	0.2689
0.2459	9.4	145418	0.2651
0.2421	9.6	148512	0.2658
0.2444	9.8	151606	0.2644
0.2373	10.0	154700	0.2613
0.2288	10.2	157794	0.2592
0.2266	10.4	160888	0.2587
0.2275	10.6	163982	0.2539
0.2236	10.8	167076	0.2532
0.2223	11.0	170170	0.2489
0.213	11.2	173264	0.2506
0.2086	11.4	176358	0.2493
0.2069	11.6	179452	0.2463
0.2031	11.8	182546	0.2450
0.2086	12.0	185640	0.2454

Framework versions

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