lemexp-task1-min_symbols_template_full-deepseek-coder-1.3b-base-ddp-12lr

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: 0.1895

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: 0.0012
• 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
0.387	0.2000	2907	0.3682
0.3641	0.4001	5814	0.3537
0.3563	0.6001	8721	0.3494
0.3467	0.8001	11628	0.3382
0.3373	1.0001	14535	0.3465
0.3347	1.2002	17442	0.3315
0.3396	1.4002	20349	0.3322
0.3317	1.6002	23256	0.3243
0.3245	1.8002	26163	0.3237
0.3219	2.0003	29070	0.3167
0.3166	2.2003	31977	0.3250
0.3179	2.4003	34884	0.3165
0.3142	2.6004	37791	0.3107
0.3102	2.8004	40698	0.3038
0.3094	3.0004	43605	0.3081
0.3039	3.2004	46512	0.2971
0.3023	3.4005	49419	0.2910
0.297	3.6005	52326	0.2912
0.2947	3.8005	55233	0.2958
0.2925	4.0006	58140	0.2862
0.2848	4.2006	61047	0.2842
0.2842	4.4006	63954	0.2797
0.2815	4.6006	66861	0.2815
0.2795	4.8007	69768	0.2784
0.2735	5.0007	72675	0.2715
0.2685	5.2007	75582	0.2766
0.2698	5.4007	78489	0.2687
0.268	5.6008	81396	0.2648
0.2623	5.8008	84303	0.2647
0.2613	6.0008	87210	0.2587
0.2568	6.2009	90117	0.2573
0.2544	6.4009	93024	0.2553
0.2519	6.6009	95931	0.2518
0.2514	6.8009	98838	0.2525
0.2519	7.0010	101745	0.2512
0.239	7.2010	104652	0.2441
0.2407	7.4010	107559	0.2456
0.2397	7.6010	110466	0.2433
0.2342	7.8011	113373	0.2364
0.2329	8.0011	116280	0.2316
0.2232	8.2011	119187	0.2307
0.2213	8.4012	122094	0.2294
0.2223	8.6012	125001	0.2230
0.2199	8.8012	127908	0.2216
0.2174	9.0012	130815	0.2211
0.208	9.2013	133722	0.2193
0.2089	9.4013	136629	0.2162
0.2047	9.6013	139536	0.2120
0.2028	9.8013	142443	0.2098
0.2005	10.0014	145350	0.2064
0.1923	10.2014	148257	0.2040
0.1926	10.4014	151164	0.2023
0.1887	10.6015	154071	0.2010
0.1883	10.8015	156978	0.2004
0.1865	11.0015	159885	0.1963
0.1769	11.2015	162792	0.1954
0.1744	11.4016	165699	0.1928
0.1721	11.6016	168606	0.1914
0.1711	11.8016	171513	0.1895

Framework versions

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