metadata

library_name: peft
license: other
base_model: deepseek-ai/deepseek-coder-1.3b-base
tags:
  - generated_from_trainer
model-index:
  - name: lemexp-task1-v2-lemma_object_full-deepseek-coder-1.3b-base-ddp-8lr-v2
    results: []

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: 0.2570

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: 6
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4506	0.2	3094	0.4509
0.4176	0.4	6188	0.4119
0.4028	0.6	9282	0.4009
0.3903	0.8	12376	0.3867
0.3791	1.0	15470	0.3844
0.3728	1.2	18564	0.3752
0.3652	1.4	21658	0.3608
0.3604	1.6	24752	0.3574
0.3549	1.8	27846	0.3554
0.3491	2.0	30940	0.3493
0.3411	2.2	34034	0.3406
0.3369	2.4	37128	0.3315
0.3304	2.6	40222	0.3313
0.3269	2.8	43316	0.3309
0.3229	3.0	46410	0.3285
0.3128	3.2	49504	0.3141
0.3128	3.4	52598	0.3127
0.3059	3.6	55692	0.3097
0.3047	3.8	58786	0.3038
0.3003	4.0	61880	0.2949
0.2881	4.2	64974	0.2886
0.2838	4.4	68068	0.2920
0.2821	4.6	71162	0.2878
0.2735	4.8	74256	0.2808
0.2698	5.0	77350	0.2764
0.2596	5.2	80444	0.2720
0.2624	5.4	83538	0.2714
0.2574	5.6	86632	0.2691
0.2542	5.8	89726	0.2630
0.2484	6.0	92820	0.2570

Framework versions

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