OpenCodeInterpreter-DS-6.7B_En__translations_size_104_epochs_10_2024-06-22_03-22-06_3557996

This model is a fine-tuned version of m-a-p/OpenCodeInterpreter-DS-6.7B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.1148
• Accuracy: 0.042
• Chrf: 0.511
• Bleu: 0.407
• Sacrebleu: 0.4
• Rouge1: 0.516
• Rouge2: 0.259
• Rougel: 0.474
• Rougelsum: 0.508
• Meteor: 0.404

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: 0.001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 3407
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 4
• total_eval_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-06
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 104
• training_steps: 1040

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Chrf	Bleu	Sacrebleu	Rouge1	Rouge2	Rougel	Rougelsum	Meteor
0.145	4.0	104	1.2220	0.035	0.711	0.572	0.6	0.657	0.424	0.585	0.646	0.499
0.1118	8.0	208	1.3598	0.034	0.708	0.563	0.6	0.66	0.421	0.586	0.641	0.519
0.1119	12.0	312	1.4868	0.037	0.719	0.586	0.6	0.674	0.441	0.597	0.659	0.522
0.4556	16.0	416	1.6801	0.04	0.704	0.587	0.6	0.697	0.468	0.613	0.682	0.525
0.1586	20.0	520	2.0607	0.043	0.672	0.552	0.6	0.659	0.422	0.584	0.65	0.46
0.2311	24.0	624	2.6808	0.039	0.57	0.449	0.4	0.574	0.312	0.504	0.566	0.42
1.0024	28.0	728	2.7805	0.055	0.569	0.462	0.5	0.572	0.322	0.52	0.56	0.413
0.2281	32.0	832	2.9914	0.044	0.546	0.44	0.4	0.548	0.296	0.504	0.542	0.409
0.3755	36.0	936	3.0919	0.045	0.508	0.406	0.4	0.522	0.261	0.479	0.515	0.407
0.2274	40.0	1040	3.1148	0.042	0.511	0.407	0.4	0.516	0.259	0.474	0.508	0.404

Framework versions

• Transformers 4.37.0
• Pytorch 2.2.1+cu121
• Datasets 2.20.0
• Tokenizers 0.15.2