deepseek-coder-6.7b-instruct_Fi__components_size_252_epochs_10_2024-06-21_09-35-14_3556545

This model is a fine-tuned version of deepseek-ai/deepseek-coder-6.7b-instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.4511
• Accuracy: 0.479
• Chrf: 0.046
• Bleu: 0.0
• Sacrebleu: 0.0
• Rouge1: 0.063
• Rouge2: 0.032
• Rougel: 0.063
• Rougelsum: 0.063
• Meteor: 0.149

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: 252
• training_steps: 2520

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Chrf	Bleu	Sacrebleu	Rouge1	Rouge2	Rougel	Rougelsum	Meteor
0.0245	4.0	252	1.2112	0.462	0.57	0.412	0.4	0.502	0.294	0.457	0.493	0.541
0.0673	8.0	504	2.4917	0.483	0.268	0.193	0.2	0.354	0.173	0.326	0.346	0.295
0.1033	12.0	756	3.5579	0.508	0.039	0.025	0.0	0.122	0.077	0.122	0.122	0.158
1.2756	16.0	1008	3.7590	0.481	0.046	0.0	0.0	0.169	0.086	0.169	0.169	0.148
0.247	20.0	1260	3.8850	0.503	0.037	0.0	0.0	0.005	0.004	0.005	0.005	0.136
0.5668	24.0	1512	3.6216	0.518	0.034	0.0	0.0	0.019	0.013	0.021	0.021	0.145
0.0462	28.0	1764	3.5317	0.517	0.037	0.0	0.0	0.008	0.007	0.008	0.008	0.15
0.2198	32.0	2016	3.5121	0.503	0.044	0.0	0.0	0.07	0.035	0.07	0.07	0.156
0.0588	36.0	2268	3.4731	0.488	0.046	0.0	0.0	0.062	0.031	0.062	0.062	0.155
0.0639	40.0	2520	3.4511	0.479	0.046	0.0	0.0	0.063	0.032	0.063	0.063	0.149

Framework versions

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