iliazlobin/phi-2-coedit

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

phi-2-coedit

This model is a fine-tuned version of microsoft/phi-2 on an unknown dataset. It achieves the following results on the evaluation set:

• —Loss: 0.7388
• —Rouge1: 0.5206
• —Rouge2: 0.4123
• —Rougel: 0.4979
• —Rougelsum: 0.5032
• —Sacreblue: 28.1346
• —Memory Used: 81917.5
• —Cuda Allocated: 10795.7861
• —Cuda Reserved: 74746.0
• —Ram Usage: 24042.6719
• —Em: 0.0
• —Gen Len: 120.6545

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: 2e-05
• —trainbatchsize: 35
• —evalbatchsize: 35
• —seed: 42
• —gradientaccumulationsteps: 4
• —totaltrainbatch_size: 140
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: linear
• —lrschedulerwarmup_steps: 1
• —num_epochs: 2
• —mixedprecisiontraining: Native AMP

Training results

Framework versions

• —Transformers 4.39.3
• —Pytorch 2.2.2
• —Datasets 2.18.0
• —Tokenizers 0.15.2