AlekseyKorshuk/6.7b-ri-reproduce-combined-4-gpu-20-val-v3

<!-- 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. -->

6.7b-ri-reproduce-combined-4-gpu-20-val-v3

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

• —Loss: 3.9434
• —Accuracy: 0.0329
• —Perplexity: 51.5916

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: 9e-07
• —trainbatchsize: 1
• —evalbatchsize: 8
• —seed: 100
• —distributed_type: multi-GPU
• —num_devices: 4
• —totaltrainbatch_size: 4
• —totalevalbatch_size: 32
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: constant
• —num_epochs: 15.0

Training results

Framework versions

• —Transformers 4.26.0.dev0
• —Pytorch 1.12.1+cu113
• —Datasets 2.3.2
• —Tokenizers 0.12.1