mrm8488's picture
update model card README.md
21f2cc9
metadata
tags:
  - generated_from_trainer
model-index:
  - name: falcon-7b-ft-codeAlpaca_20k
    results: []

falcon-7b-ft-codeAlpaca_20k

This model is a fine-tuned version of ybelkada/falcon-7b-sharded-bf16 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7470

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: 0.0002
  • train_batch_size: 64
  • eval_batch_size: 64
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: constant
  • lr_scheduler_warmup_ratio: 0.03
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss
0.7623 0.18 50 0.7899
0.7985 0.35 100 0.7680
0.7551 0.53 150 0.7570
0.7261 0.71 200 0.7499
0.8184 0.89 250 0.7461
0.7067 1.06 300 0.7480
0.6801 1.24 350 0.7463
0.6432 1.42 400 0.7423
0.7141 1.6 450 0.7398
0.669 1.77 500 0.7383
0.7177 1.95 550 0.7342
0.6419 2.13 600 0.7553
0.6395 2.3 650 0.7510
0.6255 2.48 700 0.7498
0.5556 2.66 750 0.7474
0.6592 2.84 800 0.7470

Framework versions

  • Transformers 4.31.0.dev0
  • Pytorch 2.0.1+cu118
  • Datasets 2.13.0
  • Tokenizers 0.13.3