Hugging Face's logo

HCHSmost
/
Terry-ft

PEFT
TensorBoard
Safetensors
Generated from Trainer
Model card Files Metrics Community
Terry-ft / README.md
HCHSmost's picture
HCHSmost
HCHSmost/TerryGPT-refined_version
8276728 verified
preview code
|
raw
history blame contribute delete
No virus
1.91 kB
metadata
base_model: TheBloke/Mistral-7B-Instruct-v0.2-GPTQ
library_name: peft
license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: Terry-ft
    results: []

Terry-ft

This model is a fine-tuned version of TheBloke/Mistral-7B-Instruct-v0.2-GPTQ on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7287

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: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 2
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
2.5921 0.8696 5 2.0205
1.5501 1.9130 11 1.4288
1.1411 2.9565 17 1.0682
0.8805 4.0 23 0.8740
0.8972 4.8696 28 0.8145
0.7032 5.9130 34 0.7677
0.6435 6.9565 40 0.7422
0.6246 8.0 46 0.7309
0.6399 8.6957 50 0.7287

Framework versions

  • PEFT 0.11.1
  • Transformers 4.42.4
  • Pytorch 2.1.0+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1