metadata

library_name: peft
tags:
  - trl
  - sft
  - generated_from_trainer
base_model: mistralai/Mistral-7B-Instruct-v0.1
datasets:
  - generator
model-index:
  - name: mistral_instruct_generation
    results: []

mistral_instruct_generation

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.1 on the generator dataset. It achieves the following results on the evaluation set:

Loss: 0.3233

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: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
lr_scheduler_warmup_steps: 0.03
training_steps: 500
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.05	20	0.5742
No log	0.09	40	0.5366
No log	0.14	60	0.5206
No log	0.18	80	0.5128
0.553	0.23	100	0.5081
0.553	0.27	120	0.5027
0.553	0.32	140	0.4991
0.553	0.36	160	0.4961
0.553	0.41	180	0.4938
0.498	0.45	200	0.4892
0.498	0.5	220	0.4848
0.498	0.54	240	0.4804
0.498	0.59	260	0.4764
0.498	0.63	280	0.4698
0.4806	0.68	300	0.4630
0.4806	0.72	320	0.4543
0.4806	0.77	340	0.4457
0.4806	0.81	360	0.4324
0.4806	0.86	380	0.4200
0.4393	0.9	400	0.4105
0.4393	0.95	420	0.3930
0.4393	1.0	440	0.3763
0.4393	1.04	460	0.3614
0.4393	1.09	480	0.3438
0.3692	1.13	500	0.3233

Framework versions

PEFT 0.10.1.dev0
Transformers 4.39.3
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2