metadata

base_model: meta-llama/Llama-2-13b-hf
tags:
  - generated_from_trainer
model-index:
  - name: radiopaedia-inst_240219-llama2_13b-240220
    results: []

radiopaedia-inst_240219-llama2_13b-240220

This model is a fine-tuned version of meta-llama/Llama-2-13b-hf on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.7007

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: 0.0003
train_batch_size: 8
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
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: 10
num_epochs: 2
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.8308	0.05	20	0.7775
0.767	0.11	40	0.7579
0.7219	0.16	60	0.7441
0.6127	0.21	80	0.7489
0.6799	0.27	100	0.7415
0.7201	0.32	120	0.7369
0.7045	0.37	140	0.7298
0.7259	0.42	160	0.7190
0.8055	0.48	180	0.7158
0.6834	0.53	200	0.7066
0.7885	0.58	220	0.7111
0.71	0.64	240	0.7003
0.7124	0.69	260	0.7008
0.6625	0.74	280	0.7052
0.691	0.8	300	0.6925
0.6148	0.85	320	0.6915
0.6727	0.9	340	0.6821
0.5608	0.96	360	0.6777
0.5981	1.01	380	0.6786
0.5295	1.06	400	0.7046
0.4217	1.12	420	0.7027
0.4026	1.17	440	0.7211
0.4469	1.22	460	0.7030
0.3774	1.27	480	0.7153
0.5217	1.33	500	0.7175
0.3966	1.38	520	0.6978
0.4662	1.43	540	0.7010
0.4038	1.49	560	0.6971
0.4514	1.54	580	0.7009
0.423	1.59	600	0.7069
0.3961	1.65	620	0.7030
0.3723	1.7	640	0.7008
0.3745	1.75	660	0.7008
0.4442	1.81	680	0.7087
0.4094	1.86	700	0.7040
0.3465	1.91	720	0.7015
0.3751	1.97	740	0.7007

Framework versions

Transformers 4.35.2
Pytorch 2.0.1+cu117
Datasets 2.14.6
Tokenizers 0.14.1