metadata

license: gemma
library_name: peft
tags:
  - trl
  - reward-trainer
  - generated_from_trainer
base_model: google/gemma-2b
metrics:
  - accuracy
model-index:
  - name: RM-HH-AllMix_harmless_gpt3_20000_gemma2b_shuffleFalse_extractchosenTrue
    results: []

RM-HH-AllMix_harmless_gpt3_20000_gemma2b_shuffleFalse_extractchosenTrue

This model is a fine-tuned version of google/gemma-2b on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.1748
Accuracy: 0.9237

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: 1.41e-05
train_batch_size: 1
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7416	0.04	250	0.5877	0.7074
0.7004	0.08	500	0.4614	0.7980
0.6298	0.13	750	0.3453	0.8477
0.5979	0.17	1000	0.2723	0.8774
0.5842	0.21	1250	0.2469	0.8868
0.6257	0.25	1500	0.2255	0.8973
0.5833	0.29	1750	0.2103	0.9071
0.6368	0.33	2000	0.2061	0.9082
0.5854	0.38	2250	0.2063	0.9105
0.5458	0.42	2500	0.1990	0.9127
0.6079	0.46	2750	0.1993	0.9135
0.5819	0.5	3000	0.1917	0.9165
0.5823	0.54	3250	0.1844	0.9180
0.618	0.59	3500	0.1869	0.9188
0.6075	0.63	3750	0.1885	0.9169
0.5685	0.67	4000	0.1848	0.9191
0.5718	0.71	4250	0.1848	0.9206
0.5697	0.75	4500	0.1819	0.9210
0.5719	0.79	4750	0.1769	0.9229
0.5774	0.84	5000	0.1779	0.9218
0.5331	0.88	5250	0.1745	0.9233
0.564	0.92	5500	0.1752	0.9237
0.567	0.96	5750	0.1748	0.9237

Framework versions

PEFT 0.10.0
Transformers 4.38.2
Pytorch 2.1.2+cu121
Datasets 2.18.0
Tokenizers 0.15.2