metadata

library_name: transformers
license: llama3
base_model: meta-llama/Meta-Llama-3-8B-Instruct
tags:
  - trl
  - dpo
  - generated_from_trainer
model-index:
  - name: Llama0-3-8b-v0.1-p-2-lr6e-7-e1
    results: []

Llama0-3-8b-v0.1-p-2-lr6e-7-e1

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0131
Rewards/chosen: -187.5863
Rewards/rejected: -185.7464
Rewards/accuracies: 0.4839
Rewards/margins: -1.8399
Logps/rejected: -18661.3926
Logps/chosen: -18846.9238
Logits/rejected: -6.2589
Logits/chosen: -6.2928

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: 6e-07
train_batch_size: 2
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 8
total_train_batch_size: 128
total_eval_batch_size: 64
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	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.1724	0.2137	100	0.1723	-0.1366	-0.1256	0.4274	-0.0110	-99.3180	-101.9561	-0.1170	-0.1376
0.0221	0.4275	200	0.0190	-147.9461	-146.7242	0.4798	-1.2218	-14759.1787	-14882.8994	-4.3358	-4.3358
0.0129	0.6412	300	0.0151	-161.7966	-159.7617	0.4718	-2.0349	-16062.9199	-16267.9521	-5.5083	-5.5487
0.0129	0.8549	400	0.0138	-195.4678	-193.5756	0.4637	-1.8922	-19444.3184	-19635.0742	-6.3235	-6.3558

Framework versions

Transformers 4.45.1
Pytorch 2.4.1+cu121
Datasets 3.0.0
Tokenizers 0.20.0