llama3_rm

This model is a fine-tuned version of meta-llama/Llama-3.1-8B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.3581
Accuracy: 0.8443

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: 2e-06
train_batch_size: 4
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 8
total_train_batch_size: 256
total_eval_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.03
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8063	0.0705	10	0.6141	0.7339
0.5355	0.1410	20	0.5117	0.7604
0.4811	0.2115	30	0.4766	0.7719
0.4922	0.2819	40	0.4512	0.7792
0.4919	0.3524	50	0.4428	0.8
0.476	0.4229	60	0.4174	0.8083
0.3965	0.4934	70	0.4047	0.8161
0.3523	0.5639	80	0.3918	0.8224
0.455	0.6344	90	0.3794	0.8344
0.4096	0.7048	100	0.3682	0.8396
0.3726	0.7753	110	0.3636	0.8417
0.3802	0.8458	120	0.3600	0.8448
0.3908	0.9163	130	0.3585	0.8427
0.4002	0.9868	140	0.3581	0.8443