llama3_8b_helpsteer_helpful_rm

This model is a fine-tuned version of Jennny/llama3_8b_sft_helpsteer on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.0740
• Accuracy: 0.6381

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: 1e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• total_eval_batch_size: 16
• optimizer: Use OptimizerNames.PAGED_ADAMW with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2319	0.1770	10	0.7814	0.5603
0.1672	0.3540	20	0.8783	0.6059
0.2095	0.5310	30	0.7078	0.6327
0.1445	0.7080	40	0.7842	0.6354
0.2071	0.8850	50	0.6209	0.6568
0.0539	1.0531	60	0.7969	0.6193
0.0391	1.2301	70	1.0878	0.6327
0.0172	1.4071	80	1.1465	0.6327
0.0193	1.5841	90	1.1174	0.6408
0.0277	1.7611	100	1.0800	0.6381
0.0184	1.9381	110	1.0740	0.6381

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.1