Jennny/llama3_help_rm_2ep

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: 0.3742
• Accuracy: 0.8948

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: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6626	0.0705	10	0.5923	0.6719
0.5039	0.1410	20	0.5139	0.7568
0.4865	0.2115	30	0.4816	0.7646
0.4814	0.2819	40	0.4566	0.7891
0.4932	0.3524	50	0.4449	0.7979
0.4723	0.4229	60	0.4267	0.8031
0.3906	0.4934	70	0.4042	0.8208
0.3418	0.5639	80	0.3907	0.8245
0.4427	0.6344	90	0.3736	0.8359
0.4022	0.7048	100	0.3578	0.8484
0.3738	0.7753	110	0.3470	0.8542
0.3619	0.8458	120	0.3328	0.8609
0.3266	0.9163	130	0.3256	0.8651
0.2786	0.9868	140	0.3245	0.8693
0.1685	1.0573	150	0.4035	0.8786
0.0421	1.1278	160	0.4395	0.8823
0.0655	1.1982	170	0.3843	0.8828
0.0734	1.2687	180	0.3645	0.8823
0.1441	1.3392	190	0.4277	0.8833
0.1176	1.4097	200	0.4040	0.8896
0.0826	1.4802	210	0.3609	0.8870
0.056	1.5507	220	0.3542	0.8891
0.0487	1.6211	230	0.3668	0.8927
0.0815	1.6916	240	0.3735	0.8938
0.0842	1.7621	250	0.3751	0.8943
0.0868	1.8326	260	0.3758	0.8932
0.0743	1.9031	270	0.3753	0.8938
0.0874	1.9736	280	0.3742	0.8948

Framework versions

• Transformers 4.43.4
• Pytorch 2.1.2+cu121
• Datasets 4.4.1
• Tokenizers 0.19.1