Llama-2-7b-hf-finetuned-mrpc-v3

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the glue dataset. It achieves the following results on the evaluation set:

• Loss: 0.6823
• Accuracy: 0.7475
• F1: 0.8245

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	1.0	230	0.6528	0.625	0.6982
No log	2.0	460	0.6217	0.6936	0.8159
0.6443	3.0	690	0.6033	0.6985	0.7993
0.6443	4.0	920	0.6240	0.6838	0.8089
0.6173	5.0	1150	0.5451	0.7255	0.8170
0.6173	6.0	1380	0.5380	0.7451	0.8188
0.5776	7.0	1610	0.5376	0.7426	0.8346
0.5776	8.0	1840	0.5518	0.7230	0.8243
0.5353	9.0	2070	0.5270	0.7475	0.8325
0.5353	10.0	2300	0.5381	0.7377	0.8086
0.5071	11.0	2530	0.5453	0.7181	0.7842
0.5071	12.0	2760	0.5335	0.7475	0.8341
0.5071	13.0	2990	0.5617	0.7083	0.7733
0.492	14.0	3220	0.5343	0.7426	0.8115
0.492	15.0	3450	0.5133	0.7696	0.8423
0.4608	16.0	3680	0.5573	0.7549	0.8366
0.4608	17.0	3910	0.5282	0.7721	0.8447
0.4283	18.0	4140	0.5894	0.7132	0.7710
0.4283	19.0	4370	0.5875	0.7328	0.8239
0.4042	20.0	4600	0.5447	0.7647	0.8339
0.4042	21.0	4830	0.5712	0.7598	0.8399
0.3904	22.0	5060	0.5563	0.7623	0.8301
0.3904	23.0	5290	0.5718	0.7623	0.8364
0.3597	24.0	5520	0.5592	0.7525	0.8250
0.3597	25.0	5750	0.5941	0.7574	0.8364
0.3597	26.0	5980	0.5811	0.7623	0.8370
0.3445	27.0	6210	0.6083	0.7549	0.8339
0.3445	28.0	6440	0.6049	0.75	0.8265
0.3197	29.0	6670	0.6042	0.7549	0.8311
0.3197	30.0	6900	0.6260	0.7377	0.8099
0.3	31.0	7130	0.6438	0.75	0.8229
0.3	32.0	7360	0.6319	0.7402	0.8233
0.2873	33.0	7590	0.6502	0.7402	0.8191
0.2873	34.0	7820	0.6591	0.7426	0.8187
0.2719	35.0	8050	0.6474	0.7451	0.8219
0.2719	36.0	8280	0.6803	0.7598	0.8367
0.2583	37.0	8510	0.6903	0.7475	0.8221
0.2583	38.0	8740	0.6965	0.7525	0.8279
0.2583	39.0	8970	0.6850	0.75	0.8235
0.2423	40.0	9200	0.6823	0.7475	0.8245

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3