llama_finetune_mc_20_cot

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

• Loss: 4.0293

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: 4
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.9129	1.0	70	1.6726
0.2315	2.0	140	2.3392
0.1771	3.0	210	2.4929
0.1428	4.0	280	2.6548
0.1106	5.0	350	2.6539
0.0592	6.0	420	2.8574
0.051	7.0	490	3.0477
0.0568	8.0	560	3.1988
0.0309	9.0	630	3.2646
0.0235	10.0	700	3.4138
0.0243	11.0	770	3.3431
0.0259	12.0	840	3.5628
0.0232	13.0	910	3.6910
0.0221	14.0	980	3.6820
0.0202	15.0	1050	3.8640
0.0181	16.0	1120	3.9228
0.0191	17.0	1190	3.9716
0.0184	18.0	1260	4.0096
0.0186	19.0	1330	4.0225
0.0165	20.0	1400	4.0293

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.13.1
• Tokenizers 0.14.1