llama_finetune_arc_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: 2.9424

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
1.1762	1.0	150	1.1987
0.9486	2.0	300	1.2338
0.6296	3.0	450	1.4406
0.244	4.0	600	1.7124
0.2032	5.0	750	1.8587
0.1091	6.0	900	2.0025
0.1157	7.0	1050	2.0500
0.1458	8.0	1200	2.2667
0.0639	9.0	1350	2.1986
0.0651	10.0	1500	2.3844
0.0621	11.0	1650	2.4926
0.0474	12.0	1800	2.5012
0.0524	13.0	1950	2.6149
0.0789	14.0	2100	2.6728
0.0428	15.0	2250	2.7003
0.0446	16.0	2400	2.7733
0.0762	17.0	2550	2.8436
0.0535	18.0	2700	2.9111
0.04	19.0	2850	2.9168
0.0419	20.0	3000	2.9424

Framework versions

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