llama_finetune_race_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: 3.2366

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
1.2116	1.0	150	1.3188
0.8601	2.0	300	1.3724
0.5773	3.0	450	1.4924
0.6222	4.0	600	1.6729
0.2511	5.0	750	1.8350
0.1554	6.0	900	2.0826
0.1467	7.0	1050	2.2027
0.0909	8.0	1200	2.2817
0.0713	9.0	1350	2.3923
0.0501	10.0	1500	2.6003
0.0526	11.0	1650	2.5589
0.0522	12.0	1800	2.5545
0.0485	13.0	1950	2.7130
0.0297	14.0	2100	2.8527
0.03	15.0	2250	2.8907
0.0327	16.0	2400	3.0280
0.0351	17.0	2550	3.0299
0.0381	18.0	2700	3.1626
0.031	19.0	2850	3.2051
0.028	20.0	3000	3.2366

Framework versions

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