llama2-7b-admissions-qa-merged

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6960
• F1: 0.6667

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: 0.0003
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 2
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	F1
7.8289	0.1143	2	11.7201	0.0
7.76	0.2286	4	11.1490	0.0
7.1921	0.3429	6	9.1765	0.6667
5.2447	0.4571	8	3.0182	0.6667
2.2606	0.5714	10	4.1422	0.6667
2.3043	0.6857	12	1.3106	0.6667
2.5684	0.8	14	7.5408	0.6667
4.4282	0.9143	16	4.1099	0.6667
2.8737	1.0	18	4.5994	0.6667
2.9612	1.1143	20	3.7934	0.6667
2.21	1.2286	22	2.1396	0.6667
1.2427	1.3429	24	1.2675	0.6667
0.8553	1.4571	26	1.0446	0.6667
0.6509	1.5714	28	0.7819	0.6667
0.5359	1.6857	30	0.8467	0.6667
0.5303	1.8	32	0.7678	0.6667
0.596	1.9143	34	1.0589	0.6667
0.6265	2.0	36	0.8152	0.6667
0.5092	2.1143	38	0.7012	0.6667
0.4895	2.2286	40	0.7282	0.6667
0.4743	2.3429	42	0.6964	0.6667
0.4662	2.4571	44	0.7031	0.6667
0.4744	2.5714	46	0.7032	0.6667
0.4507	2.6857	48	0.6957	0.6667
0.4737	2.8	50	0.6960	0.6667

Framework versions

• PEFT 0.14.0
• Transformers 4.50.2
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.1