MentaLLaMA-chat-7B-PsyCourse-info-chain-fold6

This model is a fine-tuned version of klyang/MentaLLaMA-chat-7B-hf on the course-info-chain-train-fold6 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0748

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use adamw_torch 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: 0.1
• num_epochs: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.3968	0.3951	10	0.3069
0.1613	0.7901	20	0.1149
0.1248	1.1852	30	0.0981
0.0973	1.5802	40	0.0846
0.119	1.9753	50	0.0818
0.0892	2.3704	60	0.0794
0.0624	2.7654	70	0.0772
0.0461	3.1605	80	0.0750
0.0736	3.5556	90	0.0756
0.0535	3.9506	100	0.0748
0.1773	4.3457	110	0.0758
0.0533	4.7407	120	0.0758

Framework versions

• PEFT 0.12.0
• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3