guilhermemoraisr/sofascore-multitask

meta-llama/Llama-2-7b-hf

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

• Loss: 0.2159
• Rouge1: 0.8791
• Rouge2: 0.7599
• Rougel: 0.8661

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel
0.2548	0.9990	513	0.2430	0.8707	0.7454	0.8577
0.2351	2.0	1027	0.2333	0.8717	0.7503	0.8613
0.2108	2.9990	1540	0.2205	0.8745	0.7543	0.8634
0.1954	4.0	2054	0.2159	0.8791	0.7599	0.8661
0.1998	4.9951	2565	0.2147	0.8822	0.7606	0.8661

Framework versions

• PEFT 0.13.1
• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0