hansken_human_hql_v2

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the hansh/hansken_hql_large dataset. It achieves the following results on the evaluation set:

• Loss: 0.3031

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: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss
0.3676	0.9994	788	0.3796
0.2968	2.0	1577	0.3381
0.2658	2.9994	2365	0.3186
0.2389	4.0	3154	0.3031
0.2098	4.9994	3942	0.3035
0.185	6.0	4731	0.3079
0.1707	6.9994	5519	0.3125
0.1578	8.0	6308	0.3237
0.1426	8.9994	7096	0.3326

Framework versions

• PEFT 0.12.0
• Transformers 4.43.1
• Pytorch 2.1.2+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1