Llama-31-8B_task-2_120-samples_config_1

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat_60_ft_t2 and the GaetanMichelet/chat_120_ft_t2 datasets. It achieves the following results on the evaluation set:

• Loss: 1.2868

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: 2
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss
1.512	0.9796	18	1.5410
1.2629	1.9592	36	1.3697
1.1822	2.9932	55	1.3049
1.1282	3.9728	73	1.2868
0.9696	4.9524	91	1.3075
0.7613	5.9864	110	1.3104
0.5878	6.9660	128	1.3776
0.4377	8.0	147	1.4299
0.3467	8.9796	165	1.4920
0.2937	9.9592	183	1.5802
0.2423	10.9932	202	1.6327

Framework versions

• PEFT 0.12.0
• Transformers 4.44.0
• Pytorch 2.1.2+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1