Llama-31-8B_task-3_180-samples_config-2_full

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-3, the GaetanMichelet/chat-120_ft_task-3 and the GaetanMichelet/chat-180_ft_task-3 datasets. It achieves the following results on the evaluation set:

• Loss: 1.1381

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
• distributed_type: multi-GPU
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• 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
1.695	0.9412	8	1.6759
1.5868	2.0	17	1.5454
1.473	2.9412	25	1.4198
1.2035	4.0	34	1.2535
1.1293	4.9412	42	1.1991
1.1361	6.0	51	1.1733
1.1333	6.9412	59	1.1571
1.0612	8.0	68	1.1462
0.9895	8.9412	76	1.1392
0.9858	10.0	85	1.1381
0.939	10.9412	93	1.1420
0.8747	12.0	102	1.1664
0.8694	12.9412	110	1.1780
0.8188	14.0	119	1.2246
0.697	14.9412	127	1.2348
0.6048	16.0	136	1.3102
0.5898	16.9412	144	1.3190

Framework versions

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