Llama-31-8B_task-3_120-samples_config-2_full_auto

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

• Loss: 1.1200

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
1.6955	0.9091	5	1.6712
1.6465	2.0	11	1.5769
1.516	2.9091	16	1.4843
1.374	4.0	22	1.3648
1.2447	4.9091	27	1.2468
1.1453	6.0	33	1.1844
1.1195	6.9091	38	1.1647
1.1005	8.0	44	1.1478
1.0586	8.9091	49	1.1371
1.0782	10.0	55	1.1279
1.0254	10.9091	60	1.1231
0.9925	12.0	66	1.1200
0.9598	12.9091	71	1.1201
0.9553	14.0	77	1.1259
0.8505	14.9091	82	1.1372
0.865	16.0	88	1.1483
0.8046	16.9091	93	1.1632
0.7313	18.0	99	1.1917
0.7125	18.9091	104	1.2146

Framework versions

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