Llama-31-8B_task-3_60-samples_config-1_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 dataset. It achieves the following results on the evaluation set:

• Loss: 1.1645

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: 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
1.6787	0.8696	5	1.6761
1.6422	1.9130	11	1.5859
1.5493	2.9565	17	1.4786
1.3749	4.0	23	1.3613
1.2143	4.8696	28	1.2559
1.1354	5.9130	34	1.2058
1.0603	6.9565	40	1.1850
1.0521	8.0	46	1.1711
1.0417	8.8696	51	1.1646
1.025	9.9130	57	1.1645
0.9238	10.9565	63	1.1722
0.8752	12.0	69	1.1791
0.8737	12.8696	74	1.1912
0.7418	13.9130	80	1.2409
0.6853	14.9565	86	1.3252
0.6937	16.0	92	1.2866
0.442	16.8696	97	1.3294

Framework versions

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