Gemma-2-2B_task-1_120-samples_config-2_full_auto

This model is a fine-tuned version of google/gemma-2-2b-it on the GaetanMichelet/chat-60_ft_task-1_auto and the GaetanMichelet/chat-120_ft_task-1_auto datasets. It achieves the following results on the evaluation set:

• Loss: 0.9282

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
2.2743	0.9091	5	2.2774
2.2016	2.0	11	1.9581
1.6799	2.9091	16	1.5711
1.3079	4.0	22	1.2308
1.0919	4.9091	27	1.0583
0.9651	6.0	33	0.9889
0.8719	6.9091	38	0.9588
0.8299	8.0	44	0.9332
0.7912	8.9091	49	0.9282
0.692	10.0	55	0.9332
0.63	10.9091	60	0.9539
0.5784	12.0	66	1.0078
0.4937	12.9091	71	1.0819
0.4029	14.0	77	1.2032
0.3117	14.9091	82	1.3912
0.2521	16.0	88	1.5847

Framework versions

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