Gemma-2-2B_task-1_120-samples_config-2_full

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

• Loss: 1.0151

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
2.5293	0.9091	5	2.4961
2.4062	2.0	11	2.1506
1.8384	2.9091	16	1.7335
1.446	4.0	22	1.3520
1.2181	4.9091	27	1.1761
1.0795	6.0	33	1.0930
0.9652	6.9091	38	1.0544
0.9317	8.0	44	1.0245
0.8531	8.9091	49	1.0151
0.7657	10.0	55	1.0207
0.7049	10.9091	60	1.0355
0.6405	12.0	66	1.0797
0.5447	12.9091	71	1.1631
0.4351	14.0	77	1.2858
0.3579	14.9091	82	1.4292
0.2641	16.0	88	1.6355

Framework versions

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