Gemma-2-2B_task-1_60-samples_config-2

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

• Loss: 1.3489

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
2.491	0.6957	2	2.2964
2.421	1.7391	5	2.1330
2.0067	2.7826	8	1.7165
1.6227	3.8261	11	1.5108
1.4202	4.8696	14	1.4400
1.3321	5.9130	17	1.3852
1.1652	6.9565	20	1.3695
1.0665	8.0	23	1.3489
0.967	8.6957	25	1.3494
0.8164	9.7391	28	1.3845
0.7346	10.7826	31	1.4393
0.5691	11.8261	34	1.5135
0.4731	12.8696	37	1.6298
0.3309	13.9130	40	1.8103
0.2569	14.9565	43	2.0895

Framework versions

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