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

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

• Loss: 0.8720

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
1.2764	1.0	11	1.3024
1.0771	2.0	22	1.1453
1.0212	3.0	33	1.0337
0.9453	4.0	44	0.9333
0.8729	5.0	55	0.8991
0.8348	6.0	66	0.8805
0.8039	7.0	77	0.8735
0.7673	8.0	88	0.8720
0.7032	9.0	99	0.8838
0.6184	10.0	110	0.9229
0.6169	11.0	121	0.9628
0.4972	12.0	132	1.0208
0.5106	13.0	143	1.0691
0.3363	14.0	154	1.1944
0.2612	15.0	165	1.2480

Framework versions

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