gemma2b-classification-gpt4o-100k

This model is a fine-tuned version of google/gemma-2b on the llama-duo/synth_classification_dataset_dedup dataset. It achieves the following results on the evaluation set:

• Loss: 2.8001

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: 0.0002
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 3
• gradient_accumulation_steps: 2
• total_train_batch_size: 48
• total_eval_batch_size: 24
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss
1.4266	0.9969	159	1.9880
1.3029	2.0	319	1.9710
1.2414	2.9969	478	1.9794
1.2012	4.0	638	2.0134
1.1513	4.9969	797	2.0583
1.0951	6.0	957	2.1084
1.0414	6.9969	1116	2.2094
1.0041	8.0	1276	2.3043
0.9481	8.9969	1435	2.3989
0.9006	10.0	1595	2.5173
0.8626	10.9969	1754	2.6419
0.8351	12.0	1914	2.7331
0.8265	12.9969	2073	2.7838
0.8167	14.0	2233	2.7990
0.8075	14.9530	2385	2.8001

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1