gemma2b-summarize-gpt4o-256k

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

• Loss: 2.5990

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: 4
• total_train_batch_size: 96
• 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: 10

Training results

Training Loss	Epoch	Step	Validation Loss
1.1174	0.9974	292	2.4482
1.0252	1.9983	585	2.4514
0.988	2.9991	878	2.4683
0.9741	4.0	1171	2.5000
0.9342	4.9974	1463	2.5203
0.9201	5.9983	1756	2.5519
0.9054	6.9991	2049	2.5763
0.8902	8.0	2342	2.5922
0.8818	8.9974	2634	2.5982
0.8852	9.9744	2920	2.5990

Framework versions

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