gemma2b-summarize-gemini1_5flash-8k

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: 2.5133

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: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• 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
2.7544	0.9630	13	2.8722
1.7723	2.0	27	2.6064
1.4023	2.9630	40	2.5710
1.2778	4.0	54	2.5349
1.1848	4.9630	67	2.5176
1.1522	6.0	81	2.5045
1.1305	6.9630	94	2.5065
1.1075	8.0	108	2.5136
1.1049	8.9630	121	2.5129
1.1048	9.6296	130	2.5133

Framework versions

• PEFT 0.11.1
• Transformers 4.40.1
• Pytorch 2.2.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1