gemma2b-coding-gpt4o-100k

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

• Loss: 1.6825

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
0.6871	0.9979	235	1.4013
0.6707	2.0	471	1.3993
0.6047	2.9979	706	1.4091
0.5773	4.0	942	1.4428
0.5548	4.9979	1177	1.4904
0.5409	6.0	1413	1.5480
0.5151	6.9979	1648	1.6102
0.4987	8.0	1884	1.6578
0.4875	8.9979	2119	1.6813
0.4904	9.9788	2350	1.6825

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1