lora_0-2_3B

This model is a fine-tuned version of meta-llama/Llama-3.2-3B on the gulaschnascher4000/stream-dataset-0-2, the identity-chatgulaschpt, the dolly_15k_de and the alpaca-gpt4_de datasets. It achieves the following results on the evaluation set:

• Loss: 1.5461

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: 8
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 4
• optimizer: Use OptimizerNames.ADAFACTOR and the args are: scale_parameter=True, relative_step=True, warmup_init=True, lr=None
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 0.1

Training results

Framework versions

• PEFT 0.12.0
• Transformers 4.46.1
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3