cooking_sft_success_new_mem

This model is a fine-tuned version of meta-llama/Llama-3.1-8B-Instruct on the identity and the cooking_sft_success_new_mem datasets. It achieves the following results on the evaluation set:

• Loss: 0.2559

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 8
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.3971	0.1097	50	0.4079
0.3678	0.2194	100	0.3670
0.3599	0.3291	150	0.3344
0.3186	0.4388	200	0.3239
0.2979	0.5485	250	0.2997
0.2996	0.6582	300	0.2766
0.2887	0.7679	350	0.2653
0.2696	0.8776	400	0.2586
0.2784	0.9872	450	0.2559

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0