llama_2_sky_o1_5_full

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.5805

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.8553	0.0838	100	0.7053
0.6924	0.1676	200	0.6750
0.6677	0.2515	300	0.6541
0.6419	0.3353	400	0.6376
0.6254	0.4191	500	0.6261
0.6089	0.5029	600	0.6110
0.594	0.5868	700	0.6031
0.5853	0.6706	800	0.5969
0.574	0.7544	900	0.5901
0.5695	0.8382	1000	0.5849
0.5628	0.9220	1100	0.5809

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1