llama381binstruct_summarize_short_merged

This model is a fine-tuned version of NousResearch/Meta-Llama-3.1-8B-Instruct on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 2.7380

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: 1
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 30
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss
1.5177	2.5	25	1.7211
0.432	5.0	50	1.8956
0.1093	7.5	75	2.5105
0.0237	10.0	100	2.4890
0.0133	12.5	125	2.5161
0.0039	15.0	150	2.5213
0.0049	17.5	175	2.5015
0.0025	20.0	200	2.6358
0.0013	22.5	225	2.7310
0.0047	25.0	250	2.5421
0.0013	27.5	275	2.6042
0.0009	30.0	300	2.6729
0.0008	32.5	325	2.6970
0.0007	35.0	350	2.7104
0.0007	37.5	375	2.7201
0.0006	40.0	400	2.7271
0.0006	42.5	425	2.7318
0.0006	45.0	450	2.7348
0.0006	47.5	475	2.7373
0.0006	50.0	500	2.7380

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.20.1