results_packing

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.4308

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: 7.5e-05
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.571	0.1632	250	0.4770
0.401	0.3264	500	0.4632
0.465	0.4896	750	0.4533
0.4655	0.6527	1000	0.4458
0.406	0.8159	1250	0.4436
0.4921	0.9791	1500	0.4450
0.5231	1.1423	1750	0.4393
0.3529	1.3055	2000	0.4324
0.3498	1.4687	2250	0.4334
0.55	1.6319	2500	0.4286
0.3265	1.7950	2750	0.4275
0.351	1.9582	3000	0.4242
0.3074	2.1214	3250	0.4334
0.3342	2.2846	3500	0.4299
0.343	2.4478	3750	0.4305
0.3406	2.6110	4000	0.4306
0.3175	2.7742	4250	0.4308
0.4474	2.9373	4500	0.4308

Framework versions

• PEFT 0.10.0
• Transformers 4.40.1
• Pytorch 2.1.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1