llama381binstruct_summarize_short

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: 1.7418

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 30
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss
1.5296	1.25	25	1.0734
0.8156	2.5	50	1.1007
0.3921	3.75	75	1.2204
0.2148	5.0	100	1.3392
0.0884	6.25	125	1.4927
0.0355	7.5	150	1.5451
0.0284	8.75	175	1.6376
0.0164	10.0	200	1.5957
0.0139	11.25	225	1.6211
0.0051	12.5	250	1.5915
0.0042	13.75	275	1.6470
0.0054	15.0	300	1.6318
0.0047	16.25	325	1.6774
0.0031	17.5	350	1.6866
0.0018	18.75	375	1.7033
0.0022	20.0	400	1.7191
0.0021	21.25	425	1.7304
0.0018	22.5	450	1.7366
0.0017	23.75	475	1.7406
0.0018	25.0	500	1.7418

Framework versions

• PEFT 0.12.0
• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1