jjee2/ashwinpatti__llama381binstruct_summarize_short

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: 2.7349

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
1.5288	2.5	25	1.5540
0.4619	5.0	50	1.8601
0.1104	7.5	75	2.2719
0.0318	10.0	100	2.3521
0.0215	12.5	125	2.3788
0.0042	15.0	150	2.4808
0.0024	17.5	175	2.5740
0.0013	20.0	200	2.6431
0.0012	22.5	225	2.6684
0.001	25.0	250	2.6918
0.0009	27.5	275	2.7032
0.0008	30.0	300	2.7120
0.0009	32.5	325	2.7180
0.0007	35.0	350	2.7223
0.0008	37.5	375	2.7261
0.0007	40.0	400	2.7291
0.0006	42.5	425	2.7314
0.0006	45.0	450	2.7330
0.0006	47.5	475	2.7340
0.0006	50.0	500	2.7349

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.2
• Tokenizers 0.20.1