train_cola_1755694493

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

• Loss: 0.3498
• Num Input Tokens Seen: 3465288

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: 5e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 123
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.2119	0.5	1924	0.2508	173872
0.1252	1.0	3848	0.2795	346872
0.2905	1.5	5772	0.2591	520296
0.31	2.0	7696	0.2402	693752
0.243	2.5	9620	0.2488	867416
0.2176	3.0	11544	0.2401	1040128
0.2172	3.5	13468	0.2428	1212976
0.2667	4.0	15392	0.2426	1386696
0.2669	4.5	17316	0.2381	1559896
0.2104	5.0	19240	0.2482	1733072
0.2037	5.5	21164	0.2389	1906160
0.1723	6.0	23088	0.2377	2079640
0.147	6.5	25012	0.2382	2253000
0.3044	7.0	26936	0.2424	2425920
0.3173	7.5	28860	0.2561	2598960
0.2224	8.0	30784	0.2512	2772144
0.1814	8.5	32708	0.3283	2944864
0.2271	9.0	34632	0.3048	3118472
0.1103	9.5	36556	0.3464	3291720
0.2212	10.0	38480	0.3498	3465288

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1