train_cola_456_1760637817

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: 0.1791
• Num Input Tokens Seen: 7334376

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: 0.001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 456
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.2715	1.0	1924	0.1828	366712
0.2248	2.0	3848	0.1504	734016
0.06	3.0	5772	0.1474	1100824
0.0987	4.0	7696	0.1370	1467248
0.1496	5.0	9620	0.1377	1834568
0.1048	6.0	11544	0.1368	2201464
0.0998	7.0	13468	0.1327	2568040
0.0692	8.0	15392	0.1429	2934360
0.1352	9.0	17316	0.1388	3301448
0.1646	10.0	19240	0.1534	3668312
0.0392	11.0	21164	0.1525	4034856
0.0098	12.0	23088	0.1952	4401344
0.0435	13.0	25012	0.2015	4767736
0.0112	14.0	26936	0.2036	5134344
0.0063	15.0	28860	0.2176	5501408
0.0082	16.0	30784	0.2322	5867920
0.0041	17.0	32708	0.2507	6234920
0.0038	18.0	34632	0.2496	6601944
0.0029	19.0	36556	0.2597	6968096
0.0039	20.0	38480	0.2629	7334376

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4