train_copa_1757340183

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

• Loss: 0.0750
• Num Input Tokens Seen: 282368

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: 4
• eval_batch_size: 4
• seed: 42
• 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
0.5514	0.5	45	0.5808	14176
0.6249	1.0	90	0.4237	28256
0.1285	1.5	135	0.0999	42240
0.0929	2.0	180	0.1014	56480
0.152	2.5	225	0.0960	70688
0.2109	3.0	270	0.0887	84736
0.0323	3.5	315	0.0812	98784
0.1381	4.0	360	0.0856	113024
0.0128	4.5	405	0.0811	127264
0.1841	5.0	450	0.0821	141440
0.0707	5.5	495	0.0799	155488
0.0243	6.0	540	0.0789	169600
0.0891	6.5	585	0.0776	183648
0.0439	7.0	630	0.0806	197792
0.1076	7.5	675	0.0791	211936
0.0192	8.0	720	0.0789	225984
0.1022	8.5	765	0.0753	239936
0.1065	9.0	810	0.0760	254112
0.0452	9.5	855	0.0764	268256
0.0075	10.0	900	0.0750	282368

Framework versions

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