train_copa_1754502828

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: 0.1487
• Num Input Tokens Seen: 281856

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: 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
0.2326	0.5	45	0.2184	14016
0.2002	1.0	90	0.1768	28096
0.1371	1.5	135	0.1628	42144
0.1187	2.0	180	0.1755	56128
0.0846	2.5	225	0.1746	70272
0.1015	3.0	270	0.1487	84352
0.0375	3.5	315	0.1653	98464
0.0238	4.0	360	0.1695	112576
0.2209	4.5	405	0.1940	126624
0.0146	5.0	450	0.1788	140832
0.0095	5.5	495	0.1735	154976
0.011	6.0	540	0.1789	169056
0.1398	6.5	585	0.1763	183200
0.0917	7.0	630	0.1792	197344
0.1555	7.5	675	0.1878	211392
0.0518	8.0	720	0.1871	225536
0.0755	8.5	765	0.1859	239680
0.0405	9.0	810	0.1918	253696
0.0005	9.5	855	0.1885	267840
0.0032	10.0	900	0.1888	281856

Framework versions

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