train_copa_1757340281

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: 0.0314
• Num Input Tokens Seen: 281312

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: 101112
• 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
0.592	0.5	45	0.5497	14144
0.8778	1.0	90	0.3723	28192
0.0636	1.5	135	0.0465	42208
0.0595	2.0	180	0.0365	56256
0.242	2.5	225	0.0338	70368
0.014	3.0	270	0.0341	84320
0.1039	3.5	315	0.0326	98400
0.0307	4.0	360	0.0314	112416
0.3158	4.5	405	0.0345	126496
0.0098	5.0	450	0.0319	140544
0.0163	5.5	495	0.0342	154592
0.0024	6.0	540	0.0315	168768
0.0792	6.5	585	0.0330	182848
0.0327	7.0	630	0.0315	196896
0.1089	7.5	675	0.0345	210912
0.0141	8.0	720	0.0326	225024
0.0397	8.5	765	0.0324	239200
0.0891	9.0	810	0.0335	253152
0.0837	9.5	855	0.0317	267040
0.1442	10.0	900	0.0330	281312

Framework versions

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