train_copa_1757340180

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: 3.8649
• 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
0.0106	0.5	45	0.1546	14176
0.0618	1.0	90	0.0645	28256
0.1491	1.5	135	0.0916	42240
0.0201	2.0	180	0.0808	56480
0.1445	2.5	225	0.0840	70688
0.7415	3.0	270	0.3720	84736
0.2923	3.5	315	0.2409	98784
0.3063	4.0	360	0.2817	113024
0.2257	4.5	405	0.2347	127264
0.2394	5.0	450	0.2330	141440
0.2429	5.5	495	0.2286	155488
0.242	6.0	540	0.2464	169600
0.2266	6.5	585	0.2348	183648
0.2456	7.0	630	0.2445	197792
0.2257	7.5	675	0.2663	211936
0.1853	8.0	720	0.2313	225984
0.1794	8.5	765	0.2428	239936
0.197	9.0	810	0.2475	254112
0.1943	9.5	855	0.2425	268256
0.16	10.0	900	0.2469	282368

Framework versions

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