train_cb_1757081466

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

• Loss: 0.3225
• Num Input Tokens Seen: 310888

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: 2
• eval_batch_size: 2
• 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
0.5795	0.5044	57	0.3516	15456
0.2361	1.0088	114	1.4061	31408
0.1954	1.5133	171	0.5751	46128
0.4297	2.0177	228	0.4388	62512
0.2269	2.5221	285	0.2298	78272
0.545	3.0265	342	0.4335	94136
0.3603	3.5310	399	0.2376	109880
0.4205	4.0354	456	0.2483	125144
0.5197	4.5398	513	0.1896	139544
0.1052	5.0442	570	0.1859	156296
0.3257	5.5487	627	0.2001	172504
0.0065	6.0531	684	0.2582	187664
0.0299	6.5575	741	0.2114	204128
0.1574	7.0619	798	0.3154	219400
0.0121	7.5664	855	0.2926	235224
0.0126	8.0708	912	0.2721	250648
0.005	8.5752	969	0.2829	266456
0.003	9.0796	1026	0.3064	282256
0.1136	9.5841	1083	0.3222	297968

Framework versions

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