train_winogrande_101112_1760638068

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

• Loss: 0.2313
• Num Input Tokens Seen: 38366624

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: 0.03
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.2313	1.0	9090	0.2314	1917952
0.2347	2.0	18180	0.2317	3835840
0.2335	3.0	27270	0.2313	5753152
0.2317	4.0	36360	0.2316	7672000
0.2319	5.0	45450	0.2314	9590080
0.2293	6.0	54540	0.2314	11509088
0.2362	7.0	63630	0.2317	13427712
0.2314	8.0	72720	0.2313	15346672
0.2329	9.0	81810	0.2315	17265344
0.2319	10.0	90900	0.2315	19184224
0.2319	11.0	99990	0.2314	21102912
0.2324	12.0	109080	0.2314	23021312
0.2324	13.0	118170	0.2314	24938688
0.2309	14.0	127260	0.2314	26857088
0.2319	15.0	136350	0.2313	28775840
0.2324	16.0	145440	0.2313	30693088
0.2298	17.0	154530	0.2313	32612480
0.2309	18.0	163620	0.2313	34530176
0.2324	19.0	172710	0.2315	36447600
0.2319	20.0	181800	0.2313	38366624

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4