train_wic_101112_1760638034

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

• Loss: 0.4611
• Num Input Tokens Seen: 8443576

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.4026	1.0	1222	0.4925	421736
0.4093	2.0	2444	0.4784	844280
0.4079	3.0	3666	0.4698	1266928
0.73	4.0	4888	0.4636	1689112
0.4841	5.0	6110	0.4682	2111392
0.4261	6.0	7332	0.4700	2533592
0.4682	7.0	8554	0.4644	2955304
0.6027	8.0	9776	0.4641	3377216
0.4048	9.0	10998	0.4659	3799208
0.3773	10.0	12220	0.4688	4221160
0.2559	11.0	13442	0.4663	4643512
0.3705	12.0	14664	0.4640	5066080
0.4083	13.0	15886	0.4644	5487840
0.4473	14.0	17108	0.4627	5910224
0.6863	15.0	18330	0.4650	6332064
0.5297	16.0	19552	0.4670	6754408
0.436	17.0	20774	0.4611	7176696
0.5006	18.0	21996	0.4628	7598912
0.4396	19.0	23218	0.4628	8021160
0.4385	20.0	24440	0.4628	8443576

Framework versions

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