llama3_1_8B_all_zhtw_lr1e5_ep1_16_32_128

This model is a fine-tuned version of meta-llama/Llama-3.1-8B-Instruct on the multi_turn_miss_func_zh_tw_function_mix_sharegpt_clean1, the multi_turn_miss_param_zh_tw_function_mix_sharegpt_clean1, the multi_turn_zh_tw_function_mix_sharegpt, the irrelevance_zh_tw_sharegpt3000 and the apigen_zhtwV3_remove_sys_sharegpt datasets. It achieves the following results on the evaluation set:

• Loss: 0.6250

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 128
• total_train_batch_size: 128
• optimizer: Use OptimizerNames.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: 0.03
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.9075	0.0673	10	1.1429
0.873	0.1347	20	1.0365
0.7009	0.2020	30	0.9636
0.8234	0.2694	40	0.8995
0.5767	0.3367	50	0.8376
0.6782	0.4041	60	0.7834
0.6734	0.4714	70	0.7381
0.6834	0.5388	80	0.7015
0.5739	0.6061	90	0.6727
0.7269	0.6735	100	0.6514
0.5797	0.7408	110	0.6378
0.5396	0.8082	120	0.6303
0.5815	0.8755	130	0.6265
0.4435	0.9429	140	0.6248

Framework versions

• PEFT 0.17.0
• Transformers 4.57.1
• Pytorch 2.4.1+cu121
• Datasets 4.0.0
• Tokenizers 0.22.2