myBit-Llama2-jp-127M-2B4TLike-aozora-sort-3epc

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.2302

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: 0.0024
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 96
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
6.9258	0.0883	100	5.2708
4.7937	0.1765	200	4.4660
4.2565	0.2648	300	4.1755
3.9951	0.3530	400	4.0060
3.8438	0.4413	500	3.8854
3.7223	0.5296	600	3.7829
3.6523	0.6178	700	3.7125
3.5985	0.7061	800	3.6535
3.5666	0.7944	900	3.6039
3.5519	0.8826	1000	3.5693
3.5365	0.9709	1100	3.5404
3.6085	1.0591	1200	3.5638
3.4953	1.1474	1300	3.4983
3.425	1.2357	1400	3.4737
3.3693	1.3239	1500	3.4579
3.3396	1.4122	1600	3.4431
3.3187	1.5004	1700	3.4259
3.3013	1.5887	1800	3.4121
3.3036	1.6770	1900	3.4004
3.2947	1.7652	2000	3.3808
3.3041	1.8535	2100	3.3653
3.304	1.9417	2200	3.3541
3.3582	2.0300	2300	3.4233
3.3097	2.1183	2400	3.3351
3.2426	2.2065	2500	3.3234
3.2034	2.2948	2600	3.3149
3.1675	2.3831	2700	3.3033
3.1611	2.4713	2800	3.2953
3.1344	2.5596	2900	3.2832
3.1391	2.6478	3000	3.2729
3.1324	2.7361	3100	3.2572
3.1355	2.8244	3200	3.2440
3.1417	2.9126	3300	3.2302

Framework versions

• Transformers 4.47.1
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1