archit11/worldmodel2

acha

This model is a fine-tuned version of nisten/Biggie-SmoLlm-0.15B-Base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.8727

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: 0.0002
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 5
• training_steps: 800

Training results

Training Loss	Epoch	Step	Validation Loss
3.7094	0.0920	50	3.5122
3.5464	0.1841	100	3.2981
3.4292	0.2761	150	3.1894
3.3576	0.3682	200	3.0917
3.35	0.4602	250	3.0353
3.3334	0.5522	300	2.9864
3.3096	0.6443	350	2.9512
3.2773	0.7363	400	2.9281
3.2343	0.8283	450	2.9118
3.2265	0.9204	500	2.9015
3.0257	1.0124	550	2.8853
2.9092	1.1045	600	2.8748
2.9109	1.1965	650	2.8732
2.9437	1.2885	700	2.8728
2.894	1.3806	750	2.8727
2.9286	1.4726	800	2.8727

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0
• Datasets 3.0.0
• Tokenizers 0.19.1