G0514HMA25H

This model is a fine-tuned version of google/gemma-2b on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: -17.9022

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_steps: 80
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.8009	0.09	10	-0.2103
-1.0796	0.18	20	-2.5087
-3.9095	0.27	30	-5.8273
-7.2606	0.36	40	-9.1358
-10.3742	0.45	50	-12.0428
-13.0694	0.54	60	-14.4715
-15.2349	0.63	70	-16.0930
-16.5217	0.73	80	-16.9969
-17.1884	0.82	90	-17.3938
-17.4707	0.91	100	-17.5554
-17.5868	1.0	110	-17.6315
-17.6564	1.09	120	-17.6735
-17.688	1.18	130	-17.7003
-17.709	1.27	140	-17.7200
-17.7262	1.36	150	-17.7362
-17.7401	1.45	160	-17.7476
-17.7557	1.54	170	-17.7664
-17.7777	1.63	180	-17.7896
-17.7948	1.72	190	-17.8078
-17.8232	1.81	200	-17.8337
-17.8393	1.9	210	-17.8518
-17.8561	1.99	220	-17.8679
-17.8673	2.08	230	-17.8730
-17.8748	2.18	240	-17.8887
-17.8874	2.27	250	-17.8931
-17.8901	2.36	260	-17.8972
-17.8918	2.45	270	-17.8974
-17.8952	2.54	280	-17.9002
-17.898	2.63	290	-17.9012
-17.8994	2.72	300	-17.9019
-17.8999	2.81	310	-17.9020
-17.9012	2.9	320	-17.9022
-17.8998	2.99	330	-17.9022

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.14.0