collapse_gemma-2-9b_hs2_replace_iter1_sftsd2

This model is a fine-tuned version of google/gemma-2-9b on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9313
• Num Input Tokens Seen: 5236236

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: 8e-06
• train_batch_size: 4
• eval_batch_size: 16
• seed: 2
• gradient_accumulation_steps: 32
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
No log	0	0	1.2335	0
1.0584	0.0511	5	1.0644	270332
1.0014	0.1021	10	0.9817	541500
0.9947	0.1532	15	0.9708	812584
0.9594	0.2043	20	0.9624	1083956
0.9482	0.2553	25	0.9565	1356232
0.9359	0.3064	30	0.9534	1618536
0.9857	0.3575	35	0.9502	1882576
0.9566	0.4086	40	0.9469	2154872
0.9121	0.4596	45	0.9447	2423272
0.917	0.5107	50	0.9428	2695312
0.9864	0.5618	55	0.9412	2965956
0.9304	0.6128	60	0.9401	3236124
0.8757	0.6639	65	0.9384	3504696
0.9522	0.7150	70	0.9370	3779560
0.9782	0.7660	75	0.9360	4047404
0.9433	0.8171	80	0.9352	4319468
0.9083	0.8682	85	0.9336	4590728
0.9983	0.9192	90	0.9326	4860344
0.9986	0.9703	95	0.9319	5128848

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1