tinylm1M-stella-2sent_32clust-2025-04-04-13-49

This model is a fine-tuned version of roneneldan/TinyStories-1M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.2485

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: 5e-06
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
2.603	0.8418	500	2.5493
2.5641	1.6835	1000	2.5072
2.5856	2.5253	1500	2.4748
2.4812	3.3670	2000	2.4526
2.3806	4.2088	2500	2.4287
2.365	5.0505	3000	2.4067
2.3335	5.8923	3500	2.3860
2.3483	6.7340	4000	2.3676
2.3202	7.5758	4500	2.3528
2.3296	8.4175	5000	2.3395
2.2774	9.2593	5500	2.3251
2.2768	10.1010	6000	2.3136
2.312	10.9428	6500	2.3000
2.2991	11.7845	7000	2.2903
2.3076	12.6263	7500	2.2813
2.1882	13.4680	8000	2.2751
2.2872	14.3098	8500	2.2669
2.3859	15.1515	9000	2.2619
2.2089	15.9933	9500	2.2582
2.3199	16.8350	10000	2.2533
2.1989	17.6768	10500	2.2504
2.2267	18.5185	11000	2.2499
2.1404	19.3603	11500	2.2485

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1