SmolLM2-135M-TemporalQuestions

This model is a fine-tuned version of HuggingFaceTB/SmolLM2-135M on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0255
• F1: 0.9832

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: 0.001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 512
• total_eval_batch_size: 128
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	F1
0.0628	1.0	445	0.0735	0.9416
0.0559	2.0	890	0.0500	0.9643
0.0291	3.0	1335	0.0460	0.9700
0.0291	4.0	1780	0.0326	0.9766
0.0065	5.0	2225	0.0385	0.9754
0.021	6.0	2670	0.0301	0.9774
0.0186	7.0	3115	0.0330	0.9783
0.0491	8.0	3560	0.0330	0.9777
0.0097	9.0	4005	0.0255	0.9832
0.0071	10.0	4450	0.0281	0.9826
0.0053	11.0	4895	0.0255	0.9830
0.0059	12.0	5340	0.0297	0.9822
0.0072	13.0	5785	0.0321	0.9817
0.0128	14.0	6230	0.0269	0.9856
0.001	15.0	6675	0.0262	0.9840
0.0066	16.0	7120	0.0326	0.9859
0.0005	17.0	7565	0.0331	0.9855
0.0067	18.0	8010	0.0328	0.9876
0.0001	19.0	8455	0.0335	0.9871
0.0002	20.0	8900	0.0354	0.9855
0.0	21.0	9345	0.0369	0.9865
0.0001	22.0	9790	0.0377	0.9877
0.0	23.0	10235	0.0444	0.9866
0.0	24.0	10680	0.0451	0.9868
0.0	25.0	11125	0.0453	0.9871
0.0	26.0	11570	0.0471	0.9872
0.0	27.0	12015	0.0466	0.9871
0.0	28.0	12460	0.0468	0.9871
0.0	29.0	12905	0.0471	0.9870
0.0001	29.9342	13320	0.0470	0.9871

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.0.1
• Tokenizers 0.21.0