smol-1.7-tq

This model is a fine-tuned version of HuggingFaceTB/SmolLM2-1.7B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4523
• = Precision: 0.0
• = Recall: 0.0
• = F1-score: 0.0
• = Support: 31.0
• • Precision: 0.0
• • Recall: 0.0
• • F1-score: 0.0
• • Support: 113.0
• < Precision: 0.6985
• < Recall: 0.7925
• < F1-score: 0.7425
• < Support: 424.0

• Precision: 0.6167

• Recall: 0.7955

• F1-score: 0.6948

• Support: 269.0

• Accuracy: 0.6571
• Macro Avg Precision: 0.3288
• Macro Avg Recall: 0.3970
• Macro Avg F1-score: 0.3593
• Macro Avg Support: 837.0
• Weighted Avg Precision: 0.5521
• Weighted Avg Recall: 0.6571
• Weighted Avg F1-score: 0.5995
• Weighted Avg Support: 837.0

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• total_eval_batch_size: 16
• 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: reduce_lr_on_plateau
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	= Precision	= Recall	= F1-score	= Support	- Precision	- Recall	- F1-score	- Support	< Precision	< Recall	< F1-score	< Support	> Precision	> Recall	> F1-score	> Support	Accuracy	Macro Avg Precision	Macro Avg Recall	Macro Avg F1-score	Macro Avg Support	Weighted Avg Precision	Weighted Avg Recall	Weighted Avg F1-score	Weighted Avg Support
3.6479	1.0	25	0.4992	0.0	0.0	0.0	31.0	0.0	0.0	0.0	113.0	0.5950	0.7830	0.6762	424.0	0.5471	0.5613	0.5541	269.0	0.5771	0.2855	0.3361	0.3076	837.0	0.4772	0.5771	0.5206	837.0
2.5144	2.0	50	0.4460	0.0	0.0	0.0	31.0	0.2222	0.0177	0.0328	113.0	0.6438	0.8184	0.7207	424.0	0.6228	0.6691	0.6452	269.0	0.6320	0.3722	0.3763	0.3497	837.0	0.5563	0.6320	0.5768	837.0
2.207	3.0	75	0.4411	0.0	0.0	0.0	31.0	0.0	0.0	0.0	113.0	0.6798	0.8160	0.7417	424.0	0.6288	0.7621	0.6891	269.0	0.6583	0.3271	0.3945	0.3577	837.0	0.5464	0.6583	0.5972	837.0
1.7458	4.0	100	0.4523	0.0	0.0	0.0	31.0	0.0	0.0	0.0	113.0	0.6985	0.7925	0.7425	424.0	0.6167	0.7955	0.6948	269.0	0.6571	0.3288	0.3970	0.3593	837.0	0.5521	0.6571	0.5995	837.0
1.6753	5.0	125	0.4816	0.0	0.0	0.0	31.0	0.0833	0.0177	0.0292	113.0	0.6744	0.8255	0.7423	424.0	0.6497	0.7100	0.6785	269.0	0.6487	0.3518	0.3883	0.3625	837.0	0.5617	0.6487	0.5980	837.0
1.244	6.0	150	0.5072	0.0	0.0	0.0	31.0	0.1412	0.1062	0.1212	113.0	0.6889	0.7783	0.7309	424.0	0.6554	0.6506	0.6530	269.0	0.6177	0.3714	0.3838	0.3763	837.0	0.5787	0.6177	0.5965	837.0
1.04	7.0	175	0.5208	0.0	0.0	0.0	31.0	0.1296	0.0619	0.0838	113.0	0.6803	0.7830	0.7281	424.0	0.6395	0.6989	0.6679	269.0	0.6296	0.3624	0.3860	0.3699	837.0	0.5676	0.6296	0.5948	837.0

Framework versions

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