metadata
library_name: transformers
license: apache-2.0
base_model: HuggingFaceTB/SmolLM2-135M
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: smol-135-tq-closure-augment-synthetic
results: []
smol-135-tq-closure-augment-synthetic
This model is a fine-tuned version of HuggingFaceTB/SmolLM2-135M on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1898
- < Precision: 0.9121
- < Recall: 0.9051
- < F1-score: 0.9086
- < Support: 7717.0
Precision: 0.9113
Recall: 0.9016
F1-score: 0.9065
Support: 7717.0
- = Precision: 0.7992
- = Recall: 0.8098
- = F1-score: 0.8045
- = Support: 3244.0
- Precision: 0.7401
- Recall: 0.7950
- F1-score: 0.7666
- Support: 1322.0
- Accuracy: 0.8810
- Macro Avg Precision: 0.8407
- Macro Avg Recall: 0.8529
- Macro Avg F1-score: 0.8465
- Macro Avg Support: 20000.0
- Weighted Avg Precision: 0.8821
- Weighted Avg Recall: 0.8810
- Weighted Avg F1-score: 0.8815
- Weighted Avg Support: 20000.0
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: 0.001
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 512
- total_eval_batch_size: 256
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.2065 | 1.0 | 2708 | 0.1948 | 0.9182 | 0.8800 | 0.8987 | 7717.0 | 0.9012 | 0.8923 | 0.8967 | 7717.0 | 0.7478 | 0.8576 | 0.7990 | 3244.0 | 0.7788 | 0.7322 | 0.7548 | 1322.0 | 0.8713 | 0.8365 | 0.8405 | 0.8373 | 20000.0 | 0.8748 | 0.8713 | 0.8722 | 20000.0 |
| 0.1833 | 2.0 | 5416 | 0.1898 | 0.9121 | 0.9051 | 0.9086 | 7717.0 | 0.9113 | 0.9016 | 0.9065 | 7717.0 | 0.7992 | 0.8098 | 0.8045 | 3244.0 | 0.7401 | 0.7950 | 0.7666 | 1322.0 | 0.8810 | 0.8407 | 0.8529 | 0.8465 | 20000.0 | 0.8821 | 0.8810 | 0.8815 | 20000.0 |
| 0.1415 | 3.0 | 8124 | 0.2006 | 0.8913 | 0.9220 | 0.9064 | 7717.0 | 0.9039 | 0.9116 | 0.9077 | 7717.0 | 0.8096 | 0.7747 | 0.7917 | 3244.0 | 0.8018 | 0.6853 | 0.7390 | 1322.0 | 0.8784 | 0.8516 | 0.8234 | 0.8362 | 20000.0 | 0.8770 | 0.8784 | 0.8772 | 20000.0 |
| 0.1136 | 4.0 | 10832 | 0.2063 | 0.9045 | 0.9136 | 0.9090 | 7717.0 | 0.9038 | 0.9106 | 0.9072 | 7717.0 | 0.7968 | 0.8039 | 0.8004 | 3244.0 | 0.7876 | 0.6899 | 0.7355 | 1322.0 | 0.8799 | 0.8482 | 0.8295 | 0.8380 | 20000.0 | 0.8790 | 0.8799 | 0.8792 | 20000.0 |
| 0.1051 | 5.0 | 13540 | 0.2285 | 0.9131 | 0.9079 | 0.9105 | 7717.0 | 0.9138 | 0.9093 | 0.9115 | 7717.0 | 0.7882 | 0.7975 | 0.7928 | 3244.0 | 0.7313 | 0.7557 | 0.7433 | 1322.0 | 0.8804 | 0.8366 | 0.8426 | 0.8395 | 20000.0 | 0.8811 | 0.8804 | 0.8807 | 20000.0 |
Framework versions
- Transformers 4.47.1
- Pytorch 2.5.1+cu124
- Datasets 3.0.1
- Tokenizers 0.21.0