metadata

library_name: transformers
license: apache-2.0
base_model: HuggingFaceTB/SmolLM2-135M-Instruct
tags:
  - generated_from_trainer
  - finance
model-index:
  - name: check-amount-deverbalizer-smollm2
    results: []
datasets:
  - TrevorJS/check-amount-verbalizer-synthetic-data
language:
  - en
pipeline_tag: text-generation

check-amount-deverbalizer-smollm2

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

Loss: 0.1520
Json Parse Rate: 0.96
Dollar Accuracy: 0.96
Cents Accuracy: 0.96
Digit Count Accuracy: 0.96
Perfect Match: 0.96

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 128
eval_batch_size: 2
seed: 42
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: linear
lr_scheduler_warmup_steps: 100
num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Json Parse Rate	Dollar Accuracy	Cents Accuracy	Digit Count Accuracy	Perfect Match
0.1588	0.2128	200	0.1580	0.9571	0.9271	0.9529	0.95	0.9186
0.1576	0.4255	400	0.1549	0.96	0.9557	0.96	0.96	0.9557
0.1546	0.6383	600	0.1539	0.96	0.9571	0.96	0.96	0.9571
0.154	0.8511	800	0.1541	0.9586	0.95	0.9586	0.9586	0.95
0.1539	1.0638	1000	0.1536	0.96	0.9557	0.96	0.96	0.9557
0.1544	1.2766	1200	0.1524	0.96	0.96	0.96	0.96	0.96
0.1535	1.4894	1400	0.1529	0.96	0.9571	0.96	0.96	0.9571
0.1537	1.7021	1600	0.1525	0.96	0.9586	0.96	0.96	0.9586
0.1548	1.9149	1800	0.1527	0.96	0.9586	0.96	0.96	0.9586
0.1545	2.1277	2000	0.1524	0.96	0.96	0.96	0.96	0.96
0.1534	2.3404	2200	0.1526	0.96	0.96	0.96	0.96	0.96
0.1533	2.5532	2400	0.1522	0.96	0.96	0.96	0.96	0.96
0.1524	2.7660	2600	0.1521	0.96	0.96	0.96	0.96	0.96
0.1527	2.9787	2800	0.1520	0.96	0.96	0.96	0.96	0.96

Framework versions

Transformers 4.48.2
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.0