---
library_name: peft
license: apache-2.0
base_model: HuggingFaceTB/SmolVLM-256M-Instruct
tags:
- base_model:adapter:HuggingFaceTB/SmolVLM-256M-Instruct
- lora
- transformers
- image
- gui
model-index:
- name: SmolVLM-256M-ScreenTask
  results: []
datasets:
- macpaw-research/Screen2AX-Task
language:
- en
pipeline_tag: image-text-to-text
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# SmolVLM-256M-ScreenTask

This model is a fine-tuned version of [HuggingFaceTB/SmolVLM-256M-Instruct](https://huggingface.co/HuggingFaceTB/SmolVLM-256M-Instruct) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8402

## 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: 0.0001
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- 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: 50
- num_epochs: 0.35
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch  | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| 2.857         | 0.0599 | 20   | 2.7762          |
| 1.8633        | 0.1199 | 40   | 1.8233          |
| 1.1164        | 0.1798 | 60   | 1.0640          |
| 0.8947        | 0.2397 | 80   | 0.8909          |
| 0.8471        | 0.2996 | 100  | 0.8402          |


### Framework versions

- PEFT 0.18.0
- Transformers 4.57.3
- Pytorch 2.9.0+cu126
- Datasets 4.0.0
- Tokenizers 0.22.1