---
language: ja
license: mit
widget:
- text: "自然言語処理が面白い"
metrics:
- accuracy
- f1
---

# bert-japanese_finetuned-sentiment-analysis

This model was trained from scratch on the Japanese Sentiment Polarity Dictionary dataset.

## Pre-trained model

jarvisx17/japanese-sentiment-analysis<br/>
Link : https://huggingface.co/jarvisx17/japanese-sentiment-analysis

## Training Data

The model was trained on Japanese Sentiment Polarity Dictionary dataset.<br/>
link : https://www.cl.ecei.tohoku.ac.jp/Open_Resources-Japanese_Sentiment_Polarity_Dictionary.html

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10

## Usage

You can use cURL to access this model:

Python API:

```
from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("minutillamolinara/bert-japanese_finetuned-sentiment-analysis")

model = AutoModelForSequenceClassification.from_pretrained("minutillamolinara/bert-japanese_finetuned-sentiment-analysis")

inputs = tokenizer("自然言語処理が面白い", return_tensors="pt")

outputs = model(**inputs)
```

### Dependencies
- !pip install fugashi
- !pip install unidic_lite

## Licenses

MIT