|
|
|
--- |
|
tags: |
|
- bertopic |
|
library_name: bertopic |
|
--- |
|
|
|
# ISSR_Visual_Model |
|
|
|
This is a [BERTopic](https://github.com/MaartenGr/BERTopic) model. |
|
BERTopic is a flexible and modular topic modeling framework that allows for the generation of easily interpretable topics from large datasets. |
|
|
|
## Usage |
|
|
|
To use this model, please install BERTopic: |
|
|
|
``` |
|
pip install -U bertopic |
|
``` |
|
|
|
You can use the model as follows: |
|
|
|
```python |
|
from bertopic import BERTopic |
|
topic_model = BERTopic.load("D0men1c0/ISSR_Visual_Model") |
|
|
|
topic_model.get_topic_info() |
|
``` |
|
|
|
You can make predictions as follows: |
|
```python |
|
val_labels = [...] # list of caption |
|
val_images = [...] # list of images |
|
|
|
topic, _ = topic_model.transform(val_labels, images=val_images) |
|
all_topic_info = [topic_model.get_topic_info(t) for t in topic] |
|
all_prediction_info = pd.concat(all_topic_info, ignore_index=True) |
|
|
|
# Visualize predictions: |
|
sample_images = 100 |
|
n_images = min(sample_images, len(val_images)) |
|
n_cols = 4 |
|
n_rows = math.ceil(n_images / n_cols) |
|
|
|
fig, axes = plt.subplots(n_rows, n_cols, figsize=(15, n_rows * 3)) |
|
axes = axes.flatten() |
|
|
|
for i, (path, (_, row)) in enumerate(zip(val_images[:n_images], all_prediction_info.iterrows())): |
|
ax = axes[i] |
|
ax.imshow(Image.open(path)) |
|
ax.axis('off') |
|
ax.set_title(f"Topic {row['Topic']}: {row['KeyBERTInspired'][0]}") |
|
|
|
# Hide unused axes |
|
for j in range(n_images, len(axes)): |
|
axes[j].axis('off') |
|
|
|
plt.tight_layout() |
|
plt.show() |
|
``` |
|
|
|
## Topic overview |
|
|
|
* Number of topics: 5 |
|
* Number of training documents: 2997 |
|
|
|
<details> |
|
<summary>Click here for an overview of all topics.</summary> |
|
|
|
| Topic ID | Topic Keywords | Topic Frequency | Label | |
|
|----------|----------------|-----------------|-------| |
|
| -1 | drug - people - gun - - | 151 | -1_drug_people_gun_ | |
|
| 0 | gun - people - drug - - | 2152 | 0_gun_people_drug_ | |
|
| 1 | drug - gun - - - | 342 | 1_drug_gun__ | |
|
| 2 | people - gun - - - | 287 | 2_people_gun__ | |
|
| 3 | people - gun - drug - - | 65 | 3_people_gun_drug_ | |
|
|
|
</details> |
|
|
|
## Training hyperparameters |
|
|
|
* calculate_probabilities: False |
|
* language: None |
|
* low_memory: False |
|
* min_topic_size: 50 |
|
* n_gram_range: (1, 3) |
|
* nr_topics: None |
|
* seed_topic_list: None |
|
* top_n_words: 5 |
|
* verbose: True |
|
* zeroshot_min_similarity: 0.7 |
|
* zeroshot_topic_list: None |
|
|
|
## Framework versions |
|
|
|
* Numpy: 1.26.4 |
|
* HDBSCAN: 0.8.36 |
|
* UMAP: 0.5.6 |
|
* Pandas: 2.2.2 |
|
* Scikit-Learn: 1.4.1.post1 |
|
* Sentence-transformers: 3.0.1 |
|
* Transformers: 4.39.3 |
|
* Numba: 0.60.0 |
|
* Plotly: 5.22.0 |
|
* Python: 3.12.4 |
|
|
