README.md

---
library_name: transformers
tags:
- chunking
- RAG
license: mit
datasets:
- bookcorpus/bookcorpus
- JeanKaddour/minipile
language:
- en
base_model:
- answerdotai/ModernBERT-large
---

# Chonky modernbert large v1

__Chonky__ is a transformer model that intelligently segments text into meaningful semantic chunks. This model can be used in the RAG systems.



## Model Description

The model processes text and divides it into semantically coherent segments. These chunks can then be fed into embedding-based retrieval systems or language models as part of a RAG pipeline.

⚠️This model was fine-tuned on sequence of length 1024 (by default ModernBERT supports sequence length up to 8192).

## How to use

I've made a small python library for this model: [chonky](https://github.com/mirth/chonky)

Here is the usage:

```
from chonky import ParagraphSplitter

# on the first run it will download the transformer model
splitter = ParagraphSplitter(
  model_id="mirth/chonky_modernbert_large_1",
  device="cpu"
)

text = """Before college the two main things I worked on, outside of school, were writing and programming. I didn't write essays. I wrote what beginning writers were supposed to write then, and probably still are: short stories. My stories were awful. They had hardly any plot, just characters with strong feelings, which I imagined made them deep. The first programs I tried writing were on the IBM 1401 that our school district used for what was then called "data processing." This was in 9th grade, so I was 13 or 14. The school district's 1401 happened to be in the basement of our junior high school, and my friend Rich Draves and I got permission to use it. It was like a mini Bond villain's lair down there, with all these alien-looking machines — CPU, disk drives, printer, card reader — sitting up on a raised floor under bright fluorescent lights."""

for chunk in splitter(text):
  print(chunk)
  print("--")
```

### Sample Output

```
Before college the two main things I worked on, outside of school, were writing and programming. I didn't write essays. I wrote what beginning writers were supposed to write then, and probably still are: short stories.
--
 My stories were awful. They had hardly any plot, just characters with strong feelings, which I imagined made them deep. The first programs I tried writing were on the IBM 1401 that our school district used for what was then called "data processing."
--
 This was in 9th grade, so I was 13 or 14. The school district's 1401 happened to be in the basement of our junior high school, and my friend Rich Draves and I got permission to use it.
--
 It was like a mini Bond villain's lair down there, with all these alien-looking machines — CPU, disk drives, printer, card reader — sitting up on a raised floor under bright fluorescent lights.
--
```

But you can use this model using standart NER pipeline:

```
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline

model_name = "mirth/chonky_modernbert_large_1"

tokenizer = AutoTokenizer.from_pretrained(model_name, model_max_length=1024)

id2label = {
    0: "O",
    1: "separator",
}
label2id = {
    "O": 0,
    "separator": 1,
}

model = AutoModelForTokenClassification.from_pretrained(
    model_name,
    num_labels=2,
    id2label=id2label,
    label2id=label2id,
)


pipe = pipeline("ner", model=model, tokenizer=tokenizer, aggregation_strategy="simple")

text = """Before college the two main things I worked on, outside of school, were writing and programming. I didn't write essays. I wrote what beginning writers were supposed to write then, and probably still are: short stories. My stories were awful. They had hardly any plot, just characters with strong feelings, which I imagined made them deep. The first programs I tried writing were on the IBM 1401 that our school district used for what was then called "data processing." This was in 9th grade, so I was 13 or 14. The school district's 1401 happened to be in the basement of our junior high school, and my friend Rich Draves and I got permission to use it. It was like a mini Bond villain's lair down there, with all these alien-looking machines — CPU, disk drives, printer, card reader — sitting up on a raised floor under bright fluorescent lights."""

pipe(text)

```

### Sample Output

```

[
  {'entity_group': 'separator', 'score': np.float32(0.91590524), 'word': ' stories.', 'start': 209, 'end': 218},
  {'entity_group': 'separator', 'score': np.float32(0.6210419), 'word': ' processing."', 'start': 455, 'end': 468},
  {'entity_group': 'separator', 'score': np.float32(0.7071036), 'word': '.', 'start': 652, 'end': 653}
]
```

## Training Data

The model was trained to split paragraphs from minipile and bookcorpus datasets.


## Metrics

Token based metrics for minipile:

| Metric   | Value |
| -------- | ------|
| F1       | 0.85  |
| Precision| 0.87  |
| Recall   | 0.82  |
| Accuracy | 0.99  |


Token based metrics for bookcorpus:

| Metric   | Value |
| -------- | ------|
| F1       | 0.79  |
| Precision| 0.85  |
| Recall   | 0.74  |
| Accuracy | 0.99  |


## Hardware

Model was fine-tuned on a single H100 for a several hours