Update README.md
Browse files
README.md
CHANGED
|
@@ -1,30 +1,30 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
| 4 |
# flan-t5-small-capitalizer
|
| 5 |
|
| 6 |
-
|
| 7 |
-
on the [agentlans/c4-en-lowercased](https://huggingface.co/datasets/agentlans/c4-en-lowercased) dataset.
|
| 8 |
|
| 9 |
-
|
| 10 |
-
|
|
|
|
|
|
|
| 11 |
|
| 12 |
-
## Intended
|
| 13 |
-
|
| 14 |
-
This model is intended for
|
| 15 |
- Capitalizing lowercased text
|
| 16 |
-
-
|
| 17 |
- Text normalization
|
| 18 |
|
| 19 |
-
|
| 20 |
-
- English language only
|
| 21 |
-
- Focused on modern prose found on the Internet
|
| 22 |
-
- May not capitalize titles correctly
|
| 23 |
-
- Not guaranteed to use a capitalization style consistently
|
| 24 |
-
- May have trouble with special terms and abbreviations that need to be capitalized
|
| 25 |
-
- Maximum input and output 1024 tokens
|
| 26 |
-
|
| 27 |
-
Usage example:
|
| 28 |
|
| 29 |
```python
|
| 30 |
import torch
|
|
@@ -41,6 +41,16 @@ print(output[0]["generated_text"])
|
|
| 41 |
# Expected output: Buzzfeed's 360-degree look at the aftermath of California's Valley Fire has been viewed more than 6 million times. Plenty of viewers have been asking how we made it.
|
| 42 |
```
|
| 43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
## Training and evaluation data
|
| 45 |
|
| 46 |
The model was trained on a subset of the C4 dataset's English configuration.
|
|
@@ -93,4 +103,4 @@ Additional training arguments included bf16 precision, automatic batch size find
|
|
| 93 |
- Transformers 4.43.3
|
| 94 |
- PyTorch 2.3.0+cu121
|
| 95 |
- Datasets 3.2.0
|
| 96 |
-
- Tokenizers 0.19.1
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
datasets:
|
| 4 |
+
- agentlans/c4-en-lowercased
|
| 5 |
+
language:
|
| 6 |
+
- en
|
| 7 |
+
base_model:
|
| 8 |
+
- google/flan-t5-small
|
| 9 |
+
tags:
|
| 10 |
+
- capitalization
|
| 11 |
---
|
| 12 |
# flan-t5-small-capitalizer
|
| 13 |
|
| 14 |
+
A specialized fine-tuned version of [google/flan-t5-small](https://huggingface.co/google/flan-t5-small)
|
| 15 |
+
trained on the [agentlans/c4-en-lowercased](https://huggingface.co/datasets/agentlans/c4-en-lowercased) dataset to restore proper capitalization.
|
| 16 |
|
| 17 |
+
## Key Features
|
| 18 |
+
- Restores proper noun and sentence capitalization
|
| 19 |
+
- Builds on FLAN-T5 small's robust NLP capabilities
|
| 20 |
+
- Designed for text normalization tasks
|
| 21 |
|
| 22 |
+
## Intended Uses
|
|
|
|
|
|
|
| 23 |
- Capitalizing lowercased text
|
| 24 |
+
- Sentence and proper noun capitalization
|
| 25 |
- Text normalization
|
| 26 |
|
| 27 |
+
## Usage Example
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
```python
|
| 30 |
import torch
|
|
|
|
| 41 |
# Expected output: Buzzfeed's 360-degree look at the aftermath of California's Valley Fire has been viewed more than 6 million times. Plenty of viewers have been asking how we made it.
|
| 42 |
```
|
| 43 |
|
| 44 |
+
## Limitations
|
| 45 |
+
|
| 46 |
+
- **Language**: English only
|
| 47 |
+
- **Text Type**: Primarily modern prose found on the Internet
|
| 48 |
+
- **Capitalization Issues**:
|
| 49 |
+
- May not capitalize titles correctly
|
| 50 |
+
- Inconsistent capitalization style across texts
|
| 51 |
+
- Difficulty with special terms and abbreviations requiring capitalization
|
| 52 |
+
- **Input/Output Constraint**: Maximum length of 1024 tokens for both input and output
|
| 53 |
+
|
| 54 |
## Training and evaluation data
|
| 55 |
|
| 56 |
The model was trained on a subset of the C4 dataset's English configuration.
|
|
|
|
| 103 |
- Transformers 4.43.3
|
| 104 |
- PyTorch 2.3.0+cu121
|
| 105 |
- Datasets 3.2.0
|
| 106 |
+
- Tokenizers 0.19.1
|