Add model card and documentation
Browse files
README.md
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: en
|
| 3 |
+
license: mit
|
| 4 |
+
tags:
|
| 5 |
+
- scibert
|
| 6 |
+
- classification
|
| 7 |
+
- technical-papers
|
| 8 |
+
- machine-learning
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# AEGIS SciBERT Technical Classifier
|
| 12 |
+
|
| 13 |
+
## Model Description
|
| 14 |
+
Fine-tuned SciBERT model for classifying technical papers into research categories.
|
| 15 |
+
|
| 16 |
+
## Training Details
|
| 17 |
+
- **Base Model**: allenai/scibert_scivocab_uncased
|
| 18 |
+
- **Training Samples**: 500
|
| 19 |
+
- **Number of Classes**: 6
|
| 20 |
+
- **Classes**: cs.AI, cs.LG, quant-ph, cs.NE, stat.ML, cs.CV
|
| 21 |
+
- **Validation Accuracy**: 1.0000
|
| 22 |
+
|
| 23 |
+
## Usage
|
| 24 |
+
```python
|
| 25 |
+
from transformers import AutoModelForSequenceClassification, AutoTokenizer
|
| 26 |
+
|
| 27 |
+
model = AutoModelForSequenceClassification.from_pretrained("gsstec/aegis-scibert-technical")
|
| 28 |
+
tokenizer = AutoTokenizer.from_pretrained("gsstec/aegis-scibert-technical")
|
| 29 |
+
|
| 30 |
+
# Example inference
|
| 31 |
+
text = "Quantum computing algorithms for machine learning"
|
| 32 |
+
inputs = tokenizer(text, return_tensors="pt")
|
| 33 |
+
outputs = model(**inputs)
|
| 34 |
+
predictions = torch.softmax(outputs.logits, dim=-1)
|
| 35 |
+
```
|
| 36 |
+
|
| 37 |
+
## Classes
|
| 38 |
+
- `cs.AI`: Class 0
|
| 39 |
+
- `cs.LG`: Class 1
|
| 40 |
+
- `quant-ph`: Class 2
|
| 41 |
+
- `cs.NE`: Class 3
|
| 42 |
+
- `stat.ML`: Class 4
|
| 43 |
+
- `cs.CV`: Class 5
|