Model card
Browse files
README.md
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
library_name: transformers.js
|
| 4 |
+
pipeline_tag: feature-extraction
|
| 5 |
+
base_model: BAAI/bge-small-en-v1.5
|
| 6 |
+
language: en
|
| 7 |
+
tags:
|
| 8 |
+
- decidekit
|
| 9 |
+
- system-one
|
| 10 |
+
- text-classification
|
| 11 |
+
- onnx
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# Decide Small
|
| 15 |
+
|
| 16 |
+
The sentence encoder behind Decide Small, a System One model that picks one of your labels for a piece of text on the user's device. English. 34 MB at 8-bit. [decidekit](https://debuff.jp/decide/) adds the trained head and calibrated confidence.
|
| 17 |
+
|
| 18 |
+
```ts
|
| 19 |
+
import { createDecider } from "decidekit";
|
| 20 |
+
|
| 21 |
+
const d = await createDecider();
|
| 22 |
+
const { choice, confidence } = await d.decide("my card still hasn't arrived", ["card_arrival", "refund"]);
|
| 23 |
+
```
|
| 24 |
+
|
| 25 |
+
- BANKING77, 77 labels, all training data: 90.7%
|
| 26 |
+
- CLINC150, 150 labels, all training data: 94.4%
|
| 27 |
+
- A decision in Chrome (worker, median): 7.7 ms
|
| 28 |
+
|
| 29 |
+
Measured on a Ryzen 7 260 laptop. Scripts and results: [eval/results-bge-small.json](https://github.com/debuffjp/decidekit/blob/main/eval/results-bge-small.json).
|
| 30 |
+
|
| 31 |
+
## Where the weights come from
|
| 32 |
+
|
| 33 |
+
These files are an unmodified copy of [Xenova/bge-small-en-v1.5](https://huggingface.co/Xenova/bge-small-en-v1.5) at revision `ea104da`, the ONNX conversion of [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5), which is MIT licensed. We did not train them. They are published here so decidekit has a stable, pinned source.
|