File size: 1,532 Bytes
b296c19 d904a5f 9acc3cb d904a5f b296c19 27fac14 0413bc3 27fac14 0413bc3 0638552 0413bc3 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 |
---
datasets:
- 0xMaka/trading-candles-subset-sc-format
language:
- en
metrics:
- accuracy
- f1
widget:
- text: 'identify candle: 17284.58,17264.41,17284.58,17264.41'
example_title: Bear
- text: 'identify candle: open: 17343.43, close: 17625.18, high: 17804.68, low: 17322.15'
example_title: Bull
license: gpl
---
# Based Bert for sequence classification
This model is a POC and shouldn't be used for any production task.
## Model description
Based Bert SC is a text classification bot for binary classification of a trading candles opening and closing prices.
## Uses and limitations
This model can reliably return the bullish or bearish status of a candle given the opening, closing, high and low, in a format shown.
It will have trouble if the order of the numbers change (even if tags are included).
### How to use
You can use this model directly with a pipeline
```python
>>> from transformers import pipeline
>>> pipe = pipeline("text-classification", model="0xMaka/based-bert-sc")
>>> text = "identify candle: open: 21788.19, close: 21900, high: 21965.23, low: 21788.19"
>>> pipe(text)
[{'label': 'Bullish', 'score': 0.9999682903289795}]
```
## Finetuning
For parameters: https://github.com/0xMaka/based-bert-sc/blob/main/trainer.py
This model was fine tuned on an RTX-3060-Mobile
```
// BUS_WIDTH = 192
// CLOCK_RATE = 1750
// DDR_MULTI = 8 // DDR6
// BWTheoretical = (((CLOCK_RATE * (10 ** 6)) * (BUS_WIDTH/8)) * DDR_MULI) / (10 ** 9)
// BWTheoretical == 336 GB/s
```
Self-measured effective (GB/s): 316.280736
|