Update README.md
Browse files
README.md
CHANGED
|
@@ -1,37 +1,63 @@
|
|
| 1 |
|
| 2 |
---
|
| 3 |
tags:
|
| 4 |
-
-
|
|
|
|
|
|
|
| 5 |
- text-classification
|
| 6 |
widget:
|
| 7 |
-
- text: "
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
datasets:
|
| 9 |
- FinBERT_market_based/autotrain-data
|
| 10 |
---
|
| 11 |
|
| 12 |
-
# Model
|
| 13 |
|
| 14 |
-
|
|
|
|
| 15 |
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
f1_macro: 0.49749240436690023
|
| 20 |
-
|
| 21 |
-
f1_micro: 0.5627105467737756
|
| 22 |
|
| 23 |
-
|
|
|
|
|
|
|
| 24 |
|
| 25 |
-
|
|
|
|
|
|
|
| 26 |
|
| 27 |
-
precision_micro: 0.5627105467737756
|
| 28 |
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
recall_macro: 0.517542664344306
|
| 32 |
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
|
| 35 |
-
recall_weighted: 0.5627105467737756
|
| 36 |
|
| 37 |
-
accuracy: 0.5627105467737756
|
|
|
|
| 1 |
|
| 2 |
---
|
| 3 |
tags:
|
| 4 |
+
- finance
|
| 5 |
+
- finbert
|
| 6 |
+
- market
|
| 7 |
- text-classification
|
| 8 |
widget:
|
| 9 |
+
- text: "Asian Stocks Set to Decline Amidst Growth Worries"
|
| 10 |
+
output:
|
| 11 |
+
- label: POSITIVE
|
| 12 |
+
score: 0.14
|
| 13 |
+
- label: INDECISIVE
|
| 14 |
+
score: 0.25
|
| 15 |
+
- label: NEGATIVE
|
| 16 |
+
score: 0.61
|
| 17 |
+
- text: "High inflation expectations becoming part of the American consumers behavioral norm"
|
| 18 |
+
output:
|
| 19 |
+
- label: POSITIVE
|
| 20 |
+
score: 0.49
|
| 21 |
+
- label: INDECISIVE
|
| 22 |
+
score: 0.30
|
| 23 |
+
- label: NEGATIVE
|
| 24 |
+
score: 0.21
|
| 25 |
datasets:
|
| 26 |
- FinBERT_market_based/autotrain-data
|
| 27 |
---
|
| 28 |
|
| 29 |
+
# Model Card for Finetuned FinBERT on Market-Based Facts
|
| 30 |
|
| 31 |
+
Our FinBERT model, finetuned on impactful news headlines about global equity markets, has shown significant performance improvements over standard models.
|
| 32 |
+
Its training on real-world market impact rather than subjective financial expert opinions sets a new standard for unbiased financial sentiment analysis. 📈
|
| 33 |
|
| 34 |
+
**Outperforms FinBERT**
|
| 35 |
+
- 🎯 +25% precision
|
| 36 |
+
- 🚀 +18% recall
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
+
**Outperforms DistilRoBERTa finetuned for finance**
|
| 39 |
+
- 🎯 +22% precision
|
| 40 |
+
- 🚀 +15% recall
|
| 41 |
|
| 42 |
+
**Outperforms GPT-4 zero-shot learning**
|
| 43 |
+
- 🎯 +15% precision
|
| 44 |
+
- 🚀 +8.2% recall
|
| 45 |
|
|
|
|
| 46 |
|
| 47 |
+
## Validation Metrics
|
|
|
|
|
|
|
| 48 |
|
| 49 |
+
| Metric | Value |
|
| 50 |
+
|--------------------|-----------------------|
|
| 51 |
+
| loss | 0.9176467061042786 |
|
| 52 |
+
| f1_macro | 0.49749240436690023 |
|
| 53 |
+
| f1_micro | 0.5627105467737756 |
|
| 54 |
+
| f1_weighted | 0.5279720746084178 |
|
| 55 |
+
| precision_macro | 0.5386355574899088 |
|
| 56 |
+
| precision_micro | 0.5627105467737756 |
|
| 57 |
+
| precision_weighted | 0.5462149036191247 |
|
| 58 |
+
| recall_macro | 0.517542664344306 |
|
| 59 |
+
| recall_micro | 0.5627105467737756 |
|
| 60 |
+
| recall_weighted | 0.5627105467737756 |
|
| 61 |
+
| accuracy | 0.5627105467737756 |
|
| 62 |
|
|
|
|
| 63 |
|
|
|