Update README.md
Browse files
README.md
CHANGED
|
@@ -57,6 +57,8 @@ configs:
|
|
| 57 |
---
|
| 58 |
# BizFinBench: A Business-Driven Real-World Financial Benchmark for Evaluating LLMs
|
| 59 |
|
|
|
|
|
|
|
| 60 |
## 📌 Overview
|
| 61 |
|
| 62 |
Large language models excel across general tasks, yet judging their reliability in logic‑heavy, precision‑critical domains such as finance, law and healthcare is still difficult. To address this challenge, we propose BizFinBench, the first benchmark grounded in real-world financial applications. BizFinBench consists of 6,781 well-annotated queries in Chinese, covering five dimensions: numerical calculation, reasoning, information extraction, prediction recognition and knowledge‐based question answering, which are mapped to nine fine-grained categories.
|
|
|
|
| 57 |
---
|
| 58 |
# BizFinBench: A Business-Driven Real-World Financial Benchmark for Evaluating LLMs
|
| 59 |
|
| 60 |
+
<a href="https://github.com/HiThink-Research/BizFinBench">Github</a></h3>
|
| 61 |
+
|
| 62 |
## 📌 Overview
|
| 63 |
|
| 64 |
Large language models excel across general tasks, yet judging their reliability in logic‑heavy, precision‑critical domains such as finance, law and healthcare is still difficult. To address this challenge, we propose BizFinBench, the first benchmark grounded in real-world financial applications. BizFinBench consists of 6,781 well-annotated queries in Chinese, covering five dimensions: numerical calculation, reasoning, information extraction, prediction recognition and knowledge‐based question answering, which are mapped to nine fine-grained categories.
|