Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/en/benchmarks/bfcl_v3.md
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,7 @@ BFCL (Berkeley Function Calling Leaderboard) v3 is the first comprehensive and e

## Evaluation Notes

- Requires `pip install bfcl-eval==2025.10.27.1` before evaluation
- Requires `pip install 'evalscope[bfcl]'` before evaluation
- Set `is_fc_model=True` for models with native function calling support
- `underscore_to_dot` parameter controls function name formatting
- See the [usage documentation](https://evalscope.readthedocs.io/en/latest/third_party/bfcl_v3.html) for detailed setup
Expand Down
2 changes: 1 addition & 1 deletion docs/en/benchmarks/bfcl_v4.md
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ BFCL-v4 (Berkeley Function-Calling Leaderboard V4) is a comprehensive benchmark

## Evaluation Notes

- **Installation Required**: `pip install bfcl-eval==2025.10.27.1`
- **Installation Required**: `pip install 'evalscope[bfcl]'`
- Primary metric: **Accuracy** per category
- Configure `is_fc_model` based on model capabilities
- Optional: Set `SERPAPI_API_KEY` for web search tasks
Expand Down
3 changes: 1 addition & 2 deletions docs/en/third_party/bfcl_v3.md
Original file line number Diff line number Diff line change
Expand Up @@ -39,8 +39,7 @@ This benchmark consists of 17 tasks, covering a wide range of function-calling s
Before running the evaluation, you need to install the following dependencies:

```bash
pip install evalscope # Install evalscope
pip install bfcl-eval==2025.10.27.1 # Install bfcl-eval
pip install 'evalscope[bfcl]' # Install evalscope with bfcl-eval and its compatible dependencies
```

## Usage
Expand Down
3 changes: 1 addition & 2 deletions docs/en/third_party/bfcl_v4.md
Original file line number Diff line number Diff line change
Expand Up @@ -62,8 +62,7 @@ Before running the evaluation, you need to install the following dependencies:


```bash
pip install evalscope
pip install bfcl-eval==2025.10.27.1
pip install 'evalscope[bfcl]'
```

## Usage
Expand Down
2 changes: 1 addition & 1 deletion docs/zh/benchmarks/bfcl_v3.md
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,7 @@ BFCL(Berkeley Function Calling Leaderboard)v3 是首个全面且可执行的

## 评估说明

- 评估前需安装 `pip install bfcl-eval==2025.10.27.1`
- 评估前需安装 `pip install 'evalscope[bfcl]'`
- 对于原生支持函数调用的模型,请设置 `is_fc_model=True`
- `underscore_to_dot` 参数控制函数名称的格式化方式
- 详细设置请参阅 [使用文档](https://evalscope.readthedocs.io/zh-cn/latest/third_party/bfcl_v3.html)
Expand Down
2 changes: 1 addition & 1 deletion docs/zh/benchmarks/bfcl_v4.md
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ BFCL-v4(Berkeley Function-Calling Leaderboard V4)是一个全面的基准测

## 评估说明

- **安装要求**:`pip install bfcl-eval==2025.10.27.1`
- **安装要求**:`pip install 'evalscope[bfcl]'`
- 主要指标:各类别下的 **准确率(Accuracy)**
- 根据模型能力设置 `is_fc_model` 参数
- 可选:配置 `SERPAPI_API_KEY` 以启用网络搜索任务
Expand Down
3 changes: 1 addition & 2 deletions docs/zh/third_party/bfcl_v3.md
Original file line number Diff line number Diff line change
Expand Up @@ -39,8 +39,7 @@ Berkeley Function Calling Leaderboard (BFCL) 专门用于评估大型语言模
在运行评测之前需要安装以下依赖:

```bash
pip install evalscope # 安装 evalscope
pip install bfcl-eval==2025.10.27.1 # 安装 bfcl-eval
pip install 'evalscope[bfcl]' # 安装 evalscope 及 bfcl-eval 和与之兼容的依赖
```

## 使用方法
Expand Down
3 changes: 1 addition & 2 deletions docs/zh/third_party/bfcl_v4.md
Original file line number Diff line number Diff line change
Expand Up @@ -61,8 +61,7 @@ SUBSET_LIST = [
在运行评测之前需要安装以下依赖:

```bash
pip install evalscope
pip install bfcl-eval==2025.10.27.1
pip install 'evalscope[bfcl]'
```

## 使用方法
Expand Down
12 changes: 6 additions & 6 deletions evalscope/benchmarks/_meta/bfcl_v3.json

Large diffs are not rendered by default.

12 changes: 6 additions & 6 deletions evalscope/benchmarks/_meta/bfcl_v4.json

Large diffs are not rendered by default.

2 changes: 2 additions & 0 deletions evalscope/benchmarks/bfcl/requirements.txt
Original file line number Diff line number Diff line change
@@ -1 +1,3 @@
bfcl-eval==2025.10.27.1
# bfcl-eval pins numpy==1.26.4; pyarrow 26 needs NumPy 2 at import time without declaring it.
pyarrow<26
2 changes: 1 addition & 1 deletion evalscope/benchmarks/bfcl/v3/bfcl_v3_adapter.py
Original file line number Diff line number Diff line change
Expand Up @@ -71,7 +71,7 @@

## Evaluation Notes

- Requires `pip install bfcl-eval==2025.10.27.1` before evaluation
- Requires `pip install 'evalscope[bfcl]'` before evaluation
- Set `is_fc_model=True` for models with native function calling support
- `underscore_to_dot` parameter controls function name formatting
- See the [usage documentation](https://evalscope.readthedocs.io/en/latest/third_party/bfcl_v3.html) for detailed setup
Expand Down
2 changes: 1 addition & 1 deletion evalscope/benchmarks/bfcl/v4/bfcl_v4_adapter.py
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@ def _normalize_openai_base_url(api_url: Optional[str]) -> str:

## Evaluation Notes

- **Installation Required**: `pip install bfcl-eval==2025.10.27.1`
- **Installation Required**: `pip install 'evalscope[bfcl]'`
- Primary metric: **Accuracy** per category
- Configure `is_fc_model` based on model capabilities
- Optional: Set `SERPAPI_API_KEY` for web search tasks
Expand Down
Loading