General-QA#
Overview#
General-QA is a customizable question answering benchmark for evaluating language models on open-ended text generation tasks. It supports flexible data formats and configurable evaluation metrics.
Task Description#
Task Type: Open-Ended Question Answering
Input: Question (with optional system prompt and conversation history)
Output: Free-form text answer
Flexibility: Supports custom datasets via local files
Key Features#
Flexible input format (query/answer or messages format)
Optional system prompt support
BLEU and Rouge evaluation metrics
Custom dataset support via local file loading
Extensible for various QA use cases
Evaluation Notes#
Default configuration uses 0-shot evaluation
Default metrics: BLEU, Rouge (Rouge-L-R as main score)
Evaluates on test split
See User Guide for dataset format
Properties#
Property |
Value |
|---|---|
Benchmark Name |
|
Dataset ID |
|
Paper |
N/A |
Tags |
|
Metrics |
|
Default Shots |
0-shot |
Evaluation Split |
|
Data Statistics#
Statistics not available.
Sample Example#
Sample example not available.
Prompt Template#
Prompt Template:
请回答问题
{question}
Usage#
Using CLI#
evalscope eval \
--model YOUR_MODEL \
--api-url OPENAI_API_COMPAT_URL \
--api-key EMPTY_TOKEN \
--datasets general_qa \
--limit 10 # Remove this line for formal evaluation
Using Python#
from evalscope import run_task
from evalscope.config import TaskConfig
task_cfg = TaskConfig(
model='YOUR_MODEL',
api_url='OPENAI_API_COMPAT_URL',
api_key='EMPTY_TOKEN',
datasets=['general_qa'],
limit=10, # Remove this line for formal evaluation
)
run_task(task_cfg=task_cfg)