Implementation:Open compass VLMEvalKit CWWrapper

Field	Value
source	VLMEvalKit
domain	Vision, API_Integration

Overview

CWWrapper provides a VLMEvalKit API adapter for CloudWalk's Congrong vision-language models.

Description

CWWrapper inherits from BaseAPI and communicates with the CloudWalk API endpoint. It supports configurable image detail levels (high/low), base64 image encoding, and standard OpenAI-compatible chat completion message formatting. Authentication is handled through the CW_API_KEY and CW_API_BASE environment variables.

Usage

Use this adapter when evaluating CloudWalk Congrong vision models (such as cw-congrong-v2.0) through the CloudWalk API.

Code Reference

Source: vlmeval/api/cloudwalk.py, Lines: L1-103
Import: from vlmeval.api.cloudwalk import CWWrapper

Signature:

class CWWrapper(BaseAPI):
    def __init__(self, model='cw-congrong-v2.0', retry=10, key=None,
                 verbose=True, system_prompt=None, temperature=0,
                 timeout=600, api_base='', max_tokens=2048,
                 img_detail='low', **kwargs): ...
    def generate_inner(self, inputs, **kwargs): ...

I/O Contract

Direction	Description
Inputs	message — text/image/video content list; model-specific params via kwargs
Outputs	generate() returns str prediction; generate_inner() returns (int, str, str) tuple

Usage Examples

# Example instantiation
model = CWWrapper(model='cw-congrong-v2.0')
response = model.generate(message)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment