Implementation:Open compass VLMEvalKit SiliconFlowAPI

Field	Value
source	VLMEvalKit
domain	Vision, API_Integration

Overview

SiliconFlowAPI provides a VLMEvalKit API adapter for models hosted on the SiliconFlow inference platform.

Description

SiliconFlowAPI inherits from BaseAPI and communicates with the SiliconFlow chat completions API endpoint. It includes image resizing and encoding utilities with configurable max dimensions, supports an optional reasoning mode that returns both content and reasoning outputs, and uses Bearer token authentication via the SiliconFlow_API_KEY environment variable.

Usage

Use this adapter when evaluating vision-language models (such as DeepSeek-V2.5) hosted on the SiliconFlow platform.

Code Reference

Source: vlmeval/api/siliconflow.py, Lines: L1-277
Import: from vlmeval.api.siliconflow import SiliconFlowAPI

Signature:

class SiliconFlowAPI(BaseAPI):
    def __init__(self, model='deepseek-ai/DeepSeek-V2.5', retry=5,
                 key=None, api_base=API_BASE, verbose=True,
                 system_prompt=None, timeout=60, reasoning=False,
                 **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 = SiliconFlowAPI(model='deepseek-ai/DeepSeek-V2.5')
response = model.generate(message)

Related Pages

Page Connections

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

Principle

Implementation

Heuristic

Environment