Implementation:Open compass VLMEvalKit Mantis

Field	Value
source	VLMEvalKit
domain	Vision, Model_Architecture

Overview

VLM adapter for the Mantis model enabling benchmark evaluation in VLMEvalKit.

Description

Mantis inherits from BaseModel and wraps the Mantis model for use within the VLMEvalKit evaluation framework. It initializes the model and tokenizer/processor from a HuggingFace model path (default: TIGER-Lab/Mantis-8B-siglip-llama3) and provides the generate_inner method for inference. Supports multiple model architectures including LLaVA, Fuyu, and IDEFICS2 backends.

Usage

Register in vlmeval/config.py via supported_VLM and invoke through the standard evaluation pipeline.

Code Reference

Source: vlmeval/vlm/mantis.py, Lines: L1-201
Import: from vlmeval.vlm.mantis import Mantis

Signature:

class Mantis(BaseModel):
    INSTALL_REQ = True
    INTERLEAVE = True
    def __init__(self, model_path='TIGER-Lab/Mantis-8B-siglip-llama3', **kwargs): ...
    def generate_inner(self, message, dataset=None): ...

I/O Contract

Direction	Description
Inputs	message — list of dicts with type (text/image) and value; dataset — optional dataset name for custom prompting
Outputs	generate_inner() returns str (model response text)

Usage Examples

from vlmeval.vlm.mantis import Mantis
model = Mantis(model_path='path/to/model')
response = model.generate_inner(message)

Related Pages

Page Connections

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

Principle

Implementation

Heuristic

Environment