InternVL-MoonViT-V2-MiniCPM5-2B-HF

This model combines the MoonViT V2 vision encoder from Kimi K3 with the MiniCPM5-2B language model.

The multimodal projector (MLP) was not trained. Please keep this in mind when evaluating or using the model.

Model architecture

• Vision encoder: MoonViT V2 from Kimi K3
• Language model: MiniCPM5-2B
• Multimodal projector (MLP): not trained

Loading the model

import torch
from transformers import AutoModelForImageTextToText, AutoProcessor

model_id = "PerRing/InternVL-MoonViT-V2-MiniCPM5-2B-HF"

processor = AutoProcessor.from_pretrained(
    model_id,
    trust_remote_code=True,
)

model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
).eval()

Note: This repository uses custom modeling and processing code, so trust_remote_code=True is required.