OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Mcore Bridge

Mcore Bridge is a Python library of large language and multimodal model definitions built for the Megatron-Core ecosystem, with functions for creating models and loading or saving weights.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Mcore Bridge — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when a model architecture appears in its supported-model list and you need model definitions and weight-conversion functions for a Megatron-Core workflow.

Who it is for

  • Teams integrating supported language or multimodal model architectures into Megatron-Core.
  • Developers adapting Transformers-style model workflows to Megatron-Core.
  • Practitioners building full-parameter or LoRA training workflows with Megatron-Core.

Documented uses

  • Create a Megatron-Core model from a supported model configuration.
  • Load model weights into the Megatron representation and save or export weights.
  • Use supported model definitions in training workflows involving Megatron-Core parallelism.
Before you choose

Requirements & limits

The project is published as an open-source Python repository under the Apache-2.0 license. The README documents pip and source-checkout installation.

  • Confirm the exact model type in the repository’s supported-model list.
  • Check runtime versions: Python 3.10 or newer, PyTorch 2.0 or newer, and Megatron-Core 0.16 to below 0.20.
  • Account for the listed Transformers, ModelScope, and Transformer Engine requirements and the repository’s CUDA guidance.
  • The example’s model download, distributed process setup, and training workflow use additional libraries and infrastructure beyond the bridge itself.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install Mcore Bridge from the repository

Install the library from a cloned source checkout.

Before you start

  • Python 3.10 or newer; Python 3.12 is recommended.
  • PyTorch 2.0 or newer and Megatron-Core 0.16 to below 0.20.
  • The repository lists Transformers, ModelScope, and Transformer Engine as runtime dependencies; review its full requirements and CUDA guidance for your environment.

Steps

  1. Clone the Mcore Bridge repository.

  2. Change into the mcore-bridge directory.

  3. Run the command below.

    Documented command
    pip install -e .

Expected outcome

The package is installed in editable mode from the local checkout.

Guide limitations

  • This covers installation only; it does not configure a distributed runtime or run model training.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
modelscope/mcore-bridge · Official homepage title · retained in its source language
How we review sourcesSuggest a correction