Ant-Multi-Modal-Framework Inquiry about Model Size and Plans for Open Sourcing Larger Models

Hello, I noticed that the model in Modelscope is m2_encoder_0.2B.ckpt. Is this the 0.4B parameter model mentioned in the paper? Will there be larger models open sourced in the future?

Mar 05 '24 08:03 209ye

Indeed, the file 'm2_encoder_0.2B.ckpt' corresponds to the model with 0.4 billion parameters referenced in our publication. Apologies for the discrepancy in naming
Certainly, we are committed to making our research as accessible as possible. In line with this commitment, we plan to open source the 1 billion and 10 billion parameter models within the upcoming months.

Mar 06 '24 12:03 LandyGuo

fixe the issue in this commit: https://github.com/alipay/Ant-Multi-Modal-Framework/commit/b38e199ecca0989c84fcba49f44823a1024a14a7, and naming in modelscope: https://www.modelscope.cn/models/M2Cognition/M2-Encoder/files

Mar 07 '24 06:03 LandyGuo

Thanks for the detailed response and sharing your future plans. I'm excited to hear about your team's intention to open source models at the 1 billion and 10 billion parameter scale. Appreciate your contributions through open sourcing valuable AI resources.

Mar 07 '24 06:03 209ye

We have released 1B and 10B modes in this PR: #14

May 27 '24 06:05 LandyGuo