7d16743184
enable pretrain.
b655153ec7
Merge pull request #2 from Yiqing-Zhou/fix-custom-models
9f8f9ecc89
[fix] fix genarate with custom models does not go to custom_models
e8d543558c
Merge pull request #1 from Yiqing-Zhou/custom-model-configs
fcb93e52c4
[feature] custom model configs
b7c27af6c8
Add research_token to dump token relationship in attention layer0.
185278f3a9
Update research_attention dump without sum.
a4fafd460f
Refine model of qwen.
11af10e710
Refine research_attention and forward model.