From b7e0ae21d6a237e779c33bc127f77661cc0f30a5 Mon Sep 17 00:00:00 2001 From: jingyaogong Date: Tue, 31 Mar 2026 13:40:16 +0800 Subject: [PATCH] [update] default model --- scripts/convert_model.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/scripts/convert_model.py b/scripts/convert_model.py index 34099e5..daccdb9 100644 --- a/scripts/convert_model.py +++ b/scripts/convert_model.py @@ -126,10 +126,10 @@ def convert_json_to_jinja(json_file_path, output_path): if __name__ == '__main__': - lm_config = MiniMindConfig(hidden_size=768, num_hidden_layers=8, max_seq_len=8192, use_moe=True) + lm_config = MiniMindConfig(hidden_size=768, num_hidden_layers=8, max_seq_len=8192, use_moe=False) # convert torch to transformers torch_path = f"../out/full_sft_{lm_config.hidden_size}{'_moe' if lm_config.use_moe else ''}.pth" - transformers_path = '../minimind-3-moe' + transformers_path = '../minimind-3' convert_torch2transformers(torch_path, transformers_path) # # merge lora