obscura/minimind
1
0
Fork 0
mirror of https://github.com/jingyaogong/minimind.git synced 2026-10-01 15:27:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

[fix] lora moe

Browse Source
This commit is contained in:
jingyaogong
2026-04-09 16:36:48 +08:00
parent aa3e6affa1
commit 5351424bf0
2 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/convert_model.py
+2 -1
View File
@@ -127,6 +127,7 @@ 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=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'
@@ -134,7 +135,7 @@ if __name__ == '__main__':
# # merge lora
# base_torch_path = f"../out/full_sft_{lm_config.hidden_size}{'_moe' if lm_config.use_moe else ''}.pth"
# lora_path = f"../out/lora_identity_{lm_config.hidden_size}.pth"
# lora_path = f"../out/lora_identity_{lm_config.hidden_size}{'_moe' if lm_config.use_moe else ''}.pth"
# merged_torch_path = f"../out/merge_identity_{lm_config.hidden_size}{'_moe' if lm_config.use_moe else ''}.pth"
# convert_merge_base_lora(base_torch_path, lora_path, merged_torch_path)