tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-28 20:25:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix a bug in the rope calculation

Browse Source
This commit is contained in:
Georgi Gerganov
2023-03-10 23:46:39 +02:00
parent 18ebda34d6
commit 70bc0b8b15
4 changed files with 92 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
convert-pth-to-ggml.py
View File

@ -73,7 +73,7 @@ fout.write(struct.pack("i", hparams["dim"]))
fout.write(struct.pack("i", hparams["multiple_of"]))
fout.write(struct.pack("i", hparams["n_heads"]))
fout.write(struct.pack("i", hparams["n_layers"]))
fout.write(struct.pack("i", 64)) # rot
fout.write(struct.pack("i", hparams["dim"] // hparams["n_heads"])) # rot (obsolete)
fout.write(struct.pack("i", ftype))
# Is this correct??