CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-03 09:22:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

convert-gptneox-h5-to-gguf.py : simplify nbytes

Browse Source
This commit is contained in:
klosax
2023-08-12 22:29:35 +02:00
committed by GitHub
parent 4cef57c81a
commit f8218477b3
1 changed files with 1 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
convert-gptneox-h5-to-gguf.py
View File

@@ -211,17 +211,7 @@ for name in list_vars.keys():
# if f16 desired, convert any float32 2-dim weight tensors to float16 # if f16 desired, convert any float32 2-dim weight tensors to float16
data_dtype = np.float16 data_dtype = np.float16
nelements = 1 data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
for i in range(n_dims):
nelements *= data.shape[n_dims - 1 - i]
data_nbytes = 0
if data_dtype == np.float16:
data_nbytes = nelements * 2
elif data_dtype == np.float32:
data_nbytes = nelements * 4
gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes) gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)