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

Update convert-gptneox-h5-to-gguf.py

Browse Source
This commit is contained in:
klosax
2023-07-31 03:00:20 +02:00
committed by GitHub
parent 2a0914673c
commit 4f5b6224be
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
convert-gptneox-h5-to-gguf.py
View File

@@ -37,6 +37,12 @@ if len(sys.argv) > 2:
sys.exit(1)
fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
hparams = json.load(f)
if hparams["architectures"][0] != "GPTNeoXForCausalLM":
print("Model architecture not supported: " + hparams["architectures"][0] )
sys.exit()
model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
list_vars = model.state_dict()
@@ -51,9 +57,6 @@ for name in list_vars.keys():
gguf_writer = gguf.GGUFWriter.open(fname_out)
with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
hparams = json.load(f)
# This must be changed when adding/deleting kv
kv_count = 14