root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-27 03:44:35 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

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

Browse Source
This commit is contained in:
klosax 2023-07-31 03:02:00 +02:00 committed by GitHub
parent 4f5b6224be
commit 6b3a7b9f4f
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

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

@ -47,6 +47,12 @@ if len(sys.argv) > 2:
sys.exit(1)
fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
if hparams["architectures"][0] != "LlamaForCausalLM":
print("Model architecture not supported: " + hparams["architectures"][0] )
sys.exit()
with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
hparams = json.load(f)
model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
list_vars = model.state_dict()
@ -61,9 +67,6 @@ 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 = 13