Public/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-19 17:24:57 +02:00
Code Issues Packages Projects Releases Wiki Activity

add --no-nextn

Browse Source
This commit is contained in:
Aman Gupta
2026-08-02 22:10:55 +08:00
parent 1445ea2993
commit 05491879eb
1 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
convert_hf_to_gguf.py
+4 -4
View File
@@ -122,8 +122,8 @@ def parse_args() -> argparse.Namespace:
help="Export only the multi-token prediction (MTP) head as a separate GGUF, suitable for use as a speculative draft. An 'mtp-' prefix will be added to the output file name.",
)
parser.add_argument(
"--no-mtp", action="store_true",
help="Exclude the multi-token prediction (MTP) head from the converted GGUF. Pair with --mtp on a second run to publish trunk and MTP as two files. Note: the split form duplicates embeddings, but even though the bundled default is more space-efficient overall, this allows differing quantization which may be more performant.",
"--no-nextn", "--no-mtp", dest="no_mtp", action="store_true",
help="Exclude NextN speculative draft tensors from the converted GGUF. Pair with --mtp or --dspark on a second run to publish target and draft as two files.",
)
parser.add_argument(
"--dspark", action="store_true",
@@ -259,7 +259,7 @@ def main() -> None:
model_class = MistralModel
if sum((args.mtp, args.no_mtp, args.dspark)) > 1:
logger.error("--mtp, --no-mtp, and --dspark are mutually exclusive")
logger.error("--mtp, --no-nextn, and --dspark are mutually exclusive")
sys.exit(1)
if args.dspark:
@@ -271,7 +271,7 @@ def main() -> None:
if args.mtp or args.no_mtp:
if not model_class.supports_mtp_export:
logger.error("--mtp / --no-mtp are not supported for %s", model_architecture)
logger.error("--mtp / --no-nextn are not supported for %s", model_architecture)
sys.exit(1)
if args.no_mtp:
model_class.no_mtp = True