happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

convert : better mtp check and fix return [no ci] (#20419)

This commit is contained in:
Sigbjørn Skjæret 2026-03-12 10:04:20 +01:00 committed by GitHub
parent 40c550d4f6
commit c3e3f9e533
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
convert_hf_to_gguf.py
View File

@ -10092,9 +10092,9 @@ class NemotronHModel(GraniteHybridModel):
# Skip Multi-Token Prediction (MTP) tensors. These are used for # Skip Multi-Token Prediction (MTP) tensors. These are used for
# for speculative decoding but we don't include them in this model # for speculative decoding but we don't include them in this model
# conversion. See https://github.com/ggml-org/llama.cpp/pull/18886 # conversion. See https://github.com/ggml-org/llama.cpp/pull/18886
if "mtp" in name: if name.startswith("mtp."):
logger.info(f"gguf: Skipping MTP (Speculative) layer: {name}") logger.info(f"gguf: Skipping MTP (Speculative) layer: {name}")
return [] return
if name.endswith("mixer.gate.e_score_correction_bias"): if name.endswith("mixer.gate.e_score_correction_bias"):
new_name = name.replace("e_score_correction_bias", "e_score_correction.bias") new_name = name.replace("e_score_correction_bias", "e_score_correction.bias")