Skip to content

convert : force patch_embd weights to F16 or F32 to avoid broken GGUFs #15367

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
CISC merged 2 commits into from
Aug 17, 2025
Merged

convert : force patch_embd weights to F16 or F32 to avoid broken GGUFs #15367

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
fe625c6
force patch_embd weights to f32
CISC Aug 17, 2025
28722d1
use MmprojModel base tensor_force_quant instead
CISC Aug 17, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
force patch_embd weights to f32
  • Loading branch information
@CISC
CISC authored Aug 17, 2025
commit fe625c621d033bace55bd9e09455d4af5803a2ba
2 changes: 1 addition & 1 deletion convert_hf_to_gguf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -302,7 +302,7 @@ def prepare_tensors(self):
data_qtype: gguf.GGMLQuantizationType | bool = self.tensor_force_quant(name, new_name, bid, n_dims)

# Most of the codebase that takes in 1D tensors or norms only handles F32 tensors
if n_dims <= 1 or new_name.endswith("_norm.weight"):
if n_dims <= 1 or new_name.endswith("_norm.weight") or ".patch_embd.weight" in new_name:
data_qtype = gguf.GGMLQuantizationType.F32

# Conditions should closely match those in llama_model_quantize_internal in llama.cpp
Expand Down