Skip to content

Support gemma 12B with quant weights. #11696

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
comfyanonymous merged 1 commit into from
Jan 7, 2026
Merged

Support gemma 12B with quant weights. #11696

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
12 changes: 6 additions & 6 deletions comfy/text_encoders/lt.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,10 +36,10 @@ def __init__(self, embedding_directory=None, tokenizer_data={}):

class Gemma3_12BModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="all", layer_idx=None, dtype=None, attention_mask=True, model_options={}):
llama_scaled_fp8 = model_options.get("gemma_scaled_fp8", None)
if llama_scaled_fp8 is not None:
llama_quantization_metadata = model_options.get("llama_quantization_metadata", None)
if llama_quantization_metadata is not None:
model_options = model_options.copy()
model_options["scaled_fp8"] = llama_scaled_fp8
model_options["quantization_metadata"] = llama_quantization_metadata

super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config={}, dtype=dtype, special_tokens={"start": 2, "pad": 0}, layer_norm_hidden_state=False, model_class=comfy.text_encoders.llama.Gemma3_12B, enable_attention_masks=attention_mask, return_attention_masks=attention_mask, model_options=model_options)

Expand Down Expand Up @@ -119,12 +119,12 @@ def load_sd(self, sd):
return self.load_state_dict(sdo, strict=False)


def ltxav_te(dtype_llama=None, llama_scaled_fp8=None):
def ltxav_te(dtype_llama=None, llama_quantization_metadata=None):
class LTXAVTEModel_(LTXAVTEModel):
def __init__(self, device="cpu", dtype=None, model_options={}):
if llama_scaled_fp8 is not None and "llama_scaled_fp8" not in model_options:
if llama_quantization_metadata is not None:
model_options = model_options.copy()
model_options["llama_scaled_fp8"] = llama_scaled_fp8
model_options["llama_quantization_metadata"] = llama_quantization_metadata
if dtype_llama is not None:
dtype = dtype_llama
super().__init__(dtype_llama=dtype_llama, device=device, dtype=dtype, model_options=model_options)
Expand Down