wylab/llama.cpp
1
0
Fork 1
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-06-09 07:16:44 +02:00
Code Issues Packages Projects Releases Wiki Activity

llama : add Gemma4 MTP (#23398)

Browse Source
This commit is contained in:
Aman Gupta
2026-06-07 20:50:54 +08:00
committed by GitHub
parent 8a091c47ab
commit 04eb4c446d
31 changed files with 644 additions and 142 deletions
Download Patch File Download Diff File Expand all files Collapse all files
conversion/__init__.py
+2
View File
@@ -75,9 +75,11 @@ TEXT_MODEL_MAP: dict[str, str] = {
"Gemma3TextModel": "gemma",
"Gemma3nForCausalLM": "gemma",
"Gemma3nForConditionalGeneration": "gemma",
"Gemma4AssistantForCausalLM": "gemma",
"Gemma4ForConditionalGeneration": "gemma",
"Gemma4ForCausalLM": "gemma",
"Gemma4UnifiedForConditionalGeneration": "gemma",
"Gemma4UnifiedAssistantForCausalLM": "gemma",
"GemmaForCausalLM": "gemma",
"Glm4ForCausalLM": "glm",
"Glm4MoeForCausalLM": "glm",
conversion/gemma.py
+10
View File
@@ -785,6 +785,16 @@ class Gemma4UnifiedModel(Gemma4Model):
self.gguf_writer.add_suppress_tokens(suppress_tokens)
@ModelBase.register("Gemma4AssistantForCausalLM", "Gemma4UnifiedAssistantForCausalLM")
class Gemma4AssistantModel(Gemma4Model):
model_arch = gguf.MODEL_ARCH.GEMMA4_ASSISTANT
def set_gguf_parameters(self):
super().set_gguf_parameters()
self.gguf_writer.add_embedding_length_out(self.hparams["backbone_hidden_size"])
self.gguf_writer.add_nextn_predict_layers(self.block_count)
@ModelBase.register("Gemma4ForConditionalGeneration")
class Gemma4VisionAudioModel(MmprojModel):
has_audio_encoder = True