yakz1/gemma-4-e4b-question-algerienne-dpo

Gemma 4 E4B Question Algérienne DPO GGUF

GGUF conversion of google/gemma-4-E4B-it after merging the LoRA adapter yakz1/gemma-4-e4b-question-algerienne-dpo-analyse-qualitative-608-2ep-lora.

Artifacts:

• —mmproj-gemma-4-e4b-question-algerienne-dpo-analyse-qualitative-608-2ep.Q8_0.gguf (Q8_0, unchanged base multimodal projector)

The DPO adapter was trained only on text language-model modules under model.language_model. If present, the mmproj artifact is therefore the unchanged base Gemma 4 E4B multimodal projector, included so llama.cpp can load this repository for image/audio-capable serving.

Conversion metadata

{
  "base_model": "google/gemma-4-E4B-it",
  "adapter_model": "yakz1/gemma-4-e4b-question-algerienne-dpo-analyse-qualitative-608-2ep-lora",
  "quantization": "Q8_0",
  "adapter_target_modules": null,
  "mmproj": "mmproj-gemma-4-e4b-question-algerienne-dpo-analyse-qualitative-608-2ep.Q8_0.gguf",
  "mmproj_note": "The projector is exported from the base Gemma 4 E4B model because the DPO adapter trains language-model modules only."
}