Gemma 4 12b It Qat Q4 0 Vindex | Sweet Tea Studio
Resources / Gemma 4 12b It Qat Q4 0 Vindex Gemma 4 12b It Qat Q4 0 Vindex google/gemma-4-12B-it-qat-q4 0 converted to larql's vindex format ( larql convert gguf-to-vindex --level inference ), plus the source GGUF and the conversion/verification scripts.
Verified source
Kind Other Base model google/gemma-4-12B-it-qat-q4_0-gguf Version v23e0804deb18cd63cb1ac2a1cb1c827695aed6ee License gemma Publisher @taykso C grade Model source
Kind Other
Base model google/gemma-4-12B-it-qat-q4_0-gguf
Version v23e0804deb18cd63cb1ac2a1cb1c827695aed6ee
License gemma
Parameters 12B
Source Hugging Face --- license: gemma base_model: google/gemma-4-12B-it-qat-q4_0-gguf tags: - gemma4 - larql - vindex - quantized --- # gemma-4-12B-it (QAT q4_0) — larql vindex google/gemma-4-12B-it-qat-q4_0 converted to [larql](https://github.com/chrishayuk/larql)'s vindex format (`larql convert gguf-to-vindex --level inference`), plus the source GGUF and the conversion/verification scripts. Verified against the HF bf16 reference: raw prompt `The capital of France is` -> top-1 `0` (30.83%); exact gemma-4 chat template -> `The` (100.00%). Conversion requires the fixes from larql PR #178; correct KV-cached decode additionally requires the gemma-4 per-layer-epilogue fix (PLE + layer_scalar in kv_dispatch/helpers.rs, 2026-07-07, `code/helpers.rs.FIXED`). Usage: `larql run vindex/ " "` — prompts must use the exact gemma-4 chat template (` user\n... \n model\n thought\n `). Subject to Google's Gemma Terms of Use: https://ai.google.dev/gemma/terms
Sources & provenance
1 active source Source evidence
3 excerpts google/gemma-4-12B-it-qat-q4 0 converted to larql's vindex format ( larql convert gguf-to-vindex --level inference ), plus the source GGUF and the conversion/verification scripts.
license: gemma basemodel: google/gemma-4-12B-it-qat-q40-gguf tags: gemma4 larql vindex quantized
taykso/gemma-4-12b-it-qat-q4_0-vindex
