From 215ec2f0f9cfb4a039bef2e94c01d357410323a7 Mon Sep 17 00:00:00 2001 From: localai-org-maint-bot <306269227+localai-org-maint-bot@users.noreply.github.com> Date: Thu, 8 Oct 2026 04:04:19 +0000 Subject: [PATCH] chore(gallery): add Gemma 4 E4B Heretic variants Add Q4_K_M and Q8_0 builds with the BF16 vision projector for llama.cpp. Pin all three assets to a Hugging Face revision and document installation. Assisted-by: Codex:gpt-6 --- docs/content/features/model-gallery.md | 19 ++++++ gallery/index.yaml | 94 ++++++++++++++++++++++++++ 2 files changed, 113 insertions(+) diff --git a/docs/content/features/model-gallery.md b/docs/content/features/model-gallery.md index 3bdfc6d41..f97ea3a75 100644 --- a/docs/content/features/model-gallery.md +++ b/docs/content/features/model-gallery.md @@ -39,6 +39,25 @@ Both views use the same model selection and store the view, search, filter, and selection in the URL. Installing from Explore does not move you away from the catalog; the entry updates in place when the operation finishes. +## Gemma 4 E4B Heretic + +[Gemma 4 E4B Heretic](https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic) +is an abliterated version of Gemma 4 E4B IT. The gallery includes Q4_K_M and Q8_0 +GGUF builds for llama.cpp. Both include the BF16 vision projector for image input +and use the embedded chat template with a 32,768-token context by default. + +Install with automatic variant selection: + +```bash +local-ai models install gemma-4-e4b-heretic-q4 +``` + +To select the Q8_0 build explicitly: + +```bash +local-ai models install --variant gemma-4-e4b-heretic-q8 gemma-4-e4b-heretic-q4 +``` + ## Cyber-Ornith 1.5 9B Cyber-Ornith 1.5 is a Qwen3.5 fine-tune for security auditing, terminal tasks, and tool use. diff --git a/gallery/index.yaml b/gallery/index.yaml index 17a7cdfa1..7c67898f0 100644 --- a/gallery/index.yaml +++ b/gallery/index.yaml @@ -13951,6 +13951,100 @@ - filename: llama-cpp/mmproj/gemma-4-E4B-it-qat-q4_0-gguf/gemma-4-E4B-it-mmproj.gguf uri: https://huggingface.co/google/gemma-4-E4B-it-qat-q4_0-gguf/resolve/main/gemma-4-E4B-it-mmproj.gguf sha256: 7498a37cb619e55f2fcf87eb931f56e99389ed6d432e4c5c66110694c0d65578 +- name: gemma-4-e4b-heretic-q4 + variants: + - model: gemma-4-e4b-heretic-q8 + url: github:mudler/LocalAI/gallery/virtual.yaml@master + urls: + - https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic + - https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF + license: apache-2.0 + icon: https://ai.google.dev/gemma/images/gemma4_banner.png + description: | + Gemma 4 E4B Heretic is llmfan46's abliterated version of Gemma 4 E4B IT. + This entry uses the Q4_K_M GGUF and BF16 vision projector for chat + and image input, with a 32,768-token context by default. + tags: + - llm + - gguf + - cpu + - gpu + - gemma4 + - multimodal + - reasoning + overrides: + backend: llama-cpp + context_size: 32768 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/gemma-4-e4b-heretic/gemma-4-E4B-it-mmproj-BF16.gguf + options: + - use_jinja:true + parameters: + model: llama-cpp/models/gemma-4-e4b-heretic/gemma-4-E4B-it-ultra-uncensored-heretic-Q4_K_M.gguf + temperature: 1 + top_p: 0.95 + top_k: 64 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/gemma-4-e4b-heretic/gemma-4-E4B-it-ultra-uncensored-heretic-Q4_K_M.gguf + uri: https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/resolve/1465f37b7dbd15e91241ae78ffebecb9f25e15de/gemma-4-E4B-it-ultra-uncensored-heretic-Q4_K_M.gguf + sha256: f9dbbe3bdf396f65aec4e801516a1b868a510d96225417a8aa06763e3c5b97dc + - filename: llama-cpp/mmproj/gemma-4-e4b-heretic/gemma-4-E4B-it-mmproj-BF16.gguf + uri: https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/resolve/1465f37b7dbd15e91241ae78ffebecb9f25e15de/gemma-4-E4B-it-mmproj-BF16.gguf + sha256: ffa64aebf7144bbadd5a1e143b77794d88e76a81cc0168d91de0770eb49e88e0 +- name: gemma-4-e4b-heretic-q8 + url: github:mudler/LocalAI/gallery/virtual.yaml@master + urls: + - https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic + - https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF + license: apache-2.0 + icon: https://ai.google.dev/gemma/images/gemma4_banner.png + description: | + Gemma 4 E4B Heretic is llmfan46's abliterated version of Gemma 4 E4B IT. + This entry uses the Q8_0 GGUF and BF16 vision projector for chat + and image input, with a 32,768-token context by default. + tags: + - llm + - gguf + - cpu + - gpu + - gemma4 + - multimodal + - reasoning + overrides: + backend: llama-cpp + context_size: 32768 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/gemma-4-e4b-heretic/gemma-4-E4B-it-mmproj-BF16.gguf + options: + - use_jinja:true + parameters: + model: llama-cpp/models/gemma-4-e4b-heretic/gemma-4-E4B-it-ultra-uncensored-heretic-Q8_0.gguf + temperature: 1 + top_p: 0.95 + top_k: 64 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/gemma-4-e4b-heretic/gemma-4-E4B-it-ultra-uncensored-heretic-Q8_0.gguf + uri: https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/resolve/1465f37b7dbd15e91241ae78ffebecb9f25e15de/gemma-4-E4B-it-ultra-uncensored-heretic-Q8_0.gguf + sha256: c1543610bcceeba8f51bd7cf38f373d2a002265f8873149aecae8a200b8fe4ba + - filename: llama-cpp/mmproj/gemma-4-e4b-heretic/gemma-4-E4B-it-mmproj-BF16.gguf + uri: https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/resolve/1465f37b7dbd15e91241ae78ffebecb9f25e15de/gemma-4-E4B-it-mmproj-BF16.gguf + sha256: ffa64aebf7144bbadd5a1e143b77794d88e76a81cc0168d91de0770eb49e88e0 - name: gemma-4-e4b-hauhaucs-aggressive-q4 variants: - model: gemma-4-e4b-hauhaucs-aggressive-q5