From 0c08c039aa43762c39c71af5b9c5316991d4f212 Mon Sep 17 00:00:00 2001 From: ModelHub XC Date: Fri, 19 Jun 2026 14:59:19 +0800 Subject: [PATCH] =?UTF-8?q?=E5=88=9D=E5=A7=8B=E5=8C=96=E9=A1=B9=E7=9B=AE?= =?UTF-8?q?=EF=BC=8C=E7=94=B1ModelHub=20XC=E7=A4=BE=E5=8C=BA=E6=8F=90?= =?UTF-8?q?=E4=BE=9B=E6=A8=A1=E5=9E=8B?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Model: mlabonne/gemma-3-4b-it-abliterated-GGUF Source: Original Platform --- .gitattributes | 42 +++++++++++++++++++++++++++ README.md | 39 +++++++++++++++++++++++++ gemma-3-4b-it-abliterated.fp16.gguf | 3 ++ gemma-3-4b-it-abliterated.q2_k.gguf | 3 ++ gemma-3-4b-it-abliterated.q3_k_m.gguf | 3 ++ gemma-3-4b-it-abliterated.q4_k_m.gguf | 3 ++ gemma-3-4b-it-abliterated.q5_k_m.gguf | 3 ++ gemma-3-4b-it-abliterated.q6_k.gguf | 3 ++ gemma-3-4b-it-abliterated.q8_0.gguf | 3 ++ 9 files changed, 102 insertions(+) create mode 100644 .gitattributes create mode 100644 README.md create mode 100644 gemma-3-4b-it-abliterated.fp16.gguf create mode 100644 gemma-3-4b-it-abliterated.q2_k.gguf create mode 100644 gemma-3-4b-it-abliterated.q3_k_m.gguf create mode 100644 gemma-3-4b-it-abliterated.q4_k_m.gguf create mode 100644 gemma-3-4b-it-abliterated.q5_k_m.gguf create mode 100644 gemma-3-4b-it-abliterated.q6_k.gguf create mode 100644 gemma-3-4b-it-abliterated.q8_0.gguf diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..4dd47eb --- /dev/null +++ b/.gitattributes @@ -0,0 +1,42 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.fp16.gguf filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.q2_k.gguf filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.q3_k_m.gguf filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.q4_k_m.gguf filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.q5_k_m.gguf filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.q6_k.gguf filter=lfs diff=lfs merge=lfs -text +gemma-3-4b-it-abliterated.q8_0.gguf filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..34fdbaf --- /dev/null +++ b/README.md @@ -0,0 +1,39 @@ +--- +license: gemma +library_name: transformers +pipeline_tag: image-text-to-text +base_model: google/gemma-3-4b-it +tags: +- autoquant +- gguf +--- + +# 💎 Gemma 3 4B IT Abliterated + +![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/WjFfc8hhj20r5XK07Yny9.png) +
Gemma 3 12B Abliterated • Gemma 3 27B Abliterated
+ +This is an uncensored version of [google/gemma-3-4b-it](https://huggingface.co/google/gemma-3-4b-it) created with a new abliteration technique. +See [this article](https://huggingface.co/blog/mlabonne/abliteration) to know more about abliteration. + +I was playing with model weights and noticed that Gemma 3 was much more resilient to abliteration than other models like Qwen 2.5. +I experimented with a few recipes to remove refusals while preserving most of the model capabilities. + +Note that this is fairly experimental, so it might not turn out as well as expected. I saw some garbled text from time to time (e.g., "It' my" instead of "It's my"). + +I recommend using these generation parameters: `temperature=1.0`, `top_k=64`, `top_p=0.95`. + +## ✂️ Layerwise abliteration + +![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/fmwidMV9SvlStlsogCcea.png) + +In the original technique, a refusal direction is computed by comparing the residual streams between target (harmful) and baseline (harmless) samples. + +Here, the model was abliterated by computing a refusal direction based on hidden states (inspired by [Sumandora's repo](https://github.com/Sumandora/remove-refusals-with-transformers/)) for most layers (layer 7 to 29), independently. +This is combined with a refusal weight that follows a symmetric pattern from 0.05 to a peak of 0.55. + +This created a very high acceptance rate (>90%) and still produced coherent outputs. + +## ⚡️ Quantization + +TBD. diff --git a/gemma-3-4b-it-abliterated.fp16.gguf b/gemma-3-4b-it-abliterated.fp16.gguf new file mode 100644 index 0000000..b865762 --- /dev/null +++ b/gemma-3-4b-it-abliterated.fp16.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d718721f1310ab76ee32efed584f62b12001848dd8c6ce1a5817872ae4f2421 +size 7767803776 diff --git a/gemma-3-4b-it-abliterated.q2_k.gguf b/gemma-3-4b-it-abliterated.q2_k.gguf new file mode 100644 index 0000000..f7ffdc9 --- /dev/null +++ b/gemma-3-4b-it-abliterated.q2_k.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b969e816f3f0644b74192655d7d19abf55501f3f8407e229a1546403b995310 +size 1729164416 diff --git a/gemma-3-4b-it-abliterated.q3_k_m.gguf b/gemma-3-4b-it-abliterated.q3_k_m.gguf new file mode 100644 index 0000000..9b4d812 --- /dev/null +++ b/gemma-3-4b-it-abliterated.q3_k_m.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3db3665fee1dcd0f37d17bb624bc099b8260366d7c927b3ac305880642f692a4 +size 2098459776 diff --git a/gemma-3-4b-it-abliterated.q4_k_m.gguf b/gemma-3-4b-it-abliterated.q4_k_m.gguf new file mode 100644 index 0000000..b9b1f4d --- /dev/null +++ b/gemma-3-4b-it-abliterated.q4_k_m.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99dc4abe5d344e1033f09e6a2c1f4e3d39093dd1f007b9d3254d8c2ef94950e1 +size 2489894016 diff --git a/gemma-3-4b-it-abliterated.q5_k_m.gguf b/gemma-3-4b-it-abliterated.q5_k_m.gguf new file mode 100644 index 0000000..e969200 --- /dev/null +++ b/gemma-3-4b-it-abliterated.q5_k_m.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0218d5cbb0ae803e18d6b31243c39261f507208dfaf9364cae9d6f1d710500cb +size 2829698176 diff --git a/gemma-3-4b-it-abliterated.q6_k.gguf b/gemma-3-4b-it-abliterated.q6_k.gguf new file mode 100644 index 0000000..dcb8fdc --- /dev/null +++ b/gemma-3-4b-it-abliterated.q6_k.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87f417fdd329e833a7c74f031359dde04da748d5bc50c67075e5b8b989243327 +size 3190740096 diff --git a/gemma-3-4b-it-abliterated.q8_0.gguf b/gemma-3-4b-it-abliterated.q8_0.gguf new file mode 100644 index 0000000..54adc54 --- /dev/null +++ b/gemma-3-4b-it-abliterated.q8_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2768a0446ecc0e51768f2836821dc75abafd342e234c3c7aa130ffdfe2a93c07 +size 4130402176