Downloads · 30 days
891
53% of all-time downloads
xCloudinfo/Gemma-4-31B-xVITA-zhTW-abliterated-GGUF
Gemma-4-31B-xVITA-zhTW-abliterated-GGUF is a image-text-to-text model from xCloudinfo. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
云碩科技 · xCloudinfo · 系列:xVITA · 消除拒絕方向 · Abliterated
Downloads · 30 days
891
53% of all-time downloads
All-time downloads
1.7K
Public
Repo size
127 GB
Likes
0
Public
Click a slice to open those files.
.gguf127 GB · 100%
From the Hugging Face model README
云碩科技 · xCloudinfo · 系列:xVITA · 消除拒絕方向 · Abliterated
使用前提 本模型以合法、且在使用者獲授權範圍內使用為前提製作與發布。經abliteration移除模型的 拒絕方向,供資安研究、紅隊評估、內容審核訓練、文件與資料分析等正當工作使用。 使用者須在中華民國法律及所適用法規範圍內使用,並對一切用途與後果負完全責任; 不得用於任何違法或造成他人危害之用途。本模型不設安全防護,請自行評估部署風險。
xCloudinfo/Gemma-4-31B-xVITA-zhTW-GGUF 的消除拒絕方向版GGUF,含視覺投影器(mmproj),可圖文對話。abliteration只對語言端殘差流的權重做正交化,視覺塔完全未更動,與一般版共用同一份mmproj。
六階量化 Q8_0 / Q6_K / Q5_K_M / Q4_K_M / IQ4_XS / IQ2_M + imatrix + 視覺投影器(mmproj)。
llama-server -m model-Q4_K_M.gguf \
--mmproj mmproj-Gemma-4-31B-xVITA-zhTW-f16.gguf -c 4096 -ngl 99
純文字可省略
--mmproj。Q4_K_M部署首選;Q8_0/Q6_K追品質;IQ4_XS/IQ2_M省資源。
google/gemma-4-31B-it:Apache-2.0 + Google Gemma 4 License(含Prohibited Use Policy)。由 云碩科技xCloudinfo於自有AI算力資源池製作。
云碩科技(xCloudinfo Corp. Limited)專注於地端私有化人工智慧解決方案。xVITA是云碩自研模型的品牌名稱。
其他模型請見 https://hf.co/xCloudinfo。
xCloudinfo (云碩科技) · Series: xVITA · refusal-direction removed · Abliterated
Usage premise This model is made and released on the premise of lawful use within the scope the user is authorized for. Abliteration has removed the model's refusal direction, for legitimate work such as security research, red-team evaluation, content-moderation training, and document/data analysis. Users must use it within the scope of the laws and applicable regulations of the Republic of China (Taiwan), and bear full responsibility for all uses and consequences; it must not be used for any illegal purpose or to cause harm to others. This model has no safety guardrails; assess your deployment risk yourself.
The refusal-direction-removed GGUF of xCloudinfo/Gemma-4-31B-xVITA-zhTW-GGUF, including the vision projector (mmproj), supporting image-text conversation. Abliteration only orthogonalizes the weights on the language-side residual stream; the vision tower is completely untouched, and it shares the same mmproj as the regular version.
Six quantization tiers Q8_0 / Q6_K / Q5_K_M / Q4_K_M / IQ4_XS / IQ2_M + imatrix + vision projector (mmproj).
llama-server -m model-Q4_K_M.gguf \
--mmproj mmproj-Gemma-4-31B-xVITA-zhTW-f16.gguf -c 4096 -ngl 99
For text only, you can omit
--mmproj. Q4_K_M is the deployment first choice; Q8_0/Q6_K for quality; IQ4_XS/IQ2_M to save resources.
google/gemma-4-31B-it: Apache-2.0 + Google Gemma 4 License (including the Prohibited Use Policy).Made by xCloudinfo (云碩科技) on its own AI compute pool.
xCloudinfo Corp. Limited (云碩科技) builds on-premises, private AI solutions. xVITA is xCloudinfo's in-house model brand.
More models: https://hf.co/xCloudinfo