Downloads · 30 days
28
34% of all-time downloads
xCloudinfo/Gemma-4-26B-A4B-Maritime-zhTW
Gemma-4-26B-A4B-Maritime-zhTW 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. It is set up for transformers. The card lists the license as apache-2.0.
云碩科技 · xCloudinfo · 系列:海事航海專業 · Maritime
Downloads · 30 days
28
34% of all-time downloads
All-time downloads
83
Public
Parameters
25.8B
51.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors51.6 GB · 100%
From the Hugging Face model README
云碩科技 · xCloudinfo · 系列:海事航海專業 · Maritime
繁體中文(台灣)+英文的海事航海專業多模態模型。以自家 Gemma-4-26B-A4B-TAIDE-zhTW(26B總 / A4B≈4B活躍 / MoE / text+vision、台灣繁中底座)為基礎,再做一輪雙語海事LoRA SFT,涵蓋避碰規則(COLREGs)、船舶操縱、航海導航、氣象航路、GMDSS通信、海事公約(SOLAS/MLC)、安全與應急等領域。保留原生影像辨識能力(LoRA僅作用於語言層,vision tower未更動)。
from transformers import AutoModelForImageTextToText, AutoProcessor
import torch
mid = "xCloudinfo/Gemma-4-26B-A4B-Maritime-zhTW"
proc = AutoProcessor.from_pretrained(mid)
model = AutoModelForImageTextToText.from_pretrained(mid, dtype="auto", device_map="auto")
msgs = [{"role":"user","content":[{"type":"text","text":"請說明 COLREGs 中直航船與讓路船的避讓義務。"}]}]
ids = proc.apply_chat_template(msgs, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt").to(model.device)
out = model.generate(**ids, max_new_tokens=512)
print(proc.decode(out[0][ids["input_ids"].shape[1]:], skip_special_tokens=True))
GGUF(llama.cpp / Ollama,含mmproj視覺)版本見 …-Maritime-zhTW-GGUF。
xCloudinfo/Gemma-4-26B-A4B-TAIDE-zhTW(底座google gemma-4 = Apache-2.0;含TAIDE衍生,使用須遵守 TAIDE模型授權條款 並保留本聲明)。由 云碩科技xCloudinfo於自有AI算力資源池製作;資料留在本地、流程可重現。
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: Maritime / nautical specialty
A maritime / nautical specialty multimodal model in Traditional Chinese (Taiwan) + English. Built on the in-house Gemma-4-26B-A4B-TAIDE-zhTW (26B total / A4B ≈ 4B active / MoE / text+vision, a Taiwanese-Chinese base), with one further round of bilingual maritime LoRA SFT, covering the collision-avoidance rules (COLREGs), ship handling, nautical navigation, weather routing, GMDSS communications, maritime conventions (SOLAS/MLC), safety and emergency response, and related fields. Native image-recognition ability is preserved (LoRA acts only on the language layers; the vision tower is unchanged).
from transformers import AutoModelForImageTextToText, AutoProcessor
import torch
mid = "xCloudinfo/Gemma-4-26B-A4B-Maritime-zhTW"
proc = AutoProcessor.from_pretrained(mid)
model = AutoModelForImageTextToText.from_pretrained(mid, dtype="auto", device_map="auto")
msgs = [{"role":"user","content":[{"type":"text","text":"請說明 COLREGs 中直航船與讓路船的避讓義務。"}]}]
ids = proc.apply_chat_template(msgs, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt").to(model.device)
out = model.generate(**ids, max_new_tokens=512)
print(proc.decode(out[0][ids["input_ids"].shape[1]:], skip_special_tokens=True))
For the GGUF version (llama.cpp / Ollama, with mmproj vision), see …-Maritime-zhTW-GGUF.
xCloudinfo/Gemma-4-26B-A4B-TAIDE-zhTW (underlying base google gemma-4 = Apache-2.0; includes a TAIDE derivative — use must comply with the TAIDE model license terms and retain this notice).Made by xCloudinfo (云碩科技) on its own AI compute resource pool; data stays local, process reproducible.
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