AI 模型库
gemma-3-12b-it
image-text-to-textimage-text-to-text
z_image_turbo
工作流：https://comfyanonymous.github.io/ComfyUI_examples/z_image/
moondream2
image-text-to-text⚠️ 本仓库包含上一代模型 Moondream 2 的最新版本。Moondream 的最新版本为 Moondream 3（预览版）。
multi-qa-mpnet-base-dot-v1
sentence-similaritymulti-qa-mpnet-base-dot-v1 这是一个句子变换器模型：它将句子和段落映射到768维的稠密向量空间，专为**语义搜索**设计。该模型基于来自不同来源的2.15亿个（问题，答案）对进行训练。如需了解语义搜索的入门介绍，请参阅
Qwen3-Coder-30B-A3B-Instruct
text-generation**Qwen3-Coder** is available in multiple sizes. Today, we're excited to introduce **Qwen3-Coder-30B-A3B-Instruct**. This streamlined model maintains impressive performance and efficiency, featuring th
Qwen3.6-35B-A3B-GGUF
image-text-to-textSee Unsloth Dynamic 2.0 GGUFs for our quantization benchmarks. NEW: Developer Role Support so Qwen3.6 can work in Codex, OpenCode a
e5-large
sentence-similarity**新闻（2023年5月）：请切换至e5-large-v2，该版本性能更优且使用方法相同。**
yolo-world-mirror
https://docs.ultralytics.com/models/yolo-world/#可用模型、支持任务及运行模式
multilingual-e5-small
feature-extractionhttps://huggingface.co/intfloat/multilingual-e5-small 附带ONNX权重，以兼容Transformers.js。
ms-marco-MiniLM-L4-v2
text-ranking该模型基于MS Marco段落排序任务进行了训练。
Qwen3-14B
text-generationQwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groun
gemma-4-26B-A4B-it-AWQ-4bit
image-text-to-textVersion 26.05.01 Calibration STEM and Agentic Languages EN ZH HI AR RU JA KO NL FR ES Model Size
Qwen2.5-14B-Instruct
text-generationQwen2.5是Qwen大语言模型的最新系列。针对Qwen2.5，我们发布了多个基础语言模型和指令微调语言模型，参数量从0.5亿到720亿不等。相较于Qwen2，Qwen2.5带来了以下改进：
deberta-v3-base
fill-maskDeBERTaV3：使用梯度解耦嵌入共享的ELECTRA式预训练改进DeBERTa
Qwen3Guard-Gen-0.6B
text-generation**Qwen3Guard** is a series of safety moderation models built upon Qwen3 and trained on a dataset of 1.19 million prompts and responses labeled for safety. The series includes models of three sizes (0.
bert-large-portuguese-cased
fill-maskBERTimbau Large（亦称“bert-large-portuguese-cased”）
bge-small-zh-v1.5
feature-extractionModel List | FAQ | Usage | Evaluation | Train | Contact | Citation | License
t5-base
translation1. 模型详情 2. 用途 3. 偏见、风险与局限性 4. 训练详情 5. 评估 6. 环境影响 7. 引用 8. 模型卡片作者 9. 如何开始使用该模型
Qwen3-VL-4B-Instruct
image-text-to-textMeet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
docling-models
此页面包含驱动PDF文档转换包docling的模型。