Skip to main content

LaKun:JEV 决策类型模型的多模态版本

灵感来源于JEV,感谢laya https://github.com/mizorewww/laya-mlx 提供的优化方式(RLCD),所以取名文LaJ(谐音垃圾),因致敬前辈故LaKun诞生 LaKun是决策类型模型的多模态版本 面对一段文本状态或一张图片,它直接回答我指定的选择、档位评分和二元判断问题, 输出每个候选项的softmax分数与最高分选项,不生成自由文本。一次调用最多可提交20道问题,目前每次最多处理一张图片。 这个PyPI安装包提供推理代码,源码包另含训练脚本;两种发行文件都不包含权重或数据集。我的完整检查点有 756,409,158 个参数;0.7B 是模型仓库名中的近似称呼。

安装与使用

先安装适合本机的 PyTorch,再安装 LaKun:

python -m pip install lakun

模型权重独立存放在我的魔塔模型仓库,不随PyPI包安装; 获取权重后,保留包含 model_config.json、权重文件、编码器配置、tokenizer 和图像处理器的完整目录。直接运行下面的命令,程序会先询问该目录的位置,再逐项询问文本状态或本地图片路径、题型和问题: 完整代码见我的个人仓库。

lakun

输入图片路径即可问单图问题;纯文本任务直接跳过图片路径。选择题的候选项用 | 分隔。--checkpoint 接受下载后的完整本地目录,也接受 hh108801/LaKun-0.7B 魔塔仓库 ID 并自动下载。需要在脚本中使用时:

from lakun import LaKunPredictor

model = LaKunPredictor.from_pretrained("hh108801/LaKun-0.7B")
result = model.predict(
    [{"type": "choice", "question": "这是什么类型的请求?", "criteria": ["咨询", "投诉", "退款"]}],
    state="用户说:我被重复扣费了,请尽快退款。",
)
print(result["answers"][0])

choice 接受 2–20 个候选项,score 接受 3–20 个有序档位,noul 固定为 false/true。predicted_index 从 0 开始。上下文最多 512 token;输入超限时,我的实现优先保留问题和候选项,再保留状态开头。。

Release files for lakun 0.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for lakun 0.1.2
File Size Uploaded
lakun-0.1.2.tar.gz 53.1 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for lakun 0.1.2
File Interpreter ABI Platform
lakun-0.1.2-py3-none-any.whl Python 3 none any Details

Total release size: 94.0 kB

Release files / lakun-0.1.2.tar.gz

Download URL lakun-0.1.2.tar.gz
Size 53.1 kB
Tags Source
SHA-256 checksum
How to use checksums
b00f958ff21b656ebc9356465553f42678a6a396e5c9558288182e487ad729a4
BLAKE2b-256 checksum
How to use checksums
d476be33e51ad8dd0777e171440f9ccb1168b2414623db09388ba8a734787830
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.12

Release files / lakun-0.1.2-py3-none-any.whl

Download URL lakun-0.1.2-py3-none-any.whl
Size 40.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
85bbe12c5d3bc83eadff3af0dd6d03adbf1dce071cc3df1d0686d13cfde0c306
BLAKE2b-256 checksum
How to use checksums
4be9d766526c3b8abeb0a728be53f6fff4d5b1b0d721b3e59f0c67787dbc7a12
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.12

Release history Release notifications | RSS feed

This release

0.1.2 This release

2 release files

0.1.1

2 release files

0.1.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page