Downloads · 30 days
0
qwen3vl-dino-lora
Sp1keeee/qwen3vl-dino-lora
qwen3vl-dino-lora is a machine learning model from Sp1keeee. Use it for the machine learning 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.
基于 Qwen3-VL-2B 的 LoRA 微调权重，用于玩 Chrome 恐龙游戏。
Access
Public
Updated Jan 25, 2026
Repo size
34.9 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors34.9 MB · 100%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Jan 25, 2026
- Updated
- Jan 25, 2026
- SHA
- 1fd1da6a
Base models
Model card
From the Hugging Face model README
Qwen3-VL Chrome Dinosaur LoRA
基于 Qwen3-VL-2B 的 LoRA 微调权重，用于玩 Chrome 恐龙游戏。
使用方法
from transformers import Qwen3VLForConditionalGeneration
from peft import PeftModel
# 加载基础模型
base_model = Qwen3VLForConditionalGeneration.from_pretrained("Qwen/Qwen3-VL-2B")
# 加载 LoRA 权重
model = PeftModel.from_pretrained(base_model, "你的用户名/qwen3vl-dino-lora")
训练参数
- 基础模型: Qwen3-VL-2B
- 训练框架: LLaMA-Factory
- 训练数据: 3233 样本 (Jump: 44%, Run: 40%, Duck: 16%)
- 训练轮数: 3 epochs
- 学习率: 1e-4
- LoRA rank: 8
模型能力
输入游戏截图，输出动作指令：
- Jump - 跳跃（躲避仙人掌）
- Duck - 下蹲（躲避飞鸟）
- Run - 继续奔跑（无障碍物）
- License
- apache-2.0
- Languages
- zh, en
- Created
- Jan 25, 2026
- Updated
- Jan 25, 2026