qwen2vl-7b-lora-apps

bmh201708/qwen2vl-7b-lora-apps

qwen2vl-7b-lora-apps is a machine learning model from bmh201708. Use it for the machine learning task on the model card, and read the license before you ship it in a product.

This repository contains LoRA (Low-Rank Adaptation) adapters for the Qwen2-VL-7B-Instruct model, fine-tuned on different mobile application datasets.

Downloads · 30 days

Access

Public

Updated Feb 2, 2026

Repo size

1.1 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors2.3 GB · 100%

At a glance

Access: Public
Created: Feb 2, 2026
Updated: Feb 2, 2026
SHA: dbaccca5

Model card

From the Hugging Face model README

Qwen2-VL-7B App LoRA Adapters

This repository contains LoRA (Low-Rank Adaptation) adapters for the Qwen2-VL-7B-Instruct model, fine-tuned on different mobile application datasets.

Model Description

These LoRA adapters were trained using federated learning approaches on app-specific datasets from the FedMABench benchmark. Each app contains:

v0: Initial training checkpoint (基础训练后的初始LoRA)
global_lora_10: Global aggregated LoRA after 10 rounds of federated learning (第10轮联邦学习聚合后的全局LoRA)

Available Apps (14 total)

App	v0 Checkpoint	Global LoRA (Round 10)	Training Version
adidas	✅ checkpoint-108	✅ global_lora_10	v0-20260131-111220
amazon	✅ checkpoint-312	✅ global_lora_10	v0-20260130-190721
calendar	✅ checkpoint-129	✅ global_lora_10	v0-20260131-131350
clock	✅ checkpoint-198	✅ global_lora_10	v0-20260131-052630
decathlon	✅ checkpoint-63	✅ global_lora_10	v0-20260131-122711
ebay	✅ checkpoint-219	✅ global_lora_10	v0-20260130-223012
etsy	✅ checkpoint-60	✅ global_lora_10	v0-20260131-202612
flipkart	✅ checkpoint-174	✅ global_lora_10	v0-20260131-010305
gmail	✅ checkpoint-201	✅ global_lora_10	v0-20260131-025059
google_drive	✅ checkpoint-63	✅ global_lora_10	v0-20260131-103300
google_maps	✅ checkpoint-60	✅ global_lora_10	v0-20260131-150323
kitchen_stories	✅ checkpoint-75	✅ global_lora_10	v0-20260131-155209
reminder	✅ checkpoint-138	✅ global_lora_10	v0-20260131-073315
youtube	✅ checkpoint-78	✅ global_lora_10	v0-20260131-093519

Directory Structure

qwen2vl-7b-lora-apps/
├── adidas/
│   ├── v0/                 # Initial checkpoint LoRA
│   └── global_lora_10/     # Round 10 federated LoRA
├── amazon/
│   ├── v0/
│   └── global_lora_10/
├── calendar/
│   ├── v0/
│   └── global_lora_10/
... (14 apps total)

Usage

Loading with PEFT

from peft import PeftModel
from transformers import Qwen2VLForConditionalGeneration

# Load base model
base_model = Qwen2VLForConditionalGeneration.from_pretrained(
    "Qwen/Qwen2-VL-7B-Instruct",
    torch_dtype="auto",
    device_map="auto"
)

# Load a specific app LoRA (e.g., amazon global_lora_10)
model = PeftModel.from_pretrained(
    base_model,
    "bmh201708/qwen2vl-7b-lora-apps",
    subfolder="amazon/global_lora_10"
)

Loading with Hugging Face Hub

from huggingface_hub import snapshot_download

# Download specific app LoRA
local_path = snapshot_download(
    repo_id="bmh201708/qwen2vl-7b-lora-apps",
    allow_patterns=["amazon/global_lora_10/*"]
)

Base Model

Model: Qwen/Qwen2-VL-7B-Instruct
LoRA Library: PEFT 0.17.1

Training Details

These LoRAs were trained as part of the FedMABench (Federated Mobile Agent Benchmark) project for mobile GUI agent tasks. Each app represents a specific mobile application domain.

Related Repositories

Category LoRAs: bmh201708/qwen2vl-7b-lora-categories - LoRAs organized by category (Entertainment, Lives, Office, Shopping, Traveling)