MergeSE - Post-hoc Model Merging for Software Engineering

Tool artifact for the ASE 2026 Tool-Track submission "MergeSE: Post-hoc Model Merging for Software Engineering Tasks Without Retraining"

The underlying model-merging approach comes from our research-track submission (under review): "A Unified Model for Cross-Domain Clone Detection via Model Merging"

MergeSE merges fine-tuned HuggingFace encoder checkpoints (CodeBERT, GraphCodeBERT, UniXcoder, CodeT5-encoder, ...) into a single model without any training data, and evaluates the result on standard software-engineering benchmarks. It ships as a single-file CLI and a web tool that share the same engine.

It implements:

• TIES-Merging (Yadav et al., NeurIPS 2023)
• DARE-TIES (Yu et al., 2024)
• WUDI-Merging (Cheng et al., ICML 2025)
• Task-vector averaging (Ilharco et al., 2022)

Plus end-to-end evaluation across the full range of SE classification tasks (clone detection, vulnerability detection, defect prediction, code-smell detection, commit classification, code-review acceptability, comment-code consistency, exception-type prediction, type inference, and any custom CSV) and one-command export to HuggingFace / ONNX / TorchScript.

---

Three ways to use MergeSE

Pick whichever fits your environment - all three sit on top of the same merging engine, so results are identical.

#	Path	Best for
1	Web tool via Docker	Easiest setup. One command brings up the UI and REST API.
2	Web tool without Docker	Same UI, but you'd rather run Flask directly in a venv.
3	CLI tool	Scripting, headless servers, reproducible runs, paper-grade evaluation.

Full references: docs/WEB.md and docs/CLI.md. Deploying to a dedicated VM: docs/DEPLOY.md.

---

Quickstart

git clone https://github.com/srlabUsask/MergeSE.git
cd MergeSE

1. Web tool with Docker

docker compose up -d --build      # -> http://localhost:8765

Open http://localhost:8765 in your browser. Stop with docker compose down.

Common first-run issues:

• permission denied ... /var/run/docker.sock - your user isn't in the docker group. One-time fix: sudo usermod -aG docker $USER, then open a fresh shell. Or prefix the one-off command with sudo.
• address already in use ... 8765 - something else is bound to port 8765. Either stop it (sudo ss -ltnp | grep :8765 to find the PID), or remap the host port in docker-compose.yml (e.g. "127.0.0.1:8766:8765") and use http://localhost:8766 instead.

2. Web tool without Docker (Flask in a venv)

python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install --index-url https://download.pytorch.org/whl/cpu torch
pip install ".[server,datasets]"

python server/app.py               # -> http://localhost:8765

For production, swap python server/app.py for gunicorn -c deploy/gunicorn.conf.py server.app:app.

3. CLI tool

python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install --index-url https://download.pytorch.org/whl/cpu torch
pip install .

mergese inspect ./model_a ./model_b --base microsoft/codebert-base
mergese merge   ./model_a ./model_b --base microsoft/codebert-base \
                --method ties --output ./merged
mergese evaluate ./merged --task clone_detection --test-file ./test.csv
mergese export   ./merged --format onnx --output ./merged.onnx

This installs a mergese console script on your $PATH. To run without installing, use python mergese.py ... after pip install -r requirements.txt.

---

Supported SE tasks

Task	Input	Metric	Known benchmarks
Clone detection	pair	binary F1	BigCloneBench, CLCDSA, GPTCloneBench, POJ-104
Vulnerability detection	single	binary F1	Devign, ReVeal, Big-Vul, D2A, Draper
Defect / bug prediction	single	binary F1	Defects4J, PROMISE, CodeXGLUE-Defect
Code-smell detection	single	binary F1	MLCQ, Qualitas
Commit classification	single	macro F1	CommitBench
Code-review acceptability	pair	binary F1	CodeReview
Comment-code consistency	pair	binary F1	comment-consistency datasets
Exception-type prediction	single	macro F1	CodeXGLUE-Exception
Type inference (closed)	single	macro F1	Typilus, Type4Py, ManyTypes4Py
Custom (any CSV)	auto	auto	-

mergese tasks (CLI) or GET /api/tasks (web) returns the same list.

Cross-task merging

When models have differently-shaped classifier heads (e.g. a 2-class clone detector + a 10-class commit classifier), MergeSE auto-detects the mismatch and runs an encoder-only merge. The base's head is preserved so you can attach a fresh task-specific head downstream. Force this with --encoder-only, or override with --include-heads.

---

Merge methods

Method	Idea	Key options
average	Mean of the task vectors θₖ - θ_base.	--weights
ties	Trim small deltas, elect a per-parameter majority sign, sum only sign-agreeing entries.	--trim-percentile, --weights
dare-ties	Randomly drop a fraction of delta entries and rescale, then apply TIES.	--drop-rate, --trim-percentile
wudi	Optimise each linear-layer delta to minimise cross-task interference; average the rest.	--wudi-steps, --wudi-lr, --device

WUDI (Weight Disentanglement Interference minimisation) refines the merged weight of every linear layer so that it interferes as little as possible with each task's own update direction. Like the other methods it needs no training data; it runs a short per-layer optimisation instead. It is the most compute- intensive method - use --device cuda when a GPU is available.

mergese merge \
    ./checkpoints/codebert_bcb \
    ./checkpoints/codebert_clcdsa \
    --base microsoft/codebert-base \
    --method wudi --wudi-steps 300 --wudi-lr 1e-5 \
    --output ./merged_wudi

---

Repository layout

MergeSE/
├── mergese.py              # the entire CLI (single file)
├── mergese_tasks.py        # task registry
├── server/
│   ├── app.py              # Flask backend
│   └── presets.json        # example workflows
├── frontend/
│   ├── index.html
│   ├── styles.css
│   ├── app.js
│   └── favicon.svg
├── deploy/
│   ├── provision.sh        # one-shot VM provisioning script
│   ├── cloud-init.yaml     # first-boot user data for a fresh VM
│   ├── nginx.conf          # reverse-proxy site
│   ├── mergese.service     # systemd unit
│   └── gunicorn.conf.py
├── data/
│   └── benchmarks/         # 200-row bundled samples + index.json
├── tests/
│   ├── test_merge_math.py
│   ├── test_tasks_and_heads.py
│   └── test_wudi.py
├── docs/
│   ├── CLI.md              # full CLI reference
│   ├── WEB.md              # full web-tool reference
│   └── DEPLOY.md           # VM deployment runbook
├── pyproject.toml
├── requirements.txt
├── Dockerfile
└── docker-compose.yml

---

Bundled benchmark samples

Name	Rows	Task	Source
bundled://bigclonebench	200 (100/100)	clone detection (Java)	CodeXGLUE / BigCloneBench
bundled://clcdsa	200 (100/100)	cross-language clones (Java<->Python)	CLCDSA Source Codes
bundled://gptclonebench	200 (100/100)	semantic clones (Java)	GPTCloneBench standalone

These are sampled from the original benchmarks for smoke-testing only. For paper-grade numbers, point --test-file at the full dataset.

---

Citing

If you use MergeSE itself, please cite the tool paper:

@inproceedings{roy2026mergese,
  author    = {Palash R. Roy and Banani Roy and Chanchal K. Roy and Kevin A. Schneider},
  title     = {MergeSE: Post-hoc Model Merging for Software Engineering Tasks Without Retraining},
  booktitle = {Proc. ASE Tool Track (under review)},
  year      = {2026}
}

If you use the merging methodology MergeSE packages, please also cite our research-track paper:

@inproceedings{roy2026unified,
  author    = {Palash R. Roy and Banani Roy and Chanchal K. Roy and Kevin A. Schneider},
  title     = {A Unified Model for Cross-Domain Clone Detection via Model Merging},
  booktitle = {Proc. ASE (Under Review)},
  year      = {2026}
}

License

Apache-2.0. See LICENSE.