Resources model

Paper code (Chinese)

• pretrain or full: train (nhưng timeout 12h) - đã có dapt, tapt, assembly

• python

• assembly: lấy từ pretrain, checkpoint để làm tiếp - assembly

• Model folder

• Demo YT 1 line

• Demo YT multi line

My improvement (English)

• dapt

• tapt

• full pretrain_mode: model fg-codebert pretrain and code

• python

• assembly

• code

• datasets

Usage Guide

1. Backend Setup (Python)

Create virtual environment

python -m venv .venv

Activate venv:

• Windows

.venv\Scripts\activate

• Linux / macOS

source .venv/bin/activate

Install dependencies

pip install -r requirements.txt

Run backend API

python asmd_http.py

---

2. VS Code Extension Setup (Node.js)

Install dependencies

npm install

Compile & watch extension

npm run compile
npm run watch

Press F5 to start Extension Development Host for debugging.

---

3. Run Tests

pytest

---

4. Package & Install Extension

Make sure vsce is installed:

npm install -g @vscode/vsce

Package extension

npx vsce package

Install extension

code --install-extension asm-suggest-0.0.2.vsix

Cli

Server

python asm_cli.py --server

Client pipe qua daemon:

python asm_cli.py --client --source "add 0x10..." --similarity "add var0..." --var var0=0x10 --var var1=esi | python asmd.py

Chạy trực tiếp (không daemon):

python asm_cli.py --source "add 0x10 to the current byte in esi" --similarity "add var0 to current byte in var1" --var var0=0x10 --var var1=esi --json

python asmd.py 

{"source":"add 0x10 to the current byte in esi","similarity":"add var0 to current byte in var1","var_map":{"var0":"0x10","var1":"esi"}}

$env:TRANSFORMERS_VERBOSITY="error"