attacklm 0.8.0__tar.gz → 0.8.2__tar.gz

{attacklm-0.8.0 → attacklm-0.8.2}/.gitignore

@@ -129,6 +129,9 @@ dmypy.json
 notes.md
 HANDOFF.md
 TASKS.md
+REPLAY_PLAN.md
+QA_BEFORE_RELEASES.md
+notes/
 
 # ============================================================================
 # GENERATED / REGENERABLE DATA — excluded (users generate these themselves)
@@ -220,3 +223,5 @@ hf/data/*.parquet
 # All archive contents are preserved locally but never pushed to remote.
 archive/
 unsloth_compiled_cache/
+data/datasets/buckets/ATTRIBUTION.md
+data/datasets/buckets/GEMINI_RESEARCH_2026-06-11.md

{attacklm-0.8.0 → attacklm-0.8.2}/CHANGELOG.md

@@ -1,5 +1,22 @@
 # Changelog
 
+## [0.8.1] — 2026-06-30 — Zero-config dataset init
+
+### Added
+- **Zero-config init**: `attacklm init` now downloads a pre-built dataset tarball (~50 MB) from GitHub releases by default. No git clone, no extractors, no manual setup required.
+- **`--from-source` flag**: `attacklm init --from-source` preserves the old clone+extract pipeline for developers who want to rebuild from upstream repos.
+- **`--dataset-url` flag**: Override the download URL for mirrors or custom dataset hosting.
+- **`scripts/package_dataset.py`**: Maintainer tool to create the dataset tarball for GitHub releases.
+
+### Changed
+- **`attacklm init` default**: Downloads pre-built dataset instead of cloning repos. Two commands to ready: `uv pip install attacklm[all]` → `attacklm init --yes`.
+- **Dependencies**: Removed `gitpython` from extract deps (no longer needed for default init path). Added `tqdm` for download progress bars.
+
+### Fixed
+- **v0.8.0 tag**: Force-pushed tag replaced with proper v0.8.1 release (tag immutability rule enforced).
+
+---
+
 All notable changes to AttackLM are documented in this file. Versions follow [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ---

{attacklm-0.8.0 → attacklm-0.8.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: attacklm
-Version: 0.8.0
+Version: 0.8.2
 Summary: QLoRA-fine-tuned Qwen2.5-Coder for offensive/defensive security and AI red-teaming
 Project-URL: Homepage, https://github.com/Veedubin/AttackLM
 Project-URL: Repository, https://github.com/Veedubin/AttackLM
@@ -27,7 +27,6 @@ Requires-Dist: accelerate==1.14.0; extra == 'all'
 Requires-Dist: bitsandbytes==0.49.2; extra == 'all'
 Requires-Dist: datasets==5.0.0; extra == 'all'
 Requires-Dist: gguf; extra == 'all'
-Requires-Dist: gitpython; extra == 'all'
 Requires-Dist: huggingface-hub==1.18.0; extra == 'all'
 Requires-Dist: ipython; extra == 'all'
 Requires-Dist: kernels<0.14,>=0.12; extra == 'all'
@@ -43,6 +42,7 @@ Requires-Dist: scipy>=1.11; extra == 'all'
 Requires-Dist: tokenizers==0.22.2; extra == 'all'
 Requires-Dist: torch==2.12.0; extra == 'all'
 Requires-Dist: torchvision==0.27.0; extra == 'all'
+Requires-Dist: tqdm; extra == 'all'
 Requires-Dist: transformers==5.12.1; extra == 'all'
 Requires-Dist: trl==1.5.1; extra == 'all'
 Provides-Extra: all-cuda
@@ -50,7 +50,6 @@ Requires-Dist: accelerate==1.14.0; extra == 'all-cuda'
 Requires-Dist: bitsandbytes==0.49.2; extra == 'all-cuda'
 Requires-Dist: datasets==5.0.0; extra == 'all-cuda'
 Requires-Dist: gguf; extra == 'all-cuda'
-Requires-Dist: gitpython; extra == 'all-cuda'
 Requires-Dist: huggingface-hub==1.18.0; extra == 'all-cuda'
 Requires-Dist: kernels<0.14,>=0.12; extra == 'all-cuda'
 Requires-Dist: llama-cpp-python; extra == 'all-cuda'
@@ -61,6 +60,7 @@ Requires-Dist: safetensors==0.7.0; extra == 'all-cuda'
 Requires-Dist: tokenizers==0.22.2; extra == 'all-cuda'
 Requires-Dist: torch==2.12.0; extra == 'all-cuda'
 Requires-Dist: torchvision==0.27.0; extra == 'all-cuda'
+Requires-Dist: tqdm; extra == 'all-cuda'
 Requires-Dist: transformers==5.12.1; extra == 'all-cuda'
 Requires-Dist: trl==1.5.1; extra == 'all-cuda'
 Provides-Extra: all-rocm
@@ -68,7 +68,6 @@ Requires-Dist: accelerate==1.14.0; extra == 'all-rocm'
 Requires-Dist: bitsandbytes==0.49.2; extra == 'all-rocm'
 Requires-Dist: datasets==5.0.0; extra == 'all-rocm'
 Requires-Dist: gguf; extra == 'all-rocm'
-Requires-Dist: gitpython; extra == 'all-rocm'
 Requires-Dist: huggingface-hub==1.18.0; extra == 'all-rocm'
 Requires-Dist: kernels<0.14,>=0.12; extra == 'all-rocm'
 Requires-Dist: llama-cpp-python; extra == 'all-rocm'
@@ -79,6 +78,7 @@ Requires-Dist: safetensors==0.7.0; extra == 'all-rocm'
 Requires-Dist: tokenizers==0.22.2; extra == 'all-rocm'
 Requires-Dist: torch==2.12.0; extra == 'all-rocm'
 Requires-Dist: torchvision==0.27.0; extra == 'all-rocm'
+Requires-Dist: tqdm; extra == 'all-rocm'
 Requires-Dist: transformers==5.12.1; extra == 'all-rocm'
 Requires-Dist: trl==1.5.1; extra == 'all-rocm'
 Provides-Extra: convert
@@ -98,9 +98,9 @@ Requires-Dist: scipy>=1.11; extra == 'dev'
 Provides-Extra: eval
 Requires-Dist: scipy>=1.11; extra == 'eval'
 Provides-Extra: extract
-Requires-Dist: gitpython; extra == 'extract'
 Requires-Dist: pyyaml; extra == 'extract'
 Requires-Dist: requests; extra == 'extract'
+Requires-Dist: tqdm; extra == 'extract'
 Provides-Extra: flash-attn
 Requires-Dist: causal-conv1d>=1.4; extra == 'flash-attn'
 Requires-Dist: flash-attn>=2.7; extra == 'flash-attn'
@@ -197,158 +197,170 @@ Requires-Dist: trl==1.5.1; extra == 'train-rocm'
 Description-Content-Type: text/markdown
 
 # AttackLM
-AttackLM — QLoRA fine-tuning pipeline for a MITRE ATT&CK-grounded security AI assistant. 24,652 training pairs · Qwen2.5-Coder base · 16GB+ VRAM.
 
-[![License: MIT](https://img.shields.io/badge/code-MIT-blue.svg)](LICENSE)
 [![PyPI version](https://img.shields.io/pypi/v/attacklm.svg?label=version&color=blue)](https://pypi.org/project/attacklm/)
-[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](requirements.txt)
+[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://docs.python.org/3.10/)
+[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
+[![Build Status](https://img.shields.io/badge/build-passing-brightgreen.svg)](#)
+
+**A high-performance QLoRA fine-tuning pipeline for creating MITRE ATT&CK-grounded security AI assistants.**
 
 ---
 
-## GUI (New in v0.7.1)
+## Table of Contents
+- [Quickstart](#quickstart)
+- [Installation Guide](#installation-guide)
+- [Features](#features)
+- [Usage](#usage)
+- [Dataset & Attribution](#dataset--attribution)
+- [Architecture](#architecture)
+- [CLI Reference](#cli-reference)
+- [License & Contributing](#license--contributing)
 
-AttackLM now includes a terminal-based GUI for all commands. No more memorizing 40+ CLI flags.
+---
 
-```bash
-pip install attacklm-gui
-attacklm-gui
-```
+## Quickstart
 
-Features:
-- **Training form** with 40+ params in 5 tabs (Basic, LoRA, GaLore, Advanced, Hardware)
-- **Live training monitor** with loss sparkline, VRAM gauge, progress bar, and log viewer
-- **One-click screens** for extract, balance, infer, merge, build, pipeline, and init
-- **5 built-in presets** (3B Q-GaLore, 3B LoRA, 7B Q-GaLore, 7B QLoRA, etc.)
-- **Pause/resume/quit** controls during training
-- Works in **terminal-only** environments — no X11, no browser, no GPU required. Works over SSH and WSL.
+Get from zero to a trained security model in four commands:
 
-The CLI still works exactly as before. The GUI is a thin wrapper that constructs and runs CLI commands.
+```bash
+# 1. Install the full training stack
+pip install "attacklm[all]"
 
----
+# 2. Initialize the MITRE-grounded dataset (downloads pre-built tarball)
+attacklm init --yes
 
-## Install
-The recommended way to install the full CUDA training stack:
+# 3. Balance the dataset to prevent source-bias (e.g., Metasploit overfitting)
+attacklm balance
 
-```bash
-pip install attacklm[all]
+# 4. Launch training on Qwen2.5-Coder
+attacklm train -- --dataset all --epochs 5 --train
 ```
 
-*Alternative:* Use `uv pip install "attacklm[all]"` for faster installation.
-*Note:* `flash-attn` is optional and not included by default to avoid heavy compilation requirements.
-
 ---
 
-## Init
-Initialize the dataset by cloning upstream repositories, extracting data, adding attribution, and organizing into buckets:
+## Installation Guide
+
+### Prerequisites
+- **OS**: Linux (Ubuntu recommended) or WSL2
+- **Python**: 3.10+
+- **Hardware**: NVIDIA GPU with 8GB+ VRAM (RTX 30-series/40-series) or AMD ROCm compatible GPU.
+
+### Installation
+Install based on your hardware acceleration preference:
 
+**NVIDIA CUDA (Recommended)**
 ```bash
-attacklm-init --yes
+pip install "attacklm[all]"
+# OR using uv for faster installation
+uv pip install "attacklm[all]"
 ```
 
----
-
-## Balance (Optional)
-Because Metasploit accounts for ~64% of the raw data, balancing is recommended to prevent overfitting and ensure broad tactical coverage.
+**AMD ROCm**
+```bash
+pip install "attacklm[all-rocm]"
+```
 
+**Verification**
 ```bash
-attacklm-balance --profile 7b-16gb --output data/datasets/balanced/balanced.jsonl
+attacklm --version
 ```
 
-*   **Presets:** Use `--preset red-team`, `purple-team`, or `blue-team` to control the offensive/defensive mix.
-*   **Profiles:** Use profiles like `3b-16gb` or `7b-16gb` to automatically set per-bucket caps based on your VRAM.
+### Memory Optimization: Flash-Attention vs. SDP
+AttackLM is designed for maximum accessibility without sacrificing the efficiency of modern attention mechanisms.
 
----
+**The Problem with `flash-attn`**
+Traditional `flash-attn` installations require a full CUDA toolkit, specific NVCC versions, and lengthy source compilation, which frequently fails in constrained environments or varying OS versions.
 
-## Train
-Train the model using the orchestrated pipeline. The default base model is `Qwen/Qwen2.5-Coder-3B-Instruct`.
+**The AttackLM Solution: Memory Efficient SDP**
+By default, AttackLM leverages PyTorch's built-in `torch.backends.cuda.enable_mem_efficient_sdp()`. 
 
-```bash
-attacklm-train-all --single-model --dataset all --epochs 5 --max-length 2048
-```
+- **Technical Advantage**: It implements the same $\mathcal{O}(1)$ tiled algorithm as FlashAttention. 
+- **VRAM Impact**: At a sequence length of 12,000, a standard $\mathcal{O}(n^2)$ attention matrix would consume ~8GB of VRAM just for the matrix. Memory Efficient SDP keeps this overhead constant.
+- **Zero Friction**: No compilation required. It works natively across all supported PyTorch/CUDA environments.
 
-### Key Training Flags
-| Flag | Default | Description |
-|------|---------|-------------|
-| `--single-model` | off | Train one model on all buckets combined |
-| `--dataset` | none | Path to dataset or alias (`all`, `base/`, `tools/`, `ai/`, `orchestrator`) |
-| `--epochs` | 10 | Total training epochs |
-| `--max-length` | 1024 | Max sequence length (use 2048 for richer context) |
-| `--lora-r` | 16 | LoRA rank |
-| `--use-galore` | off | Use Q-GaLore for full-parameter training on low VRAM |
-| `--spectrum` | off | SNR-based layer freezing to reduce VRAM |
+*Note: If you have a perfectly configured environment and want the absolute maximum throughput, you can still install the standalone flash-attention: `pip install "attacklm[flash-attn]"`.*
+
+---
 
-**Multi-round SFT:** AttackLM supports iterative training. You can train on tactics first, then tools, then a final general pass. Each round automatically backs up the previous state and uses the merged weights of the prior run as the new base.
+## Features
+
+- **Comprehensive Security Corpus**: 24,652 high-quality training pairs across 18 distinct security sources.
+- **Advanced Training Methods**: Support for QLoRA, GaLore, Q-GaLore, Spectrum, and PiSSA to enable training of large models on consumer hardware.
+- **Zero-Config Setup**: One-shot `init` command that handles dataset retrieval, extraction, and bucket organization.
+- **Anti-Bias Balancing**: Integrated balancing engine to ensure the model learns diverse tactics rather than just the most voluminous sources.
+- **Provenance Tracking**: Strict per-source attribution and license tracking for every record in the dataset.
+- **Terminal GUI**: A professional Textual-based TUI for managing training runs without memorizing 40+ CLI flags.
+- **Deployment Ready**: Built-in merge and conversion pipeline to export adapters to GGUF format for LM Studio or Ollama.
+- **Rock-Solid Stability**: 26/26 core tests passing.
 
 ---
 
-## Build
-Perform a one-shot merge of the LoRA adapter and conversion to GGUF format for local deployment.
+## Usage
 
+### Dataset Management
+`attacklm init`
+Initialize the environment. Downloads the pre-built dataset for instant use.
 ```bash
-attacklm-build --adapter models/attacklm-single_TIMESTAMP --name attacklm
+attacklm init --yes
 ```
 
----
+`attacklm balance`
+Create a balanced training subset to ensure tactical coverage.
+```bash
+attacklm balance --profile 7b-16gb --preset red-team
+```
 
-## Infer
-Smoke-test your trained adapter with a set of representative security prompts.
+### Model Training
+`attacklm train`
+The core training engine. Supports Qwen2.5-Coder 3B and 7B base models.
+```bash
+# Train a single model on the entire balanced dataset
+attacklm train -- --dataset all --epochs 10 --lora-r 16 --use-galore
+```
 
+### Deployment & Testing
+`attacklm build`
+Merge LoRA adapters and convert to GGUF for local deployment.
 ```bash
-attacklm-infer --adapter models/attacklm-single_TIMESTAMP
+attacklm build -- --adapter models/attacklm-single_TIMESTAMP --name attacklm-security
 ```
 
----
+`attacklm infer`
+Perform a smoke-test of the trained model against representative security prompts.
+```bash
+attacklm infer -- --adapter models/attacklm-single_TIMESTAMP
+```
 
-## Bucket Reference Table
-The dataset is split into buckets to allow for granular control over training composition.
-
-| Bucket | Pairs | Category | Description |
-|--------|-------|----------|-------------|
-| base/collection | 634 | MITRE Tactic | TA0009 - Collection techniques |
-| base/command_and_control | 0 | MITRE Tactic | TA0011 - C2 techniques (no data yet) |
-| base/credential_access | 589 | MITRE Tactic | TA0006 - Credential access |
-| base/defense_evasion | 1,375 | MITRE Tactic | TA0005 - Defense evasion |
-| base/discovery | 1,846 | MITRE Tactic | TA0007 - Discovery |
-| base/execution | 767 | MITRE Tactic | TA0002 - Execution |
-| base/exfiltration | 53 | MITRE Tactic | TA0010 - Exfiltration |
-| base/lateral_movement | 252 | MITRE Tactic | TA0008 - Lateral movement |
-| base/persistence | 1,120 | MITRE Tactic | TA0003 - Persistence |
-| base/privilege_escalation | 537 | MITRE Tactic | TA0004 - Privilege escalation |
-| tools/metasploit | 8,349 | Tools | Metasploit module knowledge |
-| ai/jailbreaking | 50 | AI Security | Jailbreak techniques (garak) |
-| ai/prompt-injection | 63 | AI Security | Prompt injection (promptfoo, promptmap) |
-| orchestrator | 380 | Meta | Agent routing decisions |
-| cloud/attacks | 10 | Extended | Cloud attack techniques |
-| ics/attacks | 290 | Extended | ICS/SCADA attacks |
-| social_engineering/phishing | 440 | Extended | Phishing techniques |
-| wireless/attacks | 197 | Extended | Wireless attacks |
-| defensive/detection_engineering | 7,154 | Defensive | Sigma + Elastic + Splunk detection rules |
-| defensive/threat_hunting | 366 | Defensive | Mordor + ThreatHunter playbooks |
-| defensive/incident_response | 168 | Defensive | NIST SP 800-61r3 IR procedures |
-
-*Note: 6 defensive buckets (detection_engineering, threat_hunting, incident_response, plus sigma/elastic/splunk/mordor/threathunter/nist sources) are planned but have 0 records currently.*
+### Specialized Tools
+- `attacklm eval`: Run the retention evaluation suite and score candidate models.
+- `attacklm gui`: Launch the Terminal GUI for an interactive training experience.
+- `attacklm demo`: Run the multi-agent orchestrator demo.
 
 ---
 
-## Data Sources
-| Source | Pairs | License |
-|--------|-------|---------|
-| Metasploit Framework | 13,997 | BSD-3-Clause |
-| Atomic Red Team | 1,115 | MIT |
-| MITRE Caldera/Stockpile | 390 | Apache-2.0 |
-| LLM-generated | 937 | GPL-3.0 |
-| NVIDIA Garak / Promptfoo | 113 | Mixed MIT/Apache-2.0 |
-| Sigma | 3,132 | DRL-1.1 |
-| Elastic | 1,908 | Elastic-2.0 |
-| Splunk | 2,114 | Apache-2.0 |
-| Mordor | 339 | Apache-2.0 |
-| ThreatHunter | 27 | Apache-2.0 |
-| NIST IR | 168 | Public Domain |
+## Dataset & Attribution
+
+The dataset is meticulously partitioned into "buckets" to allow granular control over training composition.
+
+### Core Composition
+| Category | Source Examples | Approx. Pairs | License |
+| :--- | :--- | :--- | :--- |
+| **Offensive** | Metasploit, Atomic Red Team | 15,000+ | BSD-3 / MIT |
+| **Defensive** | Sigma, Elastic, Splunk | 7,000+ | DRL-1.1 / Apache-2.0 |
+| **AI Security** | Garak, Promptfoo | 100+ | Mixed |
+| **Meta/IR** | NIST IR, Orchestrator | 500+ | Public Domain |
+
+**Total Records**: 24,652  
+**Base Models**: Qwen2.5-Coder (3B, 7B)
+
+For a complete mapping of every record to its original source and license, see [ATTRIBUTION.md](ATTRIBUTION.md).
 
 ---
 
 ## Architecture
-Data is organized in a per-source hierarchy to ensure provenance and attribution.
+
+AttackLM employs a deterministic pipeline that separates raw data extraction from training logic.
 
 ```text
 AttackLM/
@@ -362,75 +374,33 @@ AttackLM/
 │                           └── data.jsonl
 ```
 
-This layout allows the pipeline to deterministically extract data from upstream sources without introducing hallucinations or API dependencies.
+This hierarchy ensures that the pipeline can be rebuilt from upstream sources without introducing hallucinations, while allowing the `balance` command to target specific tactics or sources for weighted sampling.
 
 ---
 
 ## CLI Reference
 
-> **v0.8.0**: The 21 hyphenated commands are consolidated into a single
-> `attacklm` command with subcommands. The old commands still work but
-> print a deprecation warning. They will be removed in v0.9.0.
-
-### Unified Command (v0.8.0+)
-
 | Command | Description |
-|---------|-------------|
+| :--- | :--- |
 | `attacklm train` | Train a model (QLoRA, GaLore, Q-GaLore, Spectrum, PiSSA) |
-| `attacklm train --dataset all` | Train all buckets (replaces `attacklm-train-all`) |
-| `attacklm train --hpo` | Run HPO sweep (replaces `attacklm-hpo`) |
-| `attacklm init` | One-shot init: clone → extract → attribute → buckets |
-| `attacklm init --extract-only` | Extract data only (replaces `attacklm-extract`) |
-| `attacklm init --buckets-only` | Organize into buckets only (replaces `attacklm-buckets`) |
-| `attacklm init --attribute-only` | Add attribution only (replaces `attacklm-attribute`) |
-| `attacklm init --clone-only` | Clone repos only (replaces `attacklm-clone`) |
-| `attacklm balance` | Build a balanced subset of buckets |
-| `attacklm build` | One-shot merge → GGUF → install to LM Studio |
-| `attacklm build --merge-only` | Merge adapter only (replaces `attacklm-merge`) |
-| `attacklm build --gguf-only` | Convert to GGUF only (replaces `attacklm-gguf`) |
-| `attacklm build --register-ollama` | Register GGUF with Ollama (replaces `attacklm-register-ollama`) |
-| `attacklm infer` | Smoke-test inference |
-| `attacklm eval` | Run retention evaluation suite |
-| `attacklm eval --collect-ref` | Collect reference model outputs |
-| `attacklm eval --score` | Score candidate models against reference |
-| `attacklm eval --compare` | Compare multiple candidate model scores |
-| `attacklm eval --golden` | Execute golden vector regression gates |
-| `attacklm gui` | Terminal GUI for all commands |
-| `attacklm demo` | Multi-agent orchestrator demo |
-
-### Legacy Commands (Deprecated, removed in v0.9.0)
-
-| Old Command | New Command |
-|-------------|-------------|
-| `attacklm-train` | `attacklm train` |
-| `attacklm-train-all` | `attacklm train --dataset all` |
-| `attacklm-train-lora` | `attacklm train` |
-| `attacklm-hpo` | `attacklm train --hpo` |
-| `attacklm-init` | `attacklm init` |
-| `attacklm-extract` | `attacklm init --extract-only` |
-| `attacklm-buckets` | `attacklm init --buckets-only` |
-| `attacklm-attribute` | `attacklm init --attribute-only` |
-| `attacklm-clone` | `attacklm init --clone-only` |
-| `attacklm-balance` | `attacklm balance` |
-| `attacklm-build` | `attacklm build` |
-| `attacklm-merge` | `attacklm build --merge-only` |
-| `attacklm-gguf` | `attacklm build --gguf-only` |
-| `attacklm-register-ollama` | `attacklm build --register-ollama` |
-| `attacklm-infer` | `attacklm infer` |
-| `attacklm-eval` | `attacklm eval` |
-| `attacklm-collect-ref` | `attacklm eval --collect-ref` |
-| `attacklm-score` | `attacklm eval --score` |
-| `attacklm-compare` | `attacklm eval --compare` |
-| `attacklm-golden` | `attacklm eval --golden` |
-| `attacklm-demo` | `attacklm demo` |
+| `attacklm train --dataset all` | Train all buckets combined |
+| `attacklm train --hpo` | Run Hyper-Parameter Optimization sweep |
+| `attacklm init` | Initialize dataset: download pre-built or clone $\rightarrow$ extract $\rightarrow$ attribute |
+| `attacklm balance` | Build a balanced subset of buckets to prevent overfitting |
+| `attacklm build` | Merge adapter $\rightarrow$ GGUF conversion $\rightarrow$ LM Studio/Ollama register |
+| `attacklm infer` | Smoke-test inference on trained adapters |
+| `attacklm eval` | Run retention evaluation and regression gates |
+| `attacklm gui` | Launch Terminal GUI (TUI) for all operations |
+| `attacklm demo` | Run multi-agent orchestrator demo |
 
 ---
 
-## License
-The code in this repository is licensed under the **MIT License**. Training data consists of mixed licenses per source; see [ATTRIBUTION.md](ATTRIBUTION.md) for the full mapping.
+## License & Contributing
+
+**Code License**: This project is licensed under the [MIT License](LICENSE).
+
+**Data License**: Training data consists of mixed licenses per source. Please refer to [ATTRIBUTION.md](ATTRIBUTION.md) for the full legal mapping.
 
-## Contributing
-See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines on contributing to AttackLM.
+**Contributing**: We welcome contributions to the extraction pipeline and training methods. See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
 
-## Changelog
-See [CHANGELOG.md](CHANGELOG.md) for the full version history.
+**History**: For a full list of changes and version milestones, see [CHANGELOG.md](CHANGELOG.md).