bit-ttt-engine 0.6.0__tar.gz

bit_ttt_engine-0.6.0/PKG-INFO

@@ -0,0 +1,91 @@
+Metadata-Version: 2.4
+Name: bit-ttt-engine
+Version: 0.6.0
+Classifier: Development Status :: 4 - Beta
+Classifier: Programming Language :: Rust
+Classifier: Programming Language :: Python :: Implementation :: CPython
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+License-File: LICENSE
+Summary: 1.58-bit Quantization + Test-Time Training (TTT) Implementation in Pure Rust
+Keywords: llm,rust,ttt,quantization,ai
+Author: imonoonoko
+License: MIT
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
+
+# Bit-TTT Engine: High-Performance Brain Core
+
+[![Rust](https://img.shields.io/badge/rust-1.70+-orange.svg)](https://www.rust-lang.org/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
+[![PyPI](https://img.shields.io/pypi/v/bit-ttt-engine.svg)](https://pypi.org/project/bit-ttt-engine/)
+
+**1.58-bit Quantization + Test-Time Training (TTT)** Implementation in Pure Rust.
+
+This package provides Python bindings for the Bit-TTT Engine, allowing you to run ultra-light ternary LLMs with real-time adaptation.
+
+## ✨ Features
+1. **Ultra-Light**: Runs large LLMs on cheap hardware using **1.58-bit (ternary) weights**.
+2. **Adaptive (TTT)**: Learns *while* inferring, adapting to context in real-time.
+3. **Pure Rust**: High performance with minimal dependencies.
+
+## 🚀 Installation
+
+```bash
+pip install bit-ttt-engine
+```
+
+## 💻 Usage
+
+```python
+import cortex_rust
+import json
+
+# Initialize Configuration
+config = cortex_rust.BitLlamaConfig(
+    vocab_size=32000,
+    hidden_dim=512,
+    num_layers=12,
+    inner_lr=0.001
+)
+
+# Initialize Model (Inference)
+model = cortex_rust.BitLlama(
+    config=config,
+    checkpoint_path="path/to/model.safetensors",
+    device="cpu", # or "cuda"
+    tokenizer_path="path/to/tokenizer.json"
+)
+
+# Generate Text
+output = model.generate(prompt="Hello, world!", max_tokens=50)
+print(output)
+```
+
+## 🏗️ Training (TTT)
+
+```python
+trainer = cortex_rust.PyTrainer(
+    config=config,
+    checkpoint_path="path/to/model.safetensors",
+    device="cuda"
+)
+
+# Single training step
+loss = trainer.train_step(input_ids=[...], targets=[...])
+print(f"Loss: {loss}")
+
+# Save checkpoint
+trainer.save_checkpoint("model_updated.safetensors")
+```
+
+## 📖 Documentation
+For more details, please visit the [GitHub repository](https://github.com/imonoonoko/Bit-TTT-Engine).
+
+## 🙏 Acknowledgments
+This project incorporates ideas and techniques inspired by the DroPE method published by Sakana AI.
+
+## 💖 License
+MIT License
+

bit_ttt_engine-0.6.0/README_PYPI.md

@@ -0,0 +1,73 @@
+# Bit-TTT Engine: High-Performance Brain Core
+
+[![Rust](https://img.shields.io/badge/rust-1.70+-orange.svg)](https://www.rust-lang.org/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
+[![PyPI](https://img.shields.io/pypi/v/bit-ttt-engine.svg)](https://pypi.org/project/bit-ttt-engine/)
+
+**1.58-bit Quantization + Test-Time Training (TTT)** Implementation in Pure Rust.
+
+This package provides Python bindings for the Bit-TTT Engine, allowing you to run ultra-light ternary LLMs with real-time adaptation.
+
+## ✨ Features
+1. **Ultra-Light**: Runs large LLMs on cheap hardware using **1.58-bit (ternary) weights**.
+2. **Adaptive (TTT)**: Learns *while* inferring, adapting to context in real-time.
+3. **Pure Rust**: High performance with minimal dependencies.
+
+## 🚀 Installation
+
+```bash
+pip install bit-ttt-engine
+```
+
+## 💻 Usage
+
+```python
+import cortex_rust
+import json
+
+# Initialize Configuration
+config = cortex_rust.BitLlamaConfig(
+    vocab_size=32000,
+    hidden_dim=512,
+    num_layers=12,
+    inner_lr=0.001
+)
+
+# Initialize Model (Inference)
+model = cortex_rust.BitLlama(
+    config=config,
+    checkpoint_path="path/to/model.safetensors",
+    device="cpu", # or "cuda"
+    tokenizer_path="path/to/tokenizer.json"
+)
+
+# Generate Text
+output = model.generate(prompt="Hello, world!", max_tokens=50)
+print(output)
+```
+
+## 🏗️ Training (TTT)
+
+```python
+trainer = cortex_rust.PyTrainer(
+    config=config,
+    checkpoint_path="path/to/model.safetensors",
+    device="cuda"
+)
+
+# Single training step
+loss = trainer.train_step(input_ids=[...], targets=[...])
+print(f"Loss: {loss}")
+
+# Save checkpoint
+trainer.save_checkpoint("model_updated.safetensors")
+```
+
+## 📖 Documentation
+For more details, please visit the [GitHub repository](https://github.com/imonoonoko/Bit-TTT-Engine).
+
+## 🙏 Acknowledgments
+This project incorporates ideas and techniques inspired by the DroPE method published by Sakana AI.
+
+## 💖 License
+MIT License

bit_ttt_engine-0.6.0/crates/rust_engine/.cargo/config.toml

@@ -0,0 +1,2 @@
+[build]
+rustflags = ["-C", "target-cpu=native"]

bit_ttt_engine-0.6.0/crates/rust_engine/CHANGELOG.md

@@ -0,0 +1,99 @@
+# Changelog
+
+All notable changes to `cortex_rust` will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+## [0.5.0] - 2026-01-28
+
+### 🚨 Breaking Changes
+- `paged_attention_v1()` signature changed: `num_kv_heads` now required for GQA support
+- `CacheConfig::new()` parameter order reorganized for clarity
+
+### Added
+- **GQA (Grouped Query Attention) support** in PagedAttention kernel
+- `num_kv_heads` parameter for proper KV head mapping
+- Detailed profiling methods: `forward_profiled()`, `forward_timed()`
+- `PreallocKVCache` - pre-allocated KV cache to reduce allocation overhead
+
+### Fixed
+- **[CRITICAL] Long sequence quality degradation** - see "Known Issues Resolved" below
+- Shared memory overflow for sequences > 2048 tokens
+- Warp reduction bug in softmax max computation
+- Context length calculation in block manager (off-by-one)
+
+### Changed
+- Increased default `max_context_len` to 4096 for shared memory allocation
+- Improved error messages with actual tensor dimensions
+- Block size default changed from 16 to 32 for better memory efficiency
+
+### Known Issues Resolved
+#### Long Sequence Quality Degradation Root Causes
+
+1. **Query Loading Bug in PagedAttention** (`paged_attention.cu`)
+   - Query was loaded into per-thread local array `float q[128]`
+   - Each thread only loaded some elements, but inner loop accessed all
+   - **Result**: Reading uninitialized memory → garbage attention scores
+   - **Fix**: Load query into shared memory (`q_shared`)
+   - **Commit**: a6f14c8
+
+2. **Block Manager Token Count** (`block_manager.rs`)
+   - `allocate_slots()` used `blocks.len() * block_size` for current position
+   - With block_size=16, after 5 tokens: calculated 16 instead of 5
+   - **Result**: New tokens written to wrong slots (e.g., slot 16 instead of 5)
+   - **Fix**: Added `seq_to_num_tokens` HashMap to track actual token count
+   - **Commit**: f556c15
+
+#### Additional Fixes in 0.5.0
+3. **unpack_4bit offset bug** (`linear_4bit.rs`)
+   - Old: `if w > 7 { w - 16 } else { w }` (wrong)
+   - New: `w - 8` (Python compatible)
+   - **Commit**: 7ffcdc2
+
+4. **Dynamic shared memory** (`paged_attention.rs`)
+   - Old: Hardcoded `max_context_len = 2048`
+   - New: Calculated from actual `context_lens` tensor
+   - **Commit**: 9a748ec
+
+#### Verified (No Bug)
+- Warp reduction: Works correctly with `blockDim.x = 32`
+
+## [0.4.0] - 2025-01-XX
+
+### Added
+- PagedAttention CUDA kernels (`paged_attention.cu`)
+- `BlockManager` for cache block allocation
+- `CacheEngine` for GPU memory management
+- `generate_paged()` for memory-efficient inference
+- 4-bit quantized model support (`Llama4Bit`)
+- Fused CUDA kernels: `fused_silu_mul_cuda`, `softmax_cuda`
+
+### Changed
+- RoPE implementation moved to precomputed caches
+- KV cache supports both traditional (`Tensor::cat`) and paged modes
+
+### Fixed
+- Causal mask offset calculation for decode phase
+
+## [0.3.0] - 2025-01-XX
+
+### Added
+- Initial CUDA kernel infrastructure
+- 4-bit GEMM implementation (`gemm_4bit`)
+- RMSNorm layer
+
+## [0.2.0] - 2025-01-XX
+
+### Added
+- Basic transformer block implementation
+- Embedding and LM head layers
+
+## [0.1.0] - 2025-01-XX
+
+### Added
+- Initial project structure
+- Candle-core integration
+- Basic tensor operations

bit_ttt_engine-0.6.0/crates/rust_engine/Cargo.lock

@@ -0,0 +1,249 @@
+# This file is automatically @generated by Cargo.
+# It is not intended for manual editing.
+version = 4
+
+[[package]]
+name = "autocfg"
+version = "1.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c08606f8c3cbf4ce6ec8e28fb0014a2c086708fe954eaa885384a6165172e7e8"
+
+[[package]]
+name = "cfg-if"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
+
+[[package]]
+name = "cortex_rust"
+version = "0.1.0"
+dependencies = [
+ "libc",
+ "ndarray",
+ "rand",
+]
+
+[[package]]
+name = "crossbeam-deque"
+version = "0.8.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9dd111b7b7f7d55b72c0a6ae361660ee5853c9af73f70c3c2ef6858b950e2e51"
+dependencies = [
+ "crossbeam-epoch",
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "crossbeam-epoch"
+version = "0.9.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5b82ac4a3c2ca9c3460964f020e1402edd5753411d7737aa39c3714ad1b5420e"
+dependencies = [
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "crossbeam-utils"
+version = "0.8.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d0a5c400df2834b80a4c3327b3aad3a4c4cd4de0629063962b03235697506a28"
+
+[[package]]
+name = "either"
+version = "1.15.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "48c757948c5ede0e46177b7add2e67155f70e33c07fea8284df6576da70b3719"
+
+[[package]]
+name = "getrandom"
+version = "0.2.16"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "335ff9f135e4384c8150d6f27c6daed433577f86b4750418338c01a1a2528592"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "wasi",
+]
+
+[[package]]
+name = "libc"
+version = "0.2.179"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c5a2d376baa530d1238d133232d15e239abad80d05838b4b59354e5268af431f"
+
+[[package]]
+name = "matrixmultiply"
+version = "0.3.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a06de3016e9fae57a36fd14dba131fccf49f74b40b7fbdb472f96e361ec71a08"
+dependencies = [
+ "autocfg",
+ "rawpointer",
+]
+
+[[package]]
+name = "ndarray"
+version = "0.15.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "adb12d4e967ec485a5f71c6311fe28158e9d6f4bc4a447b474184d0f91a8fa32"
+dependencies = [
+ "matrixmultiply",
+ "num-complex",
+ "num-integer",
+ "num-traits",
+ "rawpointer",
+ "rayon",
+]
+
+[[package]]
+name = "num-complex"
+version = "0.4.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "73f88a1307638156682bada9d7604135552957b7818057dcef22705b4d509495"
+dependencies = [
+ "num-traits",
+]
+
+[[package]]
+name = "num-integer"
+version = "0.1.46"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7969661fd2958a5cb096e56c8e1ad0444ac2bbcd0061bd28660485a44879858f"
+dependencies = [
+ "num-traits",
+]
+
+[[package]]
+name = "num-traits"
+version = "0.2.19"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "071dfc062690e90b734c0b2273ce72ad0ffa95f0c74596bc250dcfd960262841"
+dependencies = [
+ "autocfg",
+]
+
+[[package]]
+name = "ppv-lite86"
+version = "0.2.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "85eae3c4ed2f50dcfe72643da4befc30deadb458a9b590d720cde2f2b1e97da9"
+dependencies = [
+ "zerocopy",
+]
+
+[[package]]
+name = "proc-macro2"
+version = "1.0.105"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "535d180e0ecab6268a3e718bb9fd44db66bbbc256257165fc699dadf70d16fe7"
+dependencies = [
+ "unicode-ident",
+]
+
+[[package]]
+name = "quote"
+version = "1.0.43"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "dc74d9a594b72ae6656596548f56f667211f8a97b3d4c3d467150794690dc40a"
+dependencies = [
+ "proc-macro2",
+]
+
+[[package]]
+name = "rand"
+version = "0.8.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "34af8d1a0e25924bc5b7c43c079c942339d8f0a8b57c39049bef581b46327404"
+dependencies = [
+ "libc",
+ "rand_chacha",
+ "rand_core",
+]
+
+[[package]]
+name = "rand_chacha"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e6c10a63a0fa32252be49d21e7709d4d4baf8d231c2dbce1eaa8141b9b127d88"
+dependencies = [
+ "ppv-lite86",
+ "rand_core",
+]
+
+[[package]]
+name = "rand_core"
+version = "0.6.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ec0be4795e2f6a28069bec0b5ff3e2ac9bafc99e6a9a7dc3547996c5c816922c"
+dependencies = [
+ "getrandom",
+]
+
+[[package]]
+name = "rawpointer"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "60a357793950651c4ed0f3f52338f53b2f809f32d83a07f72909fa13e4c6c1e3"
+
+[[package]]
+name = "rayon"
+version = "1.11.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "368f01d005bf8fd9b1206fb6fa653e6c4a81ceb1466406b81792d87c5677a58f"
+dependencies = [
+ "either",
+ "rayon-core",
+]
+
+[[package]]
+name = "rayon-core"
+version = "1.13.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "22e18b0f0062d30d4230b2e85ff77fdfe4326feb054b9783a3460d8435c8ab91"
+dependencies = [
+ "crossbeam-deque",
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "syn"
+version = "2.0.114"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d4d107df263a3013ef9b1879b0df87d706ff80f65a86ea879bd9c31f9b307c2a"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
+]
+
+[[package]]
+name = "unicode-ident"
+version = "1.0.22"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9312f7c4f6ff9069b165498234ce8be658059c6728633667c526e27dc2cf1df5"
+
+[[package]]
+name = "wasi"
+version = "0.11.1+wasi-snapshot-preview1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ccf3ec651a847eb01de73ccad15eb7d99f80485de043efb2f370cd654f4ea44b"
+
+[[package]]
+name = "zerocopy"
+version = "0.8.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1fabae64378cb18147bb18bca364e63bdbe72a0ffe4adf0addfec8aa166b2c56"
+dependencies = [
+ "zerocopy-derive",
+]
+
+[[package]]
+name = "zerocopy-derive"
+version = "0.8.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c9c2d862265a8bb4471d87e033e730f536e2a285cc7cb05dbce09a2a97075f90"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]

bit_ttt_engine-0.6.0/crates/rust_engine/Cargo.toml

@@ -0,0 +1,100 @@
+[package]
+name = "cortex_rust"
+version = "0.6.0"
+edition = "2021"
+readme = "README.md"
+
+[lib]
+crate-type = ["cdylib", "rlib"]
+
+[dependencies]
+rand = "0.8"
+libc = "0.2"
+candle-core = { version = "=0.8.4", default-features = false }
+candle-nn = { version = "=0.8.4", default-features = false }
+anyhow = "1.0"
+thiserror = "2.0"
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+
+# WASM specific dependencies
+wasm-bindgen = { version = "0.2", optional = true }
+js-sys = { version = "0.3", optional = true }
+web-sys = { version = "0.3", optional = true, features = ["console"] }
+console_error_panic_hook = { version = "0.1", optional = true }
+getrandom = { version = "0.3", optional = true, features = ["wasm_js"] }
+
+# CUDA specific dependencies
+cuda-runtime-sys = { version = "0.3.0-alpha.1", optional = true }
+
+# Optional dependencies
+pyo3 = { version = "0.24", features = ["extension-module", "macros"], optional = true }
+byteorder = "1.5.0"
+half = "2.3"
+rayon = "1.8"
+tracing = "0.1"
+tokenizers = { version = "0.22", optional = true }
+safetensors = { version = "0.5", optional = true }
+env_logger = { version = "0.11", optional = true }
+windows-sys = { version = "0.59", features = ["Win32_System_ProcessStatus", "Win32_System_Threading"], optional = true }
+reqwest = { version = "0.12", features = ["blocking"] }
+
+[build-dependencies]
+cc = "1.0"
+anyhow = "1.0"
+glob = "0.3"
+
+# Binaries with extra features (require tokenizers)
+[[bin]]
+name = "test_13b"
+path = "src/bin/test_13b.rs"
+required-features = ["tokenizers"]
+
+[[bin]]
+name = "bench_tinyllama"
+path = "src/bin/bench_tinyllama.rs"
+required-features = ["tokenizers"]
+
+[[bin]]
+name = "quick_gen"
+path = "src/bin/quick_gen.rs"
+required-features = ["tokenizers"]
+
+[[bin]]
+name = "bench_4bit_gpu"
+path = "src/bin/bench_4bit_gpu.rs"
+required-features = ["tokenizers"]
+
+[[bin]]
+name = "run_4bit_llama"
+path = "src/bin/run_4bit_llama.rs"
+required-features = ["tokenizers"]
+
+[[bin]]
+name = "test_4bit_inference"
+path = "src/bin/test_4bit_inference.rs"
+required-features = ["tokenizers"]
+
+[features]
+default = ["python", "tokenizers"]
+python = ["dep:pyo3"]
+safetensors = ["dep:safetensors"]
+tokenizers = ["dep:tokenizers"]
+dev-bins = ["dep:safetensors", "dep:env_logger", "dep:windows-sys", "tokenizers"]
+wasm = [
+    "dep:wasm-bindgen", 
+    "dep:js-sys",
+    "dep:web-sys", 
+    "dep:console_error_panic_hook",
+    "dep:getrandom"
+]
+cuda = [
+    "dep:cuda-runtime-sys",
+    "candle-core/cuda"
+]
+# Flash Attention for optimized attention computation
+flash-attention = []
+
+# WASM target-specific dependencies
+[target.'cfg(target_arch = "wasm32")'.dependencies]
+getrandom = { version = "0.2", features = ["js"] }

bit_ttt_engine-0.6.0/crates/rust_engine/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 imonoonoko
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

bit_ttt_engine-0.6.0/crates/rust_engine/README.md

@@ -0,0 +1,61 @@
+# Cortex Rust Engine - WebAssembly Support
+
+## Prerequisites
+- Rust (latest stable version)
+- wasm-pack
+- Node.js and npm
+
+## Installation
+1. Install wasm-pack:
+```bash
+cargo install wasm-pack
+```
+
+2. Add WebAssembly target:
+```bash
+rustup target add wasm32-unknown-unknown
+```
+
+## Building the WebAssembly Package
+```bash
+# Build WASM package
+wasm-pack build --target web --features wasm
+
+# Build npm package
+wasm-pack build --target npm --features wasm
+```
+
+## Usage in Browser
+```javascript
+import init, { wasm_infer, wasm_log, init_panic_hook } from './pkg/cortex_rust.js';
+
+async function run() {
+  // Initialize WASM module
+  await init();
+  
+  // Set up panic hook for better error reporting
+  init_panic_hook();
+
+  // Basic inference
+  try {
+    const result = wasm_infer("Your input text here");
+    console.log(result);
+  } catch (error) {
+    console.error("Inference error:", error);
+  }
+
+  // Optional logging
+  wasm_log("WebAssembly module initialized");
+}
+
+run();
+```
+
+## Limitations
+- No CUDA support (CPU-only)
+- Limited file I/O capabilities
+- Memory constraints inherent to WebAssembly
+
+## Notes
+- Requires modern browser with WebAssembly support
+- Performance may differ from native Rust execution