@small-ltsc/sdk 0.1.0 → 0.2.4

package/README.md

@@ -1,18 +1,21 @@
 # @small-ltsc/sdk
 
+[![npm](https://img.shields.io/npm/v/@small-ltsc/sdk)](https://www.npmjs.com/package/@small-ltsc/sdk)
+[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](../../LICENSE)
+
 TypeScript SDK for **Small LTSC** - Lossless Token Sequence Compression for LLMs.
 
-Reduce LLM inference costs by compressing repetitive token patterns in prompts while maintaining perfect reconstruction. The compressed format can be understood by fine-tuned models, enabling 30-50% cost reduction on structured inputs.
+Reduce LLM inference costs by compressing repetitive token patterns in prompts while maintaining perfect reconstruction. Achieve 30-60% compression on structured inputs with a format that fine-tuned models can understand.
 
 ## Features
 
-- **Lossless compression** - Perfect round-trip reconstruction
-- **High performance** - Rust/WASM core with O(n log n) algorithms
+- **Lossless compression** - Perfect round-trip reconstruction guaranteed
+- **High performance** - Rust/WASM core with O(n log n) suffix array algorithms
 - **Cross-platform** - Works in browsers, Node.js, Deno, and edge runtimes
-- **Streaming support** - Handle inputs of any size
+- **Streaming support** - Handle inputs of any size with constant memory
 - **Worker threads** - Non-blocking compression for large inputs
-- **Static dictionaries** - Pre-built patterns for common domains
-- **TypeScript-first** - Full type safety and IntelliSense
+- **Static dictionaries** - Pre-built patterns for Python, TypeScript, SQL, and more
+- **TypeScript-first** - Full type safety and IntelliSense support
 
 ## Installation
 
@@ -25,7 +28,7 @@ npm install @small-ltsc/sdk
 ```typescript
 import { compress, decompress, initWasm } from '@small-ltsc/sdk';
 
-// Initialize WASM (required once)
+// Initialize WASM module (required once)
 await initWasm();
 
 // Compress tokens
@@ -33,9 +36,9 @@ const tokens = [1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3];
 const result = await compress(tokens);
 
 console.log(`Compressed: ${result.originalLength} → ${result.compressedLength} tokens`);
-console.log(`Ratio: ${(result.compressionRatio * 100).toFixed(1)}%`);
+console.log(`Savings: ${((1 - result.compressionRatio) * 100).toFixed(1)}%`);
 
-// Decompress
+// Decompress (lossless)
 const restored = await decompress(result.serializedTokens);
 console.assert(JSON.stringify(tokens) === JSON.stringify(restored));
 ```
@@ -45,24 +48,24 @@ console.assert(JSON.stringify(tokens) === JSON.stringify(restored));
 ```typescript
 const result = await compress(tokens, {
   // Pattern discovery
-  minSubsequenceLength: 2,    // Min pattern length (default: 2)
-  maxSubsequenceLength: 8,    // Max pattern length (default: 8)
+  minSubsequenceLength: 2,    // Minimum pattern length (default: 2)
+  maxSubsequenceLength: 8,    // Maximum pattern length (default: 8)
   
   // Selection algorithm
   selectionMode: 'greedy',    // 'greedy' | 'optimal' | 'beam'
   
   // Hierarchical compression
   hierarchicalEnabled: true,  // Allow patterns of patterns
-  hierarchicalMaxDepth: 3,    // Max nesting depth
+  hierarchicalMaxDepth: 3,    // Maximum nesting depth
   
   // Verification
-  verify: true,               // Round-trip verification
+  verify: true,               // Enable round-trip verification
 });
 ```
 
 ## Static Dictionaries
 
-Use pre-built dictionaries for better compression:
+Use pre-built dictionaries for domain-specific compression:
 
 ```typescript
 const result = await compress(pythonCodeTokens, {
@@ -70,11 +73,11 @@ const result = await compress(pythonCodeTokens, {
 });
 ```
 
-Available: `python-v1`, `typescript-v1`, `markdown-v1`, `json-v1`, `sql-v1`
+Available dictionaries: `python-v1`, `typescript-v1`, `markdown-v1`, `json-v1`, `sql-v1`
 
 ## Streaming
 
-For large inputs:
+For large inputs that exceed memory constraints:
 
 ```typescript
 import { createStreamingCompressor } from '@small-ltsc/sdk';
@@ -90,7 +93,7 @@ const result = await compressor.finish();
 
 ## Worker Threads
 
-Non-blocking compression:
+Non-blocking compression for UI responsiveness:
 
 ```typescript
 import { createWorkerPool } from '@small-ltsc/sdk';
@@ -105,45 +108,58 @@ pool.terminate();
 ```html
 <script type="module">
   import { compress, initWasm } from 'https://esm.sh/@small-ltsc/sdk';
+  
   await initWasm();
   const result = await compress([1, 2, 3, 1, 2, 3]);
+  console.log('Compression ratio:', result.compressionRatio);
 </script>
 ```
 
-## API
+## API Reference
 
 ### Core Functions
 
-- `compress(tokens, config?)` - Compress token sequence
-- `decompress(tokens, config?)` - Decompress to original tokens
-- `discoverPatterns(tokens, minLen?, maxLen?)` - Find patterns without compressing
+| Function | Description |
+|----------|-------------|
+| `compress(tokens, config?)` | Compress a token sequence |
+| `decompress(tokens, config?)` | Decompress to original tokens |
+| `discoverPatterns(tokens, minLen?, maxLen?)` | Find patterns without compressing |
 
 ### Streaming
 
-- `createStreamingCompressor(config?)` - Create streaming compressor
-- `compressStream(asyncIterable, config?)` - Compress async stream
+| Function | Description |
+|----------|-------------|
+| `createStreamingCompressor(config?)` | Create a streaming compressor instance |
+| `compressStream(asyncIterable, config?)` | Compress an async iterable stream |
 
 ### Workers
 
-- `createWorkerPool(count?)` - Create worker pool
-- `compressInWorker(tokens, config?)` - Single-use worker compression
+| Function | Description |
+|----------|-------------|
+| `createWorkerPool(count?)` | Create a pool of worker threads |
+| `compressInWorker(tokens, config?)` | Single-use worker compression |
 
 ### Dictionaries
 
-- `loadStaticDictionary(id)` - Load built-in dictionary
-- `createStaticDictionary(id, patterns)` - Create custom dictionary
+| Function | Description |
+|----------|-------------|
+| `loadStaticDictionary(id)` | Load a built-in dictionary |
+| `createStaticDictionary(id, patterns)` | Create a custom dictionary |
 
 ### Utilities
 
-- `initWasm()` - Initialize WASM module
-- `isWasmInitialized()` - Check initialization status
-- `extractDictionary(tokens)` - Get dictionary from compressed tokens
-- `isCompressed(tokens)` - Check if tokens are compressed
+| Function | Description |
+|----------|-------------|
+| `initWasm()` | Initialize the WASM module |
+| `isWasmInitialized()` | Check initialization status |
+| `extractDictionary(tokens)` | Extract dictionary from compressed tokens |
+| `isCompressed(tokens)` | Check if tokens are in compressed format |
 
 ## Documentation
 
 - [Quick Start Guide](./docs/QUICKSTART.md)
 - [API Reference](./docs/API.md)
+- [Main Repository](https://github.com/triage-sec/small)
 
 ## Optional ML Features
 
@@ -154,16 +170,24 @@ npm install @small-ltsc/ml
 ```
 
 ```typescript
-import { HeuristicQualityPredictor, detectRegions } from '@small-ltsc/ml';
+import { HeuristicQualityPredictor } from '@small-ltsc/ml';
 
 const predictor = new HeuristicQualityPredictor();
 const prediction = await predictor.predict(compressionResult);
 
 if (!prediction.acceptable) {
-  // Retry with more conservative settings
+  console.log(`Recommendation: ${prediction.recommendation}`);
 }
 ```
 
 ## License
 
-MIT
+MIT License - see [LICENSE](../../LICENSE) for details.
+
+## Contributors
+
+Built by [Triage Sec](https://triage-sec.com) - an applied team of researchers and engineers working towards building resiliency for AI systems.
+
+- Nikhil Srivastava (University of California, Berkeley)
+- Omansh Bainsla (Georgia Tech)
+- Sahil Chatiwala (Georgia Tech)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@small-ltsc/sdk",
-  "version": "0.1.0",
+"version": "0.2.4",
   "description": "TypeScript SDK for Small LTSC - Lossless Token Sequence Compression",
   "type": "module",
   "main": "./dist/esm/index.js",
@@ -40,7 +40,12 @@
     "wasm",
     "lossless"
   ],
-  "author": "",
+  "author": "Triage Sec <nicks@triage-sec.com>",
+  "contributors": [
+    "Nikhil Srivastava",
+    "Omansh Bainsla",
+    "Sahil Chatiwala"
+  ],
   "license": "MIT",
   "repository": {
     "type": "git",