splintr 0.10.0

Fast Rust tokenizer (BPE + SentencePiece + WordPiece) with Python bindings
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
#!/usr/bin/env python3
"""
Benchmark: Single Text Encoding Comparison
Compares tokenizer throughput across different text types and sizes.

Usage:
    python benchmarks/benchmark_single.py
"""

import gc
import statistics
import time
from dataclasses import dataclass
from pathlib import Path

import matplotlib.pyplot as plt
import numpy as np


# Sample texts for benchmarking
SAMPLE_TEXTS = {
    "short": "Hello, world! This is a test.",
    "medium": """The quick brown fox jumps over the lazy dog.
    Machine learning models require tokenization to process text efficiently.
    Tokenizers convert text into numerical representations that models can understand.""" * 10,
    "long": """Artificial intelligence and machine learning have revolutionized
    the way we process and understand natural language. Large language models (LLMs)
    like GPT-4, Claude, and others rely heavily on efficient tokenization to handle
    vast amounts of text data. The performance of tokenizers directly impacts the
    overall throughput of these systems, making optimization crucial for production
    deployments. BPE (Byte Pair Encoding) has become the de facto standard for
    modern tokenizers due to its balance of vocabulary efficiency and handling of
    out-of-vocabulary words.""" * 50,
    "code": '''
def fibonacci(n: int) -> int:
    """Calculate the nth Fibonacci number."""
    if n <= 1:
        return n
    return fibonacci(n - 1) + fibonacci(n - 2)

class TokenizerBenchmark:
    def __init__(self, name: str):
        self.name = name
        self.results = []

    def run(self, text: str, iterations: int = 100):
        for _ in range(iterations):
            tokens = self.encode(text)
            self.results.append(len(tokens))
''' * 20,
    "multilingual": """
    English: The quick brown fox jumps over the lazy dog.
    中文: 快速的棕色狐狸跳过懒狗。
    日本語: 素早い茶色の狐が怠惰な犬を飛び越える。
    한국어: 빠른 갈색 여우가 게으른 개를 뛰어넘습니다.
    العربية: الثعلب البني السريع يقفز فوق الكلب الكسول.
    Русский: Быстрая коричневая лиса прыгает через ленивую собаку.
    """ * 20,
}

TOKENIZER_COLORS = {
    "splintr": "#2ecc71",      # Green
    "tiktoken": "#3498db",     # Blue
    "huggingface": "#e74c3c",  # Red
    "tokendagger": "#9b59b6",  # Purple
}


@dataclass
class BenchmarkResult:
    name: str
    text_type: str
    bytes_per_second: float
    tokens_per_second: float
    num_tokens: int
    num_bytes: int
    latency_ms: float
    latency_std_ms: float


def benchmark_encode(
    name: str,
    encode_fn,
    text: str,
    text_type: str,
    warmup: int = 50,
    iterations: int = 100,
) -> BenchmarkResult:
    """Benchmark a single encode function."""
    num_bytes = len(text.encode("utf-8"))

    # Warmup - use more iterations to ensure thread pools are initialized
    for _ in range(warmup):
        encode_fn(text)

    # Force garbage collection before timing
    gc.collect()

    # Benchmark
    times = []
    num_tokens = 0
    for _ in range(iterations):
        start = time.perf_counter_ns()
        tokens = encode_fn(text)
        end = time.perf_counter_ns()
        times.append((end - start) / 1e9)  # Convert to seconds
        num_tokens = len(tokens)

    avg_time = statistics.mean(times)
    std_time = statistics.stdev(times) if len(times) > 1 else 0
    bytes_per_second = num_bytes / avg_time
    tokens_per_second = num_tokens / avg_time

    return BenchmarkResult(
        name=name,
        text_type=text_type,
        bytes_per_second=bytes_per_second,
        tokens_per_second=tokens_per_second,
        num_tokens=num_tokens,
        num_bytes=num_bytes,
        latency_ms=avg_time * 1000,
        latency_std_ms=std_time * 1000,
    )


def load_tokenizers():
    """Load all available tokenizers."""
    tokenizers = {}

    # splintr
    try:
        import splintr
        enc = splintr.Tokenizer.from_pretrained("cl100k_base")
        tokenizers["splintr"] = enc.encode
        print("Loaded: splintr")
    except ImportError:
        print("splintr not available")

    # tiktoken
    try:
        import tiktoken
        enc = tiktoken.get_encoding("cl100k_base")
        tokenizers["tiktoken"] = enc.encode
        print("Loaded: tiktoken")
    except ImportError:
        print("tiktoken not available")

    # HuggingFace tokenizers
    try:
        from tokenizers import Tokenizer as HFTokenizer
        hf_enc = HFTokenizer.from_pretrained("gpt2")

        def hf_encode(text):
            return hf_enc.encode(text).ids

        tokenizers["huggingface"] = hf_encode
        print("Loaded: huggingface")
    except ImportError:
        print("huggingface not available")

    # TokenDagger
    try:
        import tokendagger
        import tiktoken
        tik_enc = tiktoken.get_encoding("cl100k_base")
        enc = tokendagger.Tokenizer(
            name="cl100k_base",
            pat_str=tik_enc._pat_str,
            mergeable_ranks=tik_enc._mergeable_ranks,
            special_tokens=tik_enc._special_tokens,
        )
        tokenizers["tokendagger"] = enc.encode
        print("Loaded: tokendagger")
    except (ImportError, Exception) as e:
        print(f"tokendagger not available: {e}")

    return tokenizers


def run_benchmarks(tokenizers: dict) -> list[BenchmarkResult]:
    """Run benchmarks for all text types."""
    results = []

    # Global warmup to initialize thread pools
    print("\nWarming up all tokenizers...")
    warmup_text = "This is a warmup text to initialize thread pools and caches." * 10
    for name, encode_fn in tokenizers.items():
        for _ in range(100):
            encode_fn(warmup_text)
    print("Warmup complete.")

    print("\n" + "=" * 70)
    print("TEXT TYPE BENCHMARKS")
    print("=" * 70)

    for text_type, text in SAMPLE_TEXTS.items():
        num_bytes = len(text.encode("utf-8"))
        print(f"\n--- {text_type.upper()} ({num_bytes:,} bytes) ---")
        print(f"{'Tokenizer':<15} {'MB/s':>10} {'Ktok/s':>10} {'Latency':>12} {'Std':>10}")
        print("-" * 60)

        for name, encode_fn in tokenizers.items():
            result = benchmark_encode(name, encode_fn, text, text_type)
            results.append(result)
            print(
                f"{name:<15} {result.bytes_per_second / 1e6:>10.2f} "
                f"{result.tokens_per_second / 1e3:>10.2f} "
                f"{result.latency_ms:>10.3f} ms "
                f"{result.latency_std_ms:>8.3f} ms"
            )

    return results


def generate_chart(results: list[BenchmarkResult], output_path: str):
    """Generate text type comparison chart."""

    # Get unique tokenizers and text types
    names = list(dict.fromkeys(r.name for r in results))
    text_types = list(dict.fromkeys(r.text_type for r in results))

    # Create figure
    fig, ax = plt.subplots(figsize=(12, 7))

    x = np.arange(len(text_types))
    width = 0.8 / len(names)

    # Create bars for each tokenizer
    for i, name in enumerate(names):
        throughputs = []
        for text_type in text_types:
            for r in results:
                if r.name == name and r.text_type == text_type:
                    throughputs.append(r.bytes_per_second / 1e6)
                    break

        offset = i * width - width * len(names) / 2 + width / 2
        bars = ax.bar(
            x + offset,
            throughputs,
            width,
            label=name,
            color=TOKENIZER_COLORS.get(name, "#95a5a6"),
        )

        # Add value labels on bars
        for bar, val in zip(bars, throughputs):
            height = bar.get_height()
            ax.annotate(
                f'{val:.1f}',
                xy=(bar.get_x() + bar.get_width() / 2, height),
                xytext=(0, 3),
                textcoords="offset points",
                ha='center',
                va='bottom',
                fontsize=8,
            )

    # Add text size annotations below x-axis
    text_sizes = []
    for text_type in text_types:
        for r in results:
            if r.text_type == text_type:
                text_sizes.append(r.num_bytes)
                break

    ax.set_xlabel("Text Type", fontsize=12)
    ax.set_ylabel("Throughput (MB/s)", fontsize=12)
    ax.set_title("Tokenizer Throughput by Text Type", fontsize=14, fontweight="bold")
    ax.set_xticks(x)

    # Create x-tick labels with size info
    xlabels = [f"{t.capitalize()}\n({text_sizes[i]:,} bytes)" for i, t in enumerate(text_types)]
    ax.set_xticklabels(xlabels)

    ax.legend(loc="upper left")
    ax.grid(axis="y", alpha=0.3)

    # Add a note about short text performance
    ax.text(
        0.98, 0.02,
        "Lower is worse for short texts due to fixed overhead",
        transform=ax.transAxes,
        fontsize=9,
        ha='right',
        va='bottom',
        style='italic',
        color='gray',
    )

    plt.tight_layout()
    plt.savefig(output_path, dpi=150, bbox_inches="tight")
    print(f"\nChart saved to: {output_path}")
    plt.close()


def generate_latency_chart(results: list[BenchmarkResult], output_path: str):
    """Generate latency comparison chart (good for seeing short text overhead)."""

    names = list(dict.fromkeys(r.name for r in results))
    text_types = list(dict.fromkeys(r.text_type for r in results))

    fig, ax = plt.subplots(figsize=(12, 7))

    x = np.arange(len(text_types))
    width = 0.8 / len(names)

    for i, name in enumerate(names):
        latencies = []
        errors = []
        for text_type in text_types:
            for r in results:
                if r.name == name and r.text_type == text_type:
                    latencies.append(r.latency_ms)
                    errors.append(r.latency_std_ms)
                    break

        offset = i * width - width * len(names) / 2 + width / 2
        bars = ax.bar(
            x + offset,
            latencies,
            width,
            label=name,
            color=TOKENIZER_COLORS.get(name, "#95a5a6"),
            yerr=errors,
            capsize=3,
        )

        # Add value labels
        for bar, val in zip(bars, latencies):
            height = bar.get_height()
            ax.annotate(
                f'{val:.2f}',
                xy=(bar.get_x() + bar.get_width() / 2, height),
                xytext=(0, 3),
                textcoords="offset points",
                ha='center',
                va='bottom',
                fontsize=8,
            )

    ax.set_xlabel("Text Type", fontsize=12)
    ax.set_ylabel("Latency (ms) - Lower is Better", fontsize=12)
    ax.set_title("Tokenizer Latency by Text Type", fontsize=14, fontweight="bold")
    ax.set_xticks(x)

    text_sizes = []
    for text_type in text_types:
        for r in results:
            if r.text_type == text_type:
                text_sizes.append(r.num_bytes)
                break

    xlabels = [f"{t.capitalize()}\n({text_sizes[i]:,} bytes)" for i, t in enumerate(text_types)]
    ax.set_xticklabels(xlabels)

    ax.legend(loc="upper left")
    ax.grid(axis="y", alpha=0.3)
    ax.set_yscale("log")  # Log scale to see small values

    plt.tight_layout()
    plt.savefig(output_path, dpi=150, bbox_inches="tight")
    print(f"Latency chart saved to: {output_path}")
    plt.close()


def analyze_short_text_overhead(tokenizers: dict):
    """Analyze why short texts are slow - measure fixed overhead."""

    print("\n" + "=" * 70)
    print("SHORT TEXT OVERHEAD ANALYSIS")
    print("=" * 70)

    # First, do a global warmup for all tokenizers
    print("\nWarming up all tokenizers...")
    warmup_text = "This is a warmup text to initialize thread pools and caches." * 10
    for name, encode_fn in tokenizers.items():
        for _ in range(100):
            encode_fn(warmup_text)
    print("Warmup complete.\n")

    # Test with progressively shorter texts
    test_texts = [
        ("1 char", "H"),
        ("5 chars", "Hello"),
        ("10 chars", "Hello worl"),
        ("29 chars (short)", "Hello, world! This is a test."),
        ("100 chars", "Hello, world! " * 7),
        ("500 chars", "Hello, world! " * 35),
    ]

    print(f"{'Text':<20} {'Size':>8} ", end="")
    for name in tokenizers.keys():
        print(f"{name:>12}", end=" ")
    print()
    print("-" * (30 + 13 * len(tokenizers)))

    for label, text in test_texts:
        num_bytes = len(text.encode("utf-8"))
        print(f"{label:<20} {num_bytes:>6} B ", end="")

        for name, encode_fn in tokenizers.items():
            # Additional per-text warmup
            for _ in range(20):
                encode_fn(text)

            # Measure
            gc.collect()
            times = []
            for _ in range(100):
                start = time.perf_counter_ns()
                encode_fn(text)
                end = time.perf_counter_ns()
                times.append((end - start) / 1e6)  # ms

            avg_ms = statistics.mean(times)
            print(f"{avg_ms:>10.4f}ms", end=" ")
        print()


def main():
    print("=" * 70)
    print("TOKENIZER BENCHMARK: TEXT TYPES")
    print("=" * 70)

    # Create output directory
    output_dir = Path(__file__).parent / "results"
    output_dir.mkdir(exist_ok=True)

    # Load tokenizers
    tokenizers = load_tokenizers()

    if len(tokenizers) < 2:
        print("\nNeed at least 2 tokenizers for comparison")
        return

    # Run main benchmarks
    results = run_benchmarks(tokenizers)

    # Generate charts
    generate_chart(results, str(output_dir / "benchmark_single.png"))
    generate_latency_chart(results, str(output_dir / "benchmark_single_latency.png"))

    # Analyze short text overhead
    analyze_short_text_overhead(tokenizers)

    print("\nDone!")


if __name__ == "__main__":
    main()