EncoderProvider.php
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | namespace Yethee\Tiktoken; |
| 6 | |
| 7 | use InvalidArgumentException; |
| 8 | use Symfony\Contracts\Service\ResetInterface; |
| 9 | use Yethee\Tiktoken\Vocab\Loader\DefaultVocabLoader; |
| 10 | use Yethee\Tiktoken\Vocab\Vocab; |
| 11 | use Yethee\Tiktoken\Vocab\VocabLoader; |
| 12 | |
| 13 | use function getenv; |
| 14 | use function sprintf; |
| 15 | use function str_starts_with; |
| 16 | use function sys_get_temp_dir; |
| 17 | |
| 18 | use const DIRECTORY_SEPARATOR; |
| 19 | |
| 20 | final class EncoderProvider implements ResetInterface |
| 21 | { |
| 22 | public const ENCODINGS = [ |
| 23 | 'r50k_base' => [ |
| 24 | 'vocab' => 'https://openaipublic.blob.core.windows.net/encodings/r50k_base.tiktoken', |
| 25 | 'pat' => '/\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+/u', |
| 26 | ], |
| 27 | 'p50k_base' => [ |
| 28 | 'vocab' => 'https://openaipublic.blob.core.windows.net/encodings/p50k_base.tiktoken', |
| 29 | 'pat' => '/\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+/u', |
| 30 | ], |
| 31 | 'p50k_edit' => [ |
| 32 | 'vocab' => 'https://openaipublic.blob.core.windows.net/encodings/p50k_base.tiktoken', |
| 33 | 'pat' => '/\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+/u', |
| 34 | ], |
| 35 | 'cl100k_base' => [ |
| 36 | 'vocab' => 'https://openaipublic.blob.core.windows.net/encodings/cl100k_base.tiktoken', |
| 37 | 'pat' => '/(?i:\'s|\'t|\'re|\'ve|\'m|\'ll|\'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}{1,3}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+/u', |
| 38 | ], |
| 39 | ]; |
| 40 | private const MODEL_PREFIX_TO_ENCODING = [ |
| 41 | 'gpt-4-' => 'cl100k_base', |
| 42 | 'gpt-3.5-turbo-' => 'cl100k_base', |
| 43 | ]; |
| 44 | private const MODEL_TO_ENCODING = [ |
| 45 | 'gpt-4' => 'cl100k_base', |
| 46 | 'gpt-3.5-turbo' => 'cl100k_base', |
| 47 | 'text-davinci-003' => 'p50k_base', |
| 48 | 'text-davinci-002' => 'p50k_base', |
| 49 | 'text-davinci-001' => 'r50k_base', |
| 50 | 'text-curie-001' => 'r50k_base', |
| 51 | 'text-babbage-001' => 'r50k_base', |
| 52 | 'text-ada-001' => 'r50k_base', |
| 53 | 'davinci' => 'r50k_base', |
| 54 | 'curie' => 'r50k_base', |
| 55 | 'babbage' => 'r50k_base', |
| 56 | 'ada' => 'r50k_base', |
| 57 | 'code-davinci-002' => 'p50k_base', |
| 58 | 'code-davinci-001' => 'p50k_base', |
| 59 | 'code-cushman-002' => 'p50k_base', |
| 60 | 'code-cushman-001' => 'p50k_base', |
| 61 | 'davinci-codex' => 'p50k_base', |
| 62 | 'cushman-codex' => 'p50k_base', |
| 63 | 'text-davinci-edit-001' => 'p50k_edit', |
| 64 | 'code-davinci-edit-001' => 'p50k_edit', |
| 65 | 'text-embedding-ada-002' => 'cl100k_base', |
| 66 | 'text-similarity-davinci-001' => 'r50k_base', |
| 67 | 'text-similarity-curie-001' => 'r50k_base', |
| 68 | 'text-similarity-babbage-001' => 'r50k_base', |
| 69 | 'text-similarity-ada-001' => 'r50k_base', |
| 70 | 'text-search-davinci-doc-001' => 'r50k_base', |
| 71 | 'text-search-curie-doc-001' => 'r50k_base', |
| 72 | 'text-search-babbage-doc-001' => 'r50k_base', |
| 73 | 'text-search-ada-doc-001' => 'r50k_base', |
| 74 | 'code-search-babbage-code-001' => 'r50k_base', |
| 75 | 'code-search-ada-code-001' => 'r50k_base', |
| 76 | ]; |
| 77 | |
| 78 | private VocabLoader|null $vocabLoader = null; |
| 79 | private string|null $vocabCacheDir; |
| 80 | |
| 81 | /** @var array<non-empty-string, Encoder> */ |
| 82 | private array $encoders = []; |
| 83 | |
| 84 | /** @var array<string, Vocab> */ |
| 85 | private array $vocabs = []; |
| 86 | |
| 87 | public function __construct() |
| 88 | { |
| 89 | $cacheDir = getenv('TIKTOKEN_CACHE_DIR'); |
| 90 | |
| 91 | if ($cacheDir === false) { |
| 92 | $cacheDir = sys_get_temp_dir() . DIRECTORY_SEPARATOR . 'tiktoken'; |
| 93 | } |
| 94 | |
| 95 | $this->vocabCacheDir = $cacheDir !== '' ? $cacheDir : null; |
| 96 | } |
| 97 | |
| 98 | /** @param non-empty-string $model */ |
| 99 | public function getForModel(string $model): Encoder |
| 100 | { |
| 101 | if (isset(self::MODEL_TO_ENCODING[$model])) { |
| 102 | return $this->get(self::MODEL_TO_ENCODING[$model]); |
| 103 | } |
| 104 | |
| 105 | foreach (self::MODEL_PREFIX_TO_ENCODING as $prefix => $modelEncoding) { |
| 106 | if (str_starts_with($model, $prefix)) { |
| 107 | return $this->get($modelEncoding); |
| 108 | } |
| 109 | } |
| 110 | |
| 111 | throw new InvalidArgumentException(sprintf('Unknown model name: %s', $model)); |
| 112 | } |
| 113 | |
| 114 | /** @param non-empty-string $encodingName */ |
| 115 | public function get(string $encodingName): Encoder |
| 116 | { |
| 117 | if (! isset(self::ENCODINGS[$encodingName])) { |
| 118 | throw new InvalidArgumentException(sprintf('Unknown encoding: %s', $encodingName)); |
| 119 | } |
| 120 | |
| 121 | if (! isset($this->encoders[$encodingName])) { |
| 122 | $options = self::ENCODINGS[$encodingName]; |
| 123 | |
| 124 | return $this->encoders[$encodingName] = new Encoder( |
| 125 | $encodingName, |
| 126 | $this->getVocab($encodingName), |
| 127 | $options['pat'], |
| 128 | ); |
| 129 | } |
| 130 | |
| 131 | return $this->encoders[$encodingName]; |
| 132 | } |
| 133 | |
| 134 | /** @param non-empty-string|null $cacheDir */ |
| 135 | public function setVocabCache(string|null $cacheDir): void |
| 136 | { |
| 137 | $this->vocabCacheDir = $cacheDir; |
| 138 | $this->vocabLoader = null; |
| 139 | } |
| 140 | |
| 141 | /** @psalm-api */ |
| 142 | public function setVocabLoader(VocabLoader $loader): void |
| 143 | { |
| 144 | $this->vocabLoader = $loader; |
| 145 | } |
| 146 | |
| 147 | public function reset(): void |
| 148 | { |
| 149 | $this->encoders = []; |
| 150 | $this->vocabs = []; |
| 151 | } |
| 152 | |
| 153 | private function getVocab(string $encodingName): Vocab |
| 154 | { |
| 155 | if (isset($this->vocabs[$encodingName])) { |
| 156 | return $this->vocabs[$encodingName]; |
| 157 | } |
| 158 | |
| 159 | $loader = $this->vocabLoader; |
| 160 | |
| 161 | if ($loader === null) { |
| 162 | $loader = $this->vocabLoader = new DefaultVocabLoader($this->vocabCacheDir); |
| 163 | } |
| 164 | |
| 165 | return $this->vocabs[$encodingName] = $loader->load(self::ENCODINGS[$encodingName]['vocab']); |
| 166 | } |
| 167 | } |
| 168 |