From 37af613dfcad659b6600b7657b3e114b98de7501 Mon Sep 17 00:00:00 2001 From: goerch Date: Sun, 1 Oct 2023 10:44:08 +0200 Subject: [PATCH] Remove unused code --- convert-falcon-hf-to-gguf.py | 24 ------------------------ convert.py | 2 -- 2 files changed, 26 deletions(-) diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py index 52fc77db3..e479b9524 100755 --- a/convert-falcon-hf-to-gguf.py +++ b/convert-falcon-hf-to-gguf.py @@ -20,28 +20,6 @@ if 'NO_LOCAL_GGUF' not in os.environ: import gguf -def bytes_to_unicode(): - # ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py - """ - Returns list of utf-8 byte and a corresponding list of unicode strings. - The reversible bpe codes work on unicode strings. - This means you need a large # of unicode characters in your vocab if you want to avoid UNKs. - When you're at something like a 10B token dataset you end up needing around 5K for decent coverage. - This is a significant percentage of your normal, say, 32K bpe vocab. - To avoid that, we want lookup tables between utf-8 bytes and unicode strings. - And avoids mapping to whitespace/control characters the bpe code barfs on. - """ - bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1)) - cs = bs[:] - n = 0 - for b in range(2**8): - if b not in bs: - bs.append(b) - cs.append(2**8+n) - n += 1 - return dict(zip(bs, (chr(n) for n in cs))) - - def count_model_parts(dir_model: Path) -> int: num_parts = 0 for filename in os.listdir(dir_model): @@ -155,8 +133,6 @@ vocab_size = hparams["vocab_size"] if "vocab_size" in hparams else len(tokenizer tokenizer = AutoTokenizer.from_pretrained(dir_model) reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.vocab.items()} -byte_encoder = bytes_to_unicode() -byte_decoder = {v: k for k, v in byte_encoder.items()} for i in range(vocab_size): tokens.append(reverse_vocab[i]) diff --git a/convert.py b/convert.py index e76fb2078..6131ce7cd 100755 --- a/convert.py +++ b/convert.py @@ -340,8 +340,6 @@ class BpeVocab: tokenizer = self.bpe_tokenizer from transformers.models.gpt2 import tokenization_gpt2 # type: ignore[import] reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.items()} - byte_encoder = tokenization_gpt2.bytes_to_unicode() - byte_decoder = {v: k for k, v in byte_encoder.items()} for i, _ in enumerate(tokenizer): yield reverse_vocab[i], 0.0, gguf.TokenType.NORMAL