root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-29 04:44:34 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

convert.py : remove GGML vocab + other obsolete stuff

Browse Source
This commit is contained in:
Georgi Gerganov 2023-07-27 16:36:35 +03:00
parent 68f53485e4
commit d2bb3ac10b
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 17 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
convert-new.py
View File

@ -278,19 +278,7 @@ class SentencePieceVocab:
return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>" return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
class GGMLVocab: Vocab = Union[SentencePieceVocab]
def __init__(self, tokens: List[Tuple[bytes, float]]):
self.tokens = tokens
self.vocab_size = len(tokens)
def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
return self.tokens
def __repr__(self) -> str:
return f"<GGMLVocab with {self.vocab_size} tokens>"
Vocab = Union[SentencePieceVocab, GGMLVocab]
def permute(weights: NDArray, n_head: int) -> NDArray: def permute(weights: NDArray, n_head: int) -> NDArray:
@ -691,7 +679,6 @@ def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], conc
def check_vocab_size(params: Params, vocab: Vocab) -> None: def check_vocab_size(params: Params, vocab: Vocab) -> None:
if params.n_vocab != vocab.vocab_size: if params.n_vocab != vocab.vocab_size:
# GGMLVocab comes from the same file as the model so shouldn't mismatch:
assert isinstance(vocab, SentencePieceVocab) assert isinstance(vocab, SentencePieceVocab)
if params.n_vocab == vocab.vocab_size_base: if params.n_vocab == vocab.vocab_size_base:
print("Ignoring added_tokens.json since model matches vocab size without it.") print("Ignoring added_tokens.json since model matches vocab size without it.")
@ -874,7 +861,7 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
if path.is_dir(): if path.is_dir():
vocab_file = "tokenizer.model" vocab_file = "tokenizer.model"
if vocabtype == 'bpe': if vocabtype == 'bpe':
vocab_file = "vocab.json" vocab_file = "vocab.json"
path2 = path / vocab_file path2 = path / vocab_file
# Use `.parent` instead of /.. to handle the symlink case better. # Use `.parent` instead of /.. to handle the symlink case better.
path3 = path.parent / vocab_file path3 = path.parent / vocab_file
@ -916,15 +903,14 @@ def do_dump_model(model_plus: ModelPlus) -> None:
def main(args_in: Optional[List[str]] = None) -> None: def main(args_in: Optional[List[str]] = None) -> None:
parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file") parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model") parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model")
parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file") parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file")
parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab") parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab")
parser.add_argument("--outtype", choices=["f32", "f16", "q4_1", "q4_0"], help="output format (default: based on input)") parser.add_argument("--outtype", choices=["f32", "f16"], help="output format (default: based on input)")
parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file") parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file")
parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input") parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input")
parser.add_argument("model", type=Path, parser.add_argument("model", type=Path, help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)") parser.add_argument("--vocabtype", choices=["spm", "bpe"], help="vocab format (default: spm)")
parser.add_argument("--vocabtype", default='spm', choices=["spm", "bpe"], help="vocab format (default: spm)")
args = parser.parse_args(args_in) args = parser.parse_args(args_in)
vocab: Vocab vocab: Vocab
@ -947,12 +933,14 @@ def main(args_in: Optional[List[str]] = None) -> None:
else: else:
vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
vocab = load_vocab(vocab_dir, args.vocabtype) vocab = load_vocab(vocab_dir, args.vocabtype)
params = Params.load(model_plus)
model = model_plus.model params = Params.load(model_plus)
model = do_necessary_conversions(model, params) model = model_plus.model
model = do_necessary_conversions(model, params)
output_type = pick_output_type(model, args.outtype) output_type = pick_output_type(model, args.outtype)
model = convert_to_output_type(model, output_type) model = convert_to_output_type(model, output_type)
outfile = args.outfile or default_outfile(model_plus.paths, output_type) outfile = args.outfile or default_outfile(model_plus.paths, output_type)
OutputFile.write_all(outfile, params, output_type, model, vocab) OutputFile.write_all(outfile, params, output_type, model, vocab)
print(f"Wrote {outfile}") print(f"Wrote {outfile}")