Skip to content

allow HF model download, allow disabling tqdm bars #1059

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
kylrth wants to merge 2 commits into
base: master
Choose a base branch
from
Open

allow HF model download, allow disabling tqdm bars #1059

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
12ff1f4
allow HF model download for layerwise models
kylrth Aug 15, 2024
9b0d7d6
setting to disable tqdm bars
kylrth Aug 15, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 12 additions & 11 deletions FlagEmbedding/flag_reranker.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -251,7 +251,7 @@ def __init__(

@torch.no_grad()
def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str, str]], batch_size: int = 256,
max_length: int = 512, normalize: bool = False) -> List[float]:
max_length: int = 512, normalize: bool = False, progress: bool = None) -> List[float]:
if self.num_gpus > 0:
batch_size = batch_size * self.num_gpus

Expand All @@ -260,8 +260,8 @@ def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str,
sentence_pairs = [sentence_pairs]

all_scores = []
for start_index in tqdm(range(0, len(sentence_pairs), batch_size), desc="Compute Scores",
disable=len(sentence_pairs) < 128):
disable = not progress if isinstance(progress, bool) else len(sentence_pairs) <= batch_size
for start_index in tqdm(range(0, len(sentence_pairs), batch_size), desc="Compute Scores", disable=disable):
sentences_batch = sentence_pairs[start_index:start_index + batch_size]
inputs = self.tokenizer(
sentences_batch,
Expand Down Expand Up @@ -331,7 +331,7 @@ def __init__(
@torch.no_grad()
def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str, str]], batch_size: int = 16,
max_length: int = 512, prompt: str = None, normalize: bool = False,
use_dataloader: bool = False, num_workers: int = None) -> List[float]:
use_dataloader: bool = False, num_workers: int = None, progress: bool = None) -> List[float]:
assert isinstance(sentence_pairs, list)
if isinstance(sentence_pairs[0], str):
sentence_pairs = [sentence_pairs]
Expand Down Expand Up @@ -373,7 +373,8 @@ def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str,
return_tensors=None,
add_special_tokens=False)['input_ids']
encode_max_length = max_length + len(sep_inputs) + len(prompt_inputs)
for batch_start in trange(0, len(sentences_sorted), batch_size):
disable = not progress if isinstance(progress, bool) else len(sentence_pairs) <= batch_size
for batch_start in trange(0, len(sentences_sorted), batch_size, disable=disable):
batch_sentences = sentences_sorted[batch_start:batch_start + batch_size]
batch_sentences = [(f'A: {q}', f'B: {p}') for q,p in batch_sentences]
queries = [s[0] for s in batch_sentences]
Expand Down Expand Up @@ -469,7 +470,6 @@ def __init__(
self.model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
cache_dir=cache_dir,
trust_remote_code=True,
local_files_only=True,
torch_dtype=torch.bfloat16 if use_bf16 else torch.float32)
if peft_path:
self.model = PeftModel.from_pretrained(self.model,peft_path)
Expand Down Expand Up @@ -508,7 +508,7 @@ def __init__(
def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str, str]], batch_size: int = 16,
max_length: int = 512, cutoff_layers: List[int] = None, prompt: str = None,
normalize: bool = False, use_dataloader: bool = False,
num_workers: int = None) -> Union[float, List[float], List[List[float]]]:
num_workers: int = None, progress: bool = None) -> Union[float, List[float], List[List[float]]]:
assert isinstance(sentence_pairs, list)
if isinstance(sentence_pairs[0], str):
sentence_pairs = [sentence_pairs]
Expand Down Expand Up @@ -558,7 +558,8 @@ def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str,
return_tensors=None,
add_special_tokens=False)['input_ids']
encode_max_length = max_length + len(sep_inputs) + len(prompt_inputs)
for batch_start in trange(0, len(sentences_sorted), batch_size):
disable = not progress if isinstance(progress, bool) else len(sentence_pairs) <= batch_size
for batch_start in trange(0, len(sentences_sorted), batch_size, disable=disable):
batch_sentences = sentences_sorted[batch_start:batch_start + batch_size]
batch_sentences = [(f'A: {q}', f'B: {p}') for q, p in batch_sentences]
queries = [s[0] for s in batch_sentences]
Expand Down Expand Up @@ -661,7 +662,6 @@ def __init__(
self.model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
cache_dir=cache_dir,
trust_remote_code=True,
local_files_only=True,
torch_dtype=torch.bfloat16 if use_bf16 else torch.float32)
if peft_path:
self.model = PeftModel.from_pretrained(self.model,peft_path)
Expand Down Expand Up @@ -700,7 +700,7 @@ def __init__(
def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str, str]], batch_size: int = 16,
max_length: int = 512,
cutoff_layers: List[int] = None, compress_layer: List[int] = [8], compress_ratio: int = 1,
prompt: str = None, normalize: bool = False) -> Union[float, List[float], List[List[float]]]:
prompt: str = None, normalize: bool = False, progress: bool = None) -> Union[float, List[float], List[List[float]]]:
assert isinstance(sentence_pairs, list)
if isinstance(sentence_pairs[0], str):
sentence_pairs = [sentence_pairs]
Expand All @@ -719,7 +719,8 @@ def compute_score(self, sentence_pairs: Union[List[Tuple[str, str]], Tuple[str,
add_special_tokens=False)['input_ids']
encode_max_length = max_length + len(sep_inputs) + len(prompt_inputs)
all_scores = []
for batch_start in trange(0, len(sentences_sorted), batch_size):
disable = not progress if isinstance(progress, bool) else len(sentence_pairs) <= batch_size
for batch_start in trange(0, len(sentences_sorted), batch_size, disable=disable):
batch_sentences = sentences_sorted[batch_start:batch_start + batch_size]
batch_sentences = [(f'A: {q}', f'B: {p}') for q, p in batch_sentences]
queries = [s[0] for s in batch_sentences]
Expand Down