import datetime import logging import logging.handlers import os import sys import requests from llava.constants import LOGDIR server_error_msg = "**NETWORK ERROR DUE TO HIGH TRAFFIC. PLEASE REGENERATE OR REFRESH THIS PAGE.**" moderation_msg = "YOUR INPUT VIOLATES OUR CONTENT MODERATION GUIDELINES. PLEASE TRY AGAIN." handler = None def build_logger(logger_name, logger_filename): global handler formatter = logging.Formatter( fmt="%(asctime)s | %(levelname)s | %(name)s | %(message)s", datefmt="%Y-%m-%d %H:%M:%S", ) # Set the format of root handlers if not logging.getLogger().handlers: logging.basicConfig(level=logging.INFO) logging.getLogger().handlers[0].setFormatter(formatter) # Redirect stdout and stderr to loggers stdout_logger = logging.getLogger("stdout") stdout_logger.setLevel(logging.INFO) sl = StreamToLogger(stdout_logger, logging.INFO) sys.stdout = sl stderr_logger = logging.getLogger("stderr") stderr_logger.setLevel(logging.ERROR) sl = StreamToLogger(stderr_logger, logging.ERROR) sys.stderr = sl # Get logger logger = logging.getLogger(logger_name) logger.setLevel(logging.INFO) # Add a file handler for all loggers if handler is None: os.makedirs(LOGDIR, exist_ok=True) filename = os.path.join(LOGDIR, logger_filename) handler = logging.handlers.TimedRotatingFileHandler( filename, when='D', utc=True, encoding='UTF-8') handler.setFormatter(formatter) for name, item in logging.root.manager.loggerDict.items(): if isinstance(item, logging.Logger): item.addHandler(handler) return logger class StreamToLogger(object): """ Fake file-like stream object that redirects writes to a logger instance. """ def __init__(self, logger, log_level=logging.INFO): self.terminal = sys.stdout self.logger = logger self.log_level = log_level self.linebuf = '' def __getattr__(self, attr): return getattr(self.terminal, attr) def write(self, buf): temp_linebuf = self.linebuf + buf self.linebuf = '' for line in temp_linebuf.splitlines(True): # From the io.TextIOWrapper docs: # On output, if newline is None, any '\n' characters written # are translated to the system default line separator. # By default sys.stdout.write() expects '\n' newlines and then # translates them so this is still cross platform. if line[-1] == '\n': self.logger.log(self.log_level, line.rstrip()) else: self.linebuf += line def flush(self): if self.linebuf != '': self.logger.log(self.log_level, self.linebuf.rstrip()) self.linebuf = '' def disable_torch_init(): """ Disable the redundant torch default initialization to accelerate model creation. """ import torch setattr(torch.nn.Linear, "reset_parameters", lambda self: None) setattr(torch.nn.LayerNorm, "reset_parameters", lambda self: None) def violates_moderation(text): """ Check whether the text violates OpenAI moderation API. """ url = "https://api.openai.com/v1/moderations" headers = {"Content-Type": "application/json", "Authorization": "Bearer " + os.environ["OPENAI_API_KEY"]} text = text.replace("\n", "") data = "{" + '"input": ' + f'"{text}"' + "}" data = data.encode("utf-8") try: ret = requests.post(url, headers=headers, data=data, timeout=5) flagged = ret.json()["results"][0]["flagged"] except requests.exceptions.RequestException as e: flagged = False except KeyError as e: flagged = False return flagged def pretty_print_semaphore(semaphore): if semaphore is None: return "None" return f"Semaphore(value={semaphore._value}, locked={semaphore.locked()})" # Modified from github.com/openai/CLIP import gzip import html import os from functools import lru_cache import ftfy import regex as re import torch @lru_cache() def default_bpe(): return os.path.join(os.path.dirname(os.path.abspath(__file__)), "bpe_simple_vocab_16e6.txt.gz") @lru_cache() def bytes_to_unicode(): """ 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 signficant 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 cs = [chr(n) for n in cs] return dict(zip(bs, cs)) def get_pairs(word): """Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings). """ pairs = set() prev_char = word[0] for char in word[1:]: pairs.add((prev_char, char)) prev_char = char return pairs def basic_clean(text): text = ftfy.fix_text(text) text = html.unescape(html.unescape(text)) return text.strip() def whitespace_clean(text): text = re.sub(r'\s+', ' ', text) text = text.strip() return text class SimpleTokenizer(object): def __init__(self, bpe_path: str = default_bpe()): self.byte_encoder = bytes_to_unicode() self.byte_decoder = {v: k for k, v in self.byte_encoder.items()} merges = gzip.open(bpe_path).read().decode("utf-8").split('\n') merges = merges[1:49152-256-2+1] merges = [tuple(merge.split()) for merge in merges] vocab = list(bytes_to_unicode().values()) vocab = vocab + [v+'' for v in vocab] for merge in merges: vocab.append(''.join(merge)) vocab.extend(['<|startoftext|>', '<|endoftext|>']) self.encoder = dict(zip(vocab, range(len(vocab)))) self.decoder = {v: k for k, v in self.encoder.items()} self.bpe_ranks = dict(zip(merges, range(len(merges)))) self.cache = {'<|startoftext|>': '<|startoftext|>', '<|endoftext|>': '<|endoftext|>'} self.pat = re.compile(r"""<\|startoftext\|>|<\|endoftext\|>|'s|'t|'re|'ve|'m|'ll|'d|[\p{L}]+|[\p{N}]|[^\s\p{L}\p{N}]+""", re.IGNORECASE) def bpe(self, token): if token in self.cache: return self.cache[token] word = tuple(token[:-1]) + ( token[-1] + '',) pairs = get_pairs(word) if not pairs: return token+'' while True: bigram = min(pairs, key = lambda pair: self.bpe_ranks.get(pair, float('inf'))) if bigram not in self.bpe_ranks: break first, second = bigram new_word = [] i = 0 while i < len(word): try: j = word.index(first, i) new_word.extend(word[i:j]) i = j except: new_word.extend(word[i:]) break if word[i] == first and i < len(word)-1 and word[i+1] == second: new_word.append(first+second) i += 2 else: new_word.append(word[i]) i += 1 new_word = tuple(new_word) word = new_word if len(word) == 1: break else: pairs = get_pairs(word) word = ' '.join(word) self.cache[token] = word return word def encode(self, text): bpe_tokens = [] text = whitespace_clean(basic_clean(text)).lower() for token in re.findall(self.pat, text): token = ''.join(self.byte_encoder[b] for b in token.encode('utf-8')) bpe_tokens.extend(self.encoder[bpe_token] for bpe_token in self.bpe(token).split(' ')) return bpe_tokens def decode(self, tokens): text = ''.join([self.decoder[token] for token in tokens]) text = bytearray([self.byte_decoder[c] for c in text]).decode('utf-8', errors="replace").replace('', ' ') return text def __call__(self, texts, context_length=77): if isinstance(texts, str): texts = [texts] sot_token = self.encoder["<|startoftext|>"] eot_token = self.encoder["<|endoftext|>"] all_tokens = [[sot_token] + self.encode(text) + [eot_token] for text in texts] result = torch.zeros(len(all_tokens), context_length, dtype=torch.long) for i, tokens in enumerate(all_tokens): tokens = tokens[:context_length] result[i, :len(tokens)] = torch.tensor(tokens) if len(result) == 1: return result[0] return result