xinference 0.14.1.post1__py3-none-any.whl → 0.14.3__py3-none-any.whl

xinference/thirdparty/fish_speech/fish_speech/text/chn_text_norm/fraction.py

@@ -0,0 +1,35 @@
+# -*- coding: utf-8 -*-
+"""FRACTION类
+分数 <=> 中文字符串 方法
+中文字符串 <=> 分数 方法
+"""
+
+__author__ = "Zhiyang Zhou <zyzhou@stu.xmu.edu.cn>"
+__data__ = "2019-05-03"
+
+from fish_speech.text.chn_text_norm.basic_util import *
+
+
+class Fraction:
+    """
+    FRACTION类
+    """
+
+    def __init__(self, fraction=None, chntext=None):
+        self.fraction = fraction
+        self.chntext = chntext
+
+    def chntext2fraction(self):
+        denominator, numerator = self.chntext.split("分之")
+        return chn2num(numerator) + "/" + chn2num(denominator)
+
+    def fraction2chntext(self):
+        numerator, denominator = self.fraction.split("/")
+        return num2chn(denominator) + "分之" + num2chn(numerator)
+
+
+if __name__ == "__main__":
+
+    # 测试程序
+    print(Fraction(fraction="2135/7230").fraction2chntext())
+    print(Fraction(chntext="五百八十一分之三百六十九").chntext2fraction())

xinference/thirdparty/fish_speech/fish_speech/text/chn_text_norm/money.py

@@ -0,0 +1,43 @@
+# -*- coding: utf-8 -*-
+"""MONEY类
+金钱 <=> 中文字符串 方法
+中文字符串 <=> 金钱 方法
+"""
+import re
+
+__author__ = "Zhiyang Zhou <zyzhou@stu.xmu.edu.cn>"
+__data__ = "2019-05-08"
+
+from fish_speech.text.chn_text_norm.cardinal import Cardinal
+
+
+class Money:
+    """
+    MONEY类
+    """
+
+    def __init__(self, money=None, chntext=None):
+        self.money = money
+        self.chntext = chntext
+
+    # def chntext2money(self):
+    #     return self.money
+
+    def money2chntext(self):
+        money = self.money
+        pattern = re.compile(r"(\d+(\.\d+)?)")
+        matchers = pattern.findall(money)
+        if matchers:
+            for matcher in matchers:
+                money = money.replace(
+                    matcher[0], Cardinal(cardinal=matcher[0]).cardinal2chntext()
+                )
+        self.chntext = money
+        return self.chntext
+
+
+if __name__ == "__main__":
+
+    # 测试
+    print(Money(money="21.5万元").money2chntext())
+    print(Money(money="230块5毛").money2chntext())

xinference/thirdparty/fish_speech/fish_speech/text/chn_text_norm/percentage.py

@@ -0,0 +1,33 @@
+# -*- coding: utf-8 -*-
+"""PERCENTAGE类
+百分数 <=> 中文字符串 方法
+中文字符串 <=> 百分数 方法
+"""
+
+__author__ = "Zhiyang Zhou <zyzhou@stu.xmu.edu.cn>"
+__data__ = "2019-05-06"
+
+from fish_speech.text.chn_text_norm.basic_util import *
+
+
+class Percentage:
+    """
+    PERCENTAGE类
+    """
+
+    def __init__(self, percentage=None, chntext=None):
+        self.percentage = percentage
+        self.chntext = chntext
+
+    def chntext2percentage(self):
+        return chn2num(self.chntext.strip().strip("百分之")) + "%"
+
+    def percentage2chntext(self):
+        return "百分之" + num2chn(self.percentage.strip().strip("%"))
+
+
+if __name__ == "__main__":
+
+    # 测试程序
+    print(Percentage(chntext="百分之五十六点零三").chntext2percentage())
+    print(Percentage(percentage="65.3%").percentage2chntext())

xinference/thirdparty/fish_speech/fish_speech/text/chn_text_norm/telephone.py

@@ -0,0 +1,51 @@
+# -*- coding: utf-8 -*-
+"""TELEPHONE类
+电话号码 <=> 中文字符串 方法
+中文字符串 <=> 电话号码 方法
+"""
+
+__author__ = "Zhiyang Zhou <zyzhou@stu.xmu.edu.cn>"
+__data__ = "2019-05-03"
+
+from fish_speech.text.chn_text_norm.basic_util import *
+
+
+class TelePhone:
+    """
+    TELEPHONE类
+    """
+
+    def __init__(self, telephone=None, raw_chntext=None, chntext=None):
+        self.telephone = telephone
+        self.raw_chntext = raw_chntext
+        self.chntext = chntext
+
+    # def chntext2telephone(self):
+    #     sil_parts = self.raw_chntext.split('<SIL>')
+    #     self.telephone = '-'.join([
+    #         str(chn2num(p)) for p in sil_parts
+    #     ])
+    #     return self.telephone
+
+    def telephone2chntext(self, fixed=False):
+
+        if fixed:
+            sil_parts = self.telephone.split("-")
+            self.raw_chntext = "<SIL>".join(
+                [num2chn(part, alt_two=False, use_units=False) for part in sil_parts]
+            )
+            self.chntext = self.raw_chntext.replace("<SIL>", "")
+        else:
+            sp_parts = self.telephone.strip("+").split()
+            self.raw_chntext = "<SP>".join(
+                [num2chn(part, alt_two=False, use_units=False) for part in sp_parts]
+            )
+            self.chntext = self.raw_chntext.replace("<SP>", "")
+        return self.chntext
+
+
+if __name__ == "__main__":
+
+    # 测试程序
+    print(TelePhone(telephone="0595-23980880").telephone2chntext())
+    # print(TelePhone(raw_chntext='零五九五杠二三八六五零九八').chntext2telephone())

xinference/thirdparty/fish_speech/fish_speech/text/chn_text_norm/text.py

@@ -0,0 +1,177 @@
+# -*- coding: utf-8 -*-
+"""
+TEXT类
+"""
+
+__author__ = "Zhiyang Zhou <zyzhou@stu.xmu.edu.cn>"
+__data__ = "2019-05-03"
+
+import re
+
+from fish_speech.text.chn_text_norm.cardinal import Cardinal
+from fish_speech.text.chn_text_norm.date import Date
+from fish_speech.text.chn_text_norm.digit import Digit
+from fish_speech.text.chn_text_norm.fraction import Fraction
+from fish_speech.text.chn_text_norm.money import Money
+from fish_speech.text.chn_text_norm.percentage import Percentage
+from fish_speech.text.chn_text_norm.telephone import TelePhone
+
+CURRENCY_NAMES = (
+    "(人民币|美元|日元|英镑|欧元|马克|法郎|加拿大元|澳元|港币|先令|芬兰马克|爱尔兰镑|"
+    "里拉|荷兰盾|埃斯库多|比塞塔|印尼盾|林吉特|新西兰元|比索|卢布|新加坡元|韩元|泰铢)"
+)
+CURRENCY_UNITS = "((亿|千万|百万|万|千|百)|(亿|千万|百万|万|千|百|)元|(亿|千万|百万|万|千|百|)块|角|毛|分)"
+COM_QUANTIFIERS = (
+    "(匹|张|座|回|场|尾|条|个|首|阙|阵|网|炮|顶|丘|棵|只|支|袭|辆|挑|担|颗|壳|窠|曲|墙|群|腔|"
+    "砣|座|客|贯|扎|捆|刀|令|打|手|罗|坡|山|岭|江|溪|钟|队|单|双|对|出|口|头|脚|板|跳|枝|件|贴|"
+    "针|线|管|名|位|身|堂|课|本|页|家|户|层|丝|毫|厘|分|钱|两|斤|担|铢|石|钧|锱|忽|(千|毫|微)克|"
+    "毫|厘|分|寸|尺|丈|里|寻|常|铺|程|(千|分|厘|毫|微)米|撮|勺|合|升|斗|石|盘|碗|碟|叠|桶|笼|盆|"
+    "盒|杯|钟|斛|锅|簋|篮|盘|桶|罐|瓶|壶|卮|盏|箩|箱|煲|啖|袋|钵|年|月|日|季|刻|时|周|天|秒|分|旬|"
+    "纪|岁|世|更|夜|春|夏|秋|冬|代|伏|辈|丸|泡|粒|颗|幢|堆|条|根|支|道|面|片|张|颗|块|人|抽)"
+)
+
+
+class Text:
+    """
+    Text类
+    """
+
+    def __init__(self, raw_text, norm_text=None):
+        self.raw_text = "^" + raw_text + "$"
+        self.norm_text = norm_text
+
+    def _particular(self):
+        text = self.norm_text
+        pattern = re.compile(r"(([a-zA-Z]+)二([a-zA-Z]+))")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('particular')
+            for matcher in matchers:
+                text = text.replace(matcher[0], matcher[1] + "2" + matcher[2], 1)
+        self.norm_text = text
+        return self.norm_text
+
+    def normalize(self):
+        text = self.raw_text
+
+        # 规范化日期
+        pattern = re.compile(
+            r"\D+((([089]\d|(19|20)\d{2})年)?(\d{1,2}月(\d{1,2}[日号])?)?)"
+        )
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('date')
+            for matcher in matchers:
+                text = text.replace(matcher[0], Date(date=matcher[0]).date2chntext(), 1)
+
+        # 规范化金钱
+        pattern = re.compile(
+            r"\D+((\d+(\.\d+)?)[多余几]?"
+            + CURRENCY_UNITS
+            + "(\d"
+            + CURRENCY_UNITS
+            + "?)?)"
+        )
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('money')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher[0], Money(money=matcher[0]).money2chntext(), 1
+                )
+
+        # 规范化固话/手机号码
+        # 手机
+        # http://www.jihaoba.com/news/show/13680
+        # 移动：139、138、137、136、135、134、159、158、157、150、151、152、188、187、182、183、184、178、198
+        # 联通：130、131、132、156、155、186、185、176
+        # 电信：133、153、189、180、181、177
+        pattern = re.compile(r"\D((\+?86 ?)?1([38]\d|5[0-35-9]|7[678]|9[89])\d{8})\D")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('telephone')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher[0], TelePhone(telephone=matcher[0]).telephone2chntext(), 1
+                )
+        # 固话
+        pattern = re.compile(r"\D((0(10|2[1-3]|[3-9]\d{2})-?)?[1-9]\d{6,7})\D")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('fixed telephone')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher[0],
+                    TelePhone(telephone=matcher[0]).telephone2chntext(fixed=True),
+                    1,
+                )
+
+        # 规范化分数
+        pattern = re.compile(r"(\d+/\d+)")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('fraction')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher, Fraction(fraction=matcher).fraction2chntext(), 1
+                )
+
+        # 规范化百分数
+        text = text.replace("％", "%")
+        pattern = re.compile(r"(\d+(\.\d+)?%)")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('percentage')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher[0],
+                    Percentage(percentage=matcher[0]).percentage2chntext(),
+                    1,
+                )
+
+        # 规范化纯数+量词
+        pattern = re.compile(r"(\d+(\.\d+)?)[多余几]?" + COM_QUANTIFIERS)
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('cardinal+quantifier')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher[0], Cardinal(cardinal=matcher[0]).cardinal2chntext(), 1
+                )
+
+        # 规范化数字编号
+        pattern = re.compile(r"(\d{4,32})")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('digit')
+            for matcher in matchers:
+                text = text.replace(matcher, Digit(digit=matcher).digit2chntext(), 1)
+
+        # 规范化纯数
+        pattern = re.compile(r"(\d+(\.\d+)?)")
+        matchers = pattern.findall(text)
+        if matchers:
+            # print('cardinal')
+            for matcher in matchers:
+                text = text.replace(
+                    matcher[0], Cardinal(cardinal=matcher[0]).cardinal2chntext(), 1
+                )
+
+        self.norm_text = text
+        self._particular()
+
+        return self.norm_text.lstrip("^").rstrip("$")
+
+
+if __name__ == "__main__":
+
+    # 测试程序
+    print(Text(raw_text="固话：0595-23865596或23880880。").normalize())
+    print(Text(raw_text="手机：+86 19859213959或15659451527。").normalize())
+    print(Text(raw_text="分数：32477/76391。").normalize())
+    print(Text(raw_text="百分数：80.03%。").normalize())
+    print(Text(raw_text="编号：31520181154418。").normalize())
+    print(Text(raw_text="纯数：2983.07克或12345.60米。").normalize())
+    print(Text(raw_text="日期：1999年2月20日或09年3月15号。").normalize())
+    print(Text(raw_text="金钱：12块5，34.5元，20.1万").normalize())
+    print(Text(raw_text="特殊：O2O或B2C。").normalize())

xinference/thirdparty/fish_speech/fish_speech/text/clean.py

@@ -0,0 +1,69 @@
+import itertools
+import re
+
+LANGUAGE_UNICODE_RANGE_MAP = {
+    "ZH": [(0x4E00, 0x9FFF)],
+    "JP": [(0x4E00, 0x9FFF), (0x3040, 0x309F), (0x30A0, 0x30FF), (0x31F0, 0x31FF)],
+    "EN": [(0x0000, 0x007F)],
+}
+
+SYMBOLS_MAPPING = {
+    "：": ",",
+    "；": ",",
+    "，": ",",
+    "。": ".",
+    "！": "!",
+    "？": "?",
+    "\n": ".",
+    "·": ",",
+    "、": ",",
+    "...": "…",
+    "“": "'",
+    "”": "'",
+    "‘": "'",
+    "’": "'",
+    "（": "'",
+    "）": "'",
+    "(": "'",
+    ")": "'",
+    "《": "'",
+    "》": "'",
+    "【": "'",
+    "】": "'",
+    "[": "'",
+    "]": "'",
+    "—": "-",
+    "～": "-",
+    "~": "-",
+    "・": "-",
+    "「": "'",
+    "」": "'",
+    ";": ",",
+    ":": ",",
+}
+
+REPLACE_SYMBOL_REGEX = re.compile(
+    "|".join(re.escape(p) for p in SYMBOLS_MAPPING.keys())
+)
+ALL_KNOWN_UTF8_RANGE = list(
+    itertools.chain.from_iterable(LANGUAGE_UNICODE_RANGE_MAP.values())
+)
+REMOVE_UNKNOWN_SYMBOL_REGEX = re.compile(
+    "[^"
+    + "".join(
+        f"{re.escape(chr(start))}-{re.escape(chr(end))}"
+        for start, end in ALL_KNOWN_UTF8_RANGE
+    )
+    + "]"
+)
+
+
+def clean_text(text):
+    # Clean the text
+    text = text.strip()
+
+    # Replace all chinese symbols with their english counterparts
+    text = REPLACE_SYMBOL_REGEX.sub(lambda x: SYMBOLS_MAPPING[x.group()], text)
+    text = REMOVE_UNKNOWN_SYMBOL_REGEX.sub("", text)
+
+    return text

xinference/thirdparty/fish_speech/fish_speech/text/spliter.py

@@ -0,0 +1,130 @@
+import re
+import string
+
+from fish_speech.text.clean import clean_text
+
+
+def utf_8_len(text):
+    return len(text.encode("utf-8"))
+
+
+def break_text(texts, length, splits: set):
+    for text in texts:
+        if utf_8_len(text) <= length:
+            yield text
+            continue
+
+        curr = ""
+        for char in text:
+            curr += char
+
+            if char in splits:
+                yield curr
+                curr = ""
+
+        if curr:
+            yield curr
+
+
+def break_text_by_length(texts, length):
+    for text in texts:
+        if utf_8_len(text) <= length:
+            yield text
+            continue
+
+        curr = ""
+        for char in text:
+            curr += char
+
+            if utf_8_len(curr) >= length:
+                yield curr
+                curr = ""
+
+        if curr:
+            yield curr
+
+
+def add_cleaned(curr, segments):
+    curr = curr.strip()
+    if curr and not all(c.isspace() or c in string.punctuation for c in curr):
+        segments.append(curr)
+
+
+def protect_float(text):
+    # Turns 3.14 into <3_f_14> to prevent splitting
+    return re.sub(r"(\d+)\.(\d+)", r"<\1_f_\2>", text)
+
+
+def unprotect_float(text):
+    # Turns <3_f_14> into 3.14
+    return re.sub(r"<(\d+)_f_(\d+)>", r"\1.\2", text)
+
+
+def split_text(text, length):
+    text = clean_text(text)
+
+    # Break the text into pieces with following rules:
+    # 1. Split the text at ".", "!", "?" if text is NOT a float
+    # 2. If the text is longer than length, split at ","
+    # 3. If the text is still longer than length, split at " "
+    # 4. If the text is still longer than length, split at any character to length
+
+    texts = [text]
+    texts = map(protect_float, texts)
+    texts = break_text(texts, length, {".", "!", "?"})
+    texts = map(unprotect_float, texts)
+    texts = break_text(texts, length, {","})
+    texts = break_text(texts, length, {" "})
+    texts = list(break_text_by_length(texts, length))
+
+    # Then, merge the texts into segments with length <= length
+    segments = []
+    curr = ""
+
+    for text in texts:
+        if utf_8_len(curr) + utf_8_len(text) <= length:
+            curr += text
+        else:
+            add_cleaned(curr, segments)
+            curr = text
+
+    if curr:
+        add_cleaned(curr, segments)
+
+    return segments
+
+
+if __name__ == "__main__":
+    # Test the split_text function
+
+    text = "This is a test sentence. This is another test sentence. And a third one."
+
+    assert split_text(text, 50) == [
+        "This is a test sentence.",
+        "This is another test sentence. And a third one.",
+    ]
+    assert split_text("a,aaaaaa3.14", 10) == ["a,", "aaaaaa3.14"]
+    assert split_text("   ", 10) == []
+    assert split_text("a", 10) == ["a"]
+
+    text = "This is a test sentence with only commas, and no dots, and no exclamation marks, and no question marks, and no newlines."
+    assert split_text(text, 50) == [
+        "This is a test sentence with only commas,",
+        "and no dots, and no exclamation marks,",
+        "and no question marks, and no newlines.",
+    ]
+
+    text = "This is a test sentence This is a test sentence This is a test sentence. This is a test sentence, This is a test sentence, This is a test sentence."
+    # First half split at " ", second half split at ","
+    assert split_text(text, 50) == [
+        "This is a test sentence This is a test sentence",
+        "This is a test sentence. This is a test sentence,",
+        "This is a test sentence, This is a test sentence.",
+    ]
+
+    text = "这是一段很长的中文文本,而且没有句号,也没有感叹号,也没有问号,也没有换行符。"
+    assert split_text(text, 50) == [
+        "这是一段很长的中文文本,",
+        "而且没有句号,也没有感叹号,",
+        "也没有问号,也没有换行符.",
+    ]

xinference/thirdparty/fish_speech/fish_speech/train.py

@@ -0,0 +1,139 @@
+import os
+import sys
+from typing import Optional
+
+import hydra
+import lightning as L
+# import pyrootutils
+import torch
+from lightning import Callback, LightningDataModule, LightningModule, Trainer
+from lightning.pytorch.loggers import Logger
+from lightning.pytorch.strategies import DDPStrategy
+from omegaconf import DictConfig, OmegaConf
+
+os.environ.pop("SLURM_NTASKS", None)
+os.environ.pop("SLURM_JOB_NAME", None)
+os.environ.pop("SLURM_NTASKS_PER_NODE", None)
+
+# register eval resolver and root
+# pyrootutils.setup_root(__file__, indicator=".project-root", pythonpath=True)
+
+# Allow TF32 on Ampere GPUs
+torch.set_float32_matmul_precision("high")
+torch.backends.cudnn.allow_tf32 = True
+
+# register eval resolver
+OmegaConf.register_new_resolver("eval", eval)
+
+import fish_speech.utils as utils
+
+log = utils.RankedLogger(__name__, rank_zero_only=True)
+
+
+@utils.task_wrapper
+def train(cfg: DictConfig) -> tuple[dict, dict]:
+    """Trains the model. Can additionally evaluate on a testset, using best weights obtained during
+    training.
+    This method is wrapped in optional @task_wrapper decorator, that controls the behavior during
+    failure. Useful for multiruns, saving info about the crash, etc.
+    Args:
+        cfg (DictConfig): Configuration composed by Hydra.
+    Returns:
+        Tuple[dict, dict]: Dict with metrics and dict with all instantiated objects.
+    """  # noqa: E501
+
+    # set seed for random number generators in pytorch, numpy and python.random
+    if cfg.get("seed"):
+        L.seed_everything(cfg.seed, workers=False)
+
+    if cfg.get("deterministic"):
+        torch.use_deterministic_algorithms(True)
+
+    log.info(f"Instantiating datamodule <{cfg.data._target_}>")
+    datamodule: LightningDataModule = hydra.utils.instantiate(cfg.data)
+
+    log.info(f"Instantiating model <{cfg.model._target_}>")
+    model: LightningModule = hydra.utils.instantiate(cfg.model)
+
+    log.info("Instantiating callbacks...")
+    callbacks: list[Callback] = utils.instantiate_callbacks(cfg.get("callbacks"))
+
+    log.info("Instantiating loggers...")
+    logger: list[Logger] = utils.instantiate_loggers(cfg.get("logger"))
+
+    log.info(f"Instantiating trainer <{cfg.trainer._target_}>")
+    trainer: Trainer = hydra.utils.instantiate(
+        cfg.trainer,
+        callbacks=callbacks,
+        logger=logger,
+    )
+
+    object_dict = {
+        "cfg": cfg,
+        "datamodule": datamodule,
+        "model": model,
+        "callbacks": callbacks,
+        "logger": logger,
+        "trainer": trainer,
+    }
+
+    if logger:
+        log.info("Logging hyperparameters!")
+        utils.log_hyperparameters(object_dict)
+
+    if cfg.get("train"):
+        log.info("Starting training!")
+
+        ckpt_path = cfg.get("ckpt_path")
+        auto_resume = False
+
+        resume_ckpt_path = utils.get_latest_checkpoint(cfg.paths.ckpt_dir)
+        if resume_ckpt_path is not None:
+            ckpt_path = resume_ckpt_path
+            auto_resume = True
+
+        if ckpt_path is not None:
+            log.info(f"Resuming from checkpoint: {ckpt_path}")
+
+        # resume weights only is disabled for auto-resume
+        if cfg.get("resume_weights_only") and auto_resume is False:
+            log.info("Resuming weights only!")
+            ckpt = torch.load(ckpt_path, map_location=model.device)
+            if "state_dict" in ckpt:
+                ckpt = ckpt["state_dict"]
+            err = model.load_state_dict(ckpt, strict=False)
+            log.info(f"Error loading state dict: {err}")
+            ckpt_path = None
+
+        trainer.fit(model=model, datamodule=datamodule, ckpt_path=ckpt_path)
+
+    train_metrics = trainer.callback_metrics
+
+    if cfg.get("test"):
+        log.info("Starting testing!")
+        ckpt_path = trainer.checkpoint_callback.best_model_path
+        if ckpt_path == "":
+            log.warning("Best ckpt not found! Using current weights for testing...")
+            ckpt_path = cfg.get("ckpt_path")
+
+        trainer.test(model=model, datamodule=datamodule, ckpt_path=ckpt_path)
+        log.info(f"Best ckpt path: {ckpt_path}")
+
+    test_metrics = trainer.callback_metrics
+
+    # merge train and test metrics
+    metric_dict = {**train_metrics, **test_metrics}
+
+    return metric_dict, object_dict
+
+
+@hydra.main(
+    version_base="1.3", config_path="./configs", config_name="llama_pretrain.yaml"
+)
+def main(cfg: DictConfig) -> Optional[float]:
+    # train the model
+    train(cfg)
+
+
+if __name__ == "__main__":
+    main()