From 24aea85b5bc3f354d683201fa9e37968f3f1638f Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期四, 21 三月 2024 14:01:45 +0800
Subject: [PATCH] trainer
---
setup.py | 60 +++++++++++++++++++++++++++++++++++++++++-------------------
1 files changed, 41 insertions(+), 19 deletions(-)
diff --git a/setup.py b/setup.py
index 30e1f1a..9a69fcd 100644
--- a/setup.py
+++ b/setup.py
@@ -10,38 +10,40 @@
requirements = {
"install": [
- "setuptools>=38.5.1",
- "typeguard>=3.0.1",
- "humanfriendly",
"scipy>=1.4.1",
"librosa",
"jamo", # For kss
"PyYAML>=5.1.2",
- "soundfile>=0.10.2",
- "h5py>=2.10.0",
+ "soundfile>=0.12.1",
"kaldiio>=2.17.0",
"torch_complex",
- "nltk>=3.4.5",
- # ASR
- "sentencepiece",
- # TTS
- "pypinyin>=0.44.0",
- "espnet_tts_frontend",
+ # "nltk>=3.4.5",
+ "sentencepiece", # train
+ "jieba",
+ "rotary_embedding_torch",
+ # "ffmpeg-python",
+ # "pypinyin>=0.44.0",
+ # "espnet_tts_frontend",
# ENH
"pytorch_wpe",
"editdistance>=0.5.2",
- "tensorboard",
- "g2p",
+ # "g2p",
+ # "nara_wpe",
# PAI
"oss2",
- "edit-distance",
- "textgrid",
- "protobuf",
+ # "edit-distance",
+ # "textgrid",
+ # "protobuf",
+ "tqdm",
+ "umap_learn",
+ "jaconv",
+ "hydra-core>=1.3.2",
+ "tensorboardX",
+ "rotary_embedding_torch",
],
# train: The modules invoked when training only.
"train": [
"editdistance",
- "wandb",
],
# all: The modules should be optionally installled due to some reason.
# Please consider moving them to "install" occasionally
@@ -50,6 +52,7 @@
"torch_optimizer",
"fairscale",
"transformers",
+ "openai-whisper"
],
"setup": [
"numpy",
@@ -79,8 +82,20 @@
"sphinx-markdown-tables>=0.0.12",
"configargparse>=1.2.1"
],
+ "llm":[
+ "transformers>=4.32.0",
+ "accelerate",
+ "tiktoken",
+ "einops",
+ "transformers_stream_generator>=0.0.4",
+ "scipy",
+ "torchvision",
+ "pillow",
+ "matplotlib",
+ ],
}
requirements["all"].extend(requirements["train"])
+requirements["all"].extend(requirements["llm"])
requirements["test"].extend(requirements["train"])
install_requires = requirements["install"]
@@ -98,7 +113,7 @@
name="funasr",
version=version,
url="https://github.com/alibaba-damo-academy/FunASR.git",
- author="Speech Lab of DAMO Academy, Alibaba Group",
+ author="Speech Lab of Alibaba Group",
author_email="funasr@list.alibaba-inc.com",
description="FunASR: A Fundamental End-to-End Speech Recognition Toolkit",
long_description=open(os.path.join(dirname, "README.md"), encoding="utf-8").read(),
@@ -123,4 +138,11 @@
"License :: OSI Approved :: Apache Software License",
"Topic :: Software Development :: Libraries :: Python Modules",
],
-)
\ No newline at end of file
+ entry_points={"console_scripts": [
+ "funasr = funasr.bin.inference:main_hydra",
+ "funasr-train = funasr.bin.train:main_hydra",
+ "funasr-export = funasr.bin.export:main_hydra",
+ "scp2jsonl = funasr.datasets.audio_datasets.scp2jsonl:main_hydra",
+ "jsonl2scp = funasr.datasets.audio_datasets.jsonl2scp:main_hydra",
+ ]},
+)
--
Gitblit v1.9.1