Add ruff.toml

.gitignore

@@ -145,3 +145,4 @@ install-sw.sh
 install-sw1.sh
 win10-install-memo.txt
 model-s
+model-s-v2

.ruff.toml

@@ -0,0 +1,22 @@
+# Assume Python 3.10.
+target-version = "py310"
+# Decrease the maximum line length to 79 characters.
+line-length = 300
+
+# pyflakes, pycodestyle, isort
+# flake8 YTT, pydocstyle D, pylint PLC
+lint.select = ["F", "E", "W", "I001", "YTT", "D", "PLC"]
+# select = ["ALL"]
+
+# D103 Missing docstring in public function
+# D101 Missing docstring in public class
+# `multi-line-summary-first-line` (D212)
+# `one-blank-line-before-class` (D203)
+# imported but unused (F401)
+# assigned to but never used (F841)
+# Missing dashed underline after section ("Args") (D407)
+# D413 [*] Missing blank line after last section ("Returns")
+# D213 [*] Multi-line docstring summary should start at the second line
+lint.extend-ignore = ["D103", "D101", "D212", "D203", "F401", "F841", "D407", "D413", "D213"]
+
+exclude = [".venv"]

st_mlbee/gen_cmat.py

@@ -1,20 +1,18 @@
 """Gen cmat for de/en text."""
 # pylint: disable=invalid-name, too-many-branches
-
 from typing import List, Optional
 
 import more_itertools as mit
 import numpy as np
 
+# from logzero import logger
+from loguru import logger
 from tqdm import tqdm
 
 # from model_pool import load_model_s
 # from hf_model_s_cpu import model_s  # load_model_s directly
 from st_mlbee.load_model_s import load_model_s
 
-# from logzero import logger
-from loguru import logger
-
 # from st_mlbee.cos_matrix2 import cos_matrix2
 from .cos_matrix2 import cos_matrix2
 
@@ -29,7 +27,10 @@ except Exception as exc:
 try:
     # model = model_s()
     # model = model_s(alive_bar_on=True)
-    model = load_model_s()
+
+    # default model-s mikeee/model_s_512
+    model_s = load_model_s()
+    # model_s_v2 = load_model_s("model_s_512v2")  # model-s mikeee/model-s-512v2
 except Exception as _:
     logger.error(_)
     raise
@@ -38,20 +39,31 @@ except Exception as _:
 def gen_cmat(
     text1: List[str],
     text2: List[str],
-    bsize: int = 50
+    bsize: int = 32,  # default batch_size of model.encode
+    model=None,
 ) -> np.ndarray:
     """Gen corr matrix for texts.
 
     Args:
-        text1: typically '''...''' splitlines()
-        text2: typically '''...''' splitlines()
-        bsize: batch size, default 50
+    ----
+    text1: typically '''...''' splitlines()
+    text2: typically '''...''' splitlines()
+    bsize: batch size, default 50
+    model: for encoding list of strings, default model-s of mikeee/model_s_512
+
     text1 = 'this is a test'
     text2 = 'another test'
+
+    Returns:
+    -------
+    numpy array of cmat
+
     """
+    if model is None:
+        model = model_s
     bsize = int(bsize)
     if bsize <= 0:
-        bsize = 50
+        bsize = 32
 
     if isinstance(text1, str):
         text1 = [text1]

st_mlbee/info.py

@@ -1,6 +1,5 @@
 """Present info about st-mlbee."""
 from textwrap import dedent
-
 import streamlit as st
 
 from st_mlbee import __version__
@@ -10,7 +9,6 @@ from st_mlbee.utils import msg
 
 def info():
     """Prep info page."""
-
     st.subheader(f"st-mlbee {__version__}")
 
     st.markdown(msg, unsafe_allow_html=True)

st_mlbee/load_model_s.py

@@ -3,13 +3,13 @@ Load model_s from hf.
 
 cf aslo align-model-pool\model_pool\load_model.py and ycco make-upload-model-s.ipynb.
 """
-import torch
 import joblib
 from huggingface_hub import hf_hub_download
 from loguru import logger
 
 try:
     loc = hf_hub_download("mikeee/model_s_512", "model-s", local_dir=".")
+    # loc2 = hf_hub_download("mikeee/model_s_512v2", "model-s-v2", local_dir=".")
 except Exception as exc:
     logger.error(exc)
     raise SystemExit(1) from exc

st_mlbee/utils.py

@@ -290,7 +290,8 @@ def to_excel(df):
 
 
 def get_table_download_link(df):
-    """Generates a link.
+    """
+    Generate a link.
 
     Allowing the data in a given panda dataframe
     to be downloaded.
@@ -307,7 +308,8 @@ def get_table_download_link(df):
 
 
 def get_table_download_link_sents(df):
-    """Generates a link.
+    """
+    Generate a link.
 
     Allowing the data in a given panda dataframe to be
     downloaded for sents aligned.