sentence-transformers How to use fyaronskiy/english_code_retriever with sentence-transformers:
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("fyaronskiy/english_code_retriever")
sentences = [
"search_query: Finds the top long, short, and absolute positions.\n\n Parameters\n ----------\n positions : pd.DataFrame\n The positions that the strategy takes over time.\n top : int, optional\n How many of each to find (default 10).\n\n Returns\n -------\n df_top_long : pd.DataFrame\n Top long positions.\n df_top_short : pd.DataFrame\n Top short positions.\n df_top_abs : pd.DataFrame\n Top absolute positions.",
"search_document: def symmetric_ema(xolds, yolds, low=None, high=None, n=512, decay_steps=1., low_counts_threshold=1e-8):\n '''\n perform symmetric EMA (exponential moving average)\n smoothing and resampling to an even grid with n points.\n Does not do extrapolation, so we assume\n xolds[0] <= low && high <= xolds[-1]\n\n Arguments:\n\n xolds: array or list - x values of data. Needs to be sorted in ascending order\n yolds: array of list - y values of data. Has to have the same length as xolds\n\n low: float - min value of the new x grid. By default equals to xolds[0]\n high: float - max value of the new x grid. By default equals to xolds[-1]\n\n n: int - number of points in new x grid\n\n decay_steps: float - EMA decay factor, expressed in new x grid steps.\n\n low_counts_threshold: float or int\n - y values with counts less than this value will be set to NaN\n\n Returns:\n tuple sum_ys, count_ys where\n xs - array with new x grid\n ys - array of EMA of y at each point of the new x grid\n count_ys - array of EMA of y counts at each point of the new x grid\n\n '''\n xs, ys1, count_ys1 = one_sided_ema(xolds, yolds, low, high, n, decay_steps, low_counts_threshold=0)\n _, ys2, count_ys2 = one_sided_ema(-xolds[::-1], yolds[::-1], -high, -low, n, decay_steps, low_counts_threshold=0)\n ys2 = ys2[::-1]\n count_ys2 = count_ys2[::-1]\n count_ys = count_ys1 + count_ys2\n ys = (ys1 * count_ys1 + ys2 * count_ys2) / count_ys\n ys[count_ys < low_counts_threshold] = np.nan\n return xs, ys, count_ys",
"search_document: def project(self, from_shape, to_shape):\n \"\"\"\n Project the polygon onto an image with different shape.\n\n The relative coordinates of all points remain the same.\n E.g. a point at (x=20, y=20) on an image (width=100, height=200) will be\n projected on a new image (width=200, height=100) to (x=40, y=10).\n\n This is intended for cases where the original image is resized.\n It cannot be used for more complex changes (e.g. padding, cropping).\n\n Parameters\n ----------\n from_shape : tuple of int\n Shape of the original image. (Before resize.)\n\n to_shape : tuple of int\n Shape of the new image. (After resize.)\n\n Returns\n -------\n imgaug.Polygon\n Polygon object with new coordinates.\n\n \"\"\"\n if from_shape[0:2] == to_shape[0:2]:\n return self.copy()\n ls_proj = self.to_line_string(closed=False).project(\n from_shape, to_shape)\n return self.copy(exterior=ls_proj.coords)",
"search_document: def get_top_long_short_abs(positions, top=10):\n \"\"\"\n Finds the top long, short, and absolute positions.\n\n Parameters\n ----------\n positions : pd.DataFrame\n The positions that the strategy takes over time.\n top : int, optional\n How many of each to find (default 10).\n\n Returns\n -------\n df_top_long : pd.DataFrame\n Top long positions.\n df_top_short : pd.DataFrame\n Top short positions.\n df_top_abs : pd.DataFrame\n Top absolute positions.\n \"\"\"\n\n positions = positions.drop('cash', axis='columns')\n df_max = positions.max()\n df_min = positions.min()\n df_abs_max = positions.abs().max()\n df_top_long = df_max[df_max > 0].nlargest(top)\n df_top_short = df_min[df_min < 0].nsmallest(top)\n df_top_abs = df_abs_max.nlargest(top)\n return df_top_long, df_top_short, df_top_abs"
]
embeddings = model.encode(sentences)
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [4, 4] 
