.

utils/utils_inference.py

@@ -0,0 +1,36 @@
+#!/usr/bin/python3
+## Author: Raeid Saqur
+
+### -------- CONSTANTS -------- ###
+
+LABELS = ["Fall", "Neutral", "Rise"]
+LABEL_MAP = {"Rise": 2, "Neutral": 1, "Fall": 0}
+NUMERIC_LABEL_MAP = {v: k for k, v in LABEL_MAP.items()}
+SEEDS = [0, 13, 42]
+
+SYSTEM_ROLE_DEF_1 = "You are a helpful assistant and a financial technical analyst."
+SYSTEM_ROLE_DEF_2 = ("You are a helpful financial market technical analyst. "
+                     "You specialize in financial stock and equities market, a top expert in assessing market index movement direction from events and news. ")
+
+
+def get_truncated_user_prompt_for_nifty(user_prompt: str, drop_percent: float = 0.5) -> str:
+    """Keeps instruction and context unchanged, drops p% of news headlines randomly
+    Usage e.g.:
+        user_prompt = get_truncated_user_prompt_for_nifty(user_prompt, drop_percent=drop_percent)
+    """
+    import random
+
+    splits = user_prompt.split("\n\n")
+    context, news = splits[:-1], splits[-1]
+    news_headlines = news.split("\n")
+    news_headlines, suffix = news_headlines[:-1], news_headlines[-1]
+    N = len(news_headlines)
+    N_truncated = int(N * drop_percent)
+    random.shuffle(news_headlines)
+    truncated_news_headlines = news_headlines[:N_truncated] + [suffix]
+    truncated_news_string = "\n".join(truncated_news_headlines)
+    truncated_user_prompt = context + [truncated_news_string]
+    truncated_user_prompt = "\n\n".join(truncated_user_prompt)
+
+    return truncated_user_prompt
+