symbolicai 1.5.0__py3-none-any.whl → 1.6.0__py3-none-any.whl

symai/extended/strategies/cot.py

@@ -1,40 +0,0 @@
-from ...models import LLMDataModel
-from ...prompts import PromptLanguage, PromptRegistry
-from ...strategy import BaseStrategy
-
-registry = PromptRegistry()
-registry.register_instruction(PromptLanguage.ENGLISH, "static_context_chain_of_thoughts",
-"""
-{
-    "step_by_step": "Q: Your warehouse has 5 pallets of widgets. You purchase 2 more shipments of widgets. Each shipment contains 3 pallets. How many pallets of widgets do you have now?
-A: The warehouse started with 5 pallets of widgets. 2 shipments of 3 pallets each is 6 pallets. 5 pallets + 6 pallets = 11 pallets. The answer is 11 pallets.
-Q: Your finance department has $23,000 in the budget. If they allocate $20,000 for a marketing campaign and add $6,000 from other savings, how much is left in the budget?"
-    "answer": "The finance department started with $23,000. They allocated $20,000 for a marketing campaign and added $6,000 from other savings. $23,000 - $20,000 + $6,000 = $9,000. The answer is $9,000."
-}
-"""+
-"Think step by step as in the example above before answering."+
-"Return the 'step_by_step' and 'answer' properties of the JSON format."
-"Always ONLY return a valid JSON format that solve the task.")
-
-
-class Result(LLMDataModel):
-    step_by_step: str
-    answer: str
-
-
-class ChainOfThought(BaseStrategy):
-    def __init__(self, data_model=Result, *args, **kwargs):
-        super().__init__(data_model=data_model, *args, **kwargs)
-
-    @property
-    def task(self):
-        task_str = (
-            "Think step by step to solve a problem and answer." \
-                + r'JSON format: {"step_by_step": "string", "answer": "string"}'
-        )
-        self.print_verbose(task_str)
-        return task_str
-
-    @property
-    def static_context(self):
-        return registry.instruction("static_context_chain_of_thoughts")