Per module lm history #8199
Merged
Per module lm history #8199
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
okhat
reviewed
May 22, 2025
| @@ -17,10 +17,10 @@ | |||
|
|
|||
| class Predict(Module, Parameter): | |||
| def __init__(self, signature, callbacks=None, **config): | |||
| super().__init__(callbacks=callbacks) | |||
There was a problem hiding this comment.
Btw typical custom DSPy programs don't call super().__init__(...).
Will that be an issue?
There was a problem hiding this comment.
Nothing will crash, but they will lose some perks from dspy.Module. We can follow what PyTorch does - recommending calling super().__init__() when customizing modules: https://docs.pytorch.org/tutorials/beginner/examples_nn/polynomial_module.html#pytorch-custom-nn-modules
okhat
reviewed
May 22, 2025
| @@ -20,26 +21,38 @@ def _base_init(self): | |||
| def __init__(self, callbacks=None): | |||
| self.callbacks = callbacks or [] | |||
| self._compiled = False | |||
| # LM calling history of the module. | |||
| self.history = [] | |||
There was a problem hiding this comment.
Cool! Does this work if there are multiple LMs used in the module for different sub-modules?
There was a problem hiding this comment.
yes! This per-module history works the same for multi-lm situation.
Just tested with the following code and no issue:
import dspy
class MyProgram(dspy.Module):
def __init__(self):
super().__init__()
self.cot = dspy.ChainOfThought("question -> answer")
self.cot2 = dspy.ChainOfThought("question, answer -> judgement")
self.cot2.predict.lm = dspy.LM("openai/gpt-4o", cache=False)
def forward(self, question: str, **kwargs) -> str:
answer = self.cot(question=question).answer
return self.cot2(question=question, answer=answer)
dspy.settings.configure(lm=dspy.LM("openai/gpt-4o-mini", cache=False), adapter=dspy.JSONAdapter())
program = MyProgram()
program(question="What is the capital of France?")
print(program.history)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Make it possible to do
module.inspect_history, which gives the LM history of the module. In nesting case, the same LM history message will be available to all the ancestor callers.Example code:
It will give:
You can also access the history by
program.cot.inspect_history.