dspy

dspy copied to clipboard

Triton Inference Server

1

When it comes to fully production grade inference servers, TIS is very much optimized and open sourced. So an integration of this in dspy along with trt llm (#1094) would...

Anindyadeep

feature(dspy): adding spBLEU and cosine similarity as new metrics

3

Added support for spBLEU (SentencePiece BLEU) and cosine similarity as new evaluation metrics in the dspy module to enhance text similarity and performance analysis.

fsndzomga

feat(dspy): Experiment with adding image data with GPT-4o and Gemini

6

Add support for Vision data for various LLM vendor (Gemini, GPT, Azure OpenAI GPT). This implements feature requested in https://github.com/stanfordnlp/dspy/issues/624 This adds the `is_image` property to `InputField`. We expect this...

dat-boris

Feature request: Allow specifying chat history for LMs

5

Hi DSPy developers, First of all, thanks a lot for this great work! Recently I've been trying to integrate DSPy into my work, but I stumbled upon the chat history...

ShaojieJiang

Roadmap: Terminology Improvements

2

I think one of the great improvements that DSPy has made was going from the term "Teleprompter" to "Optimizer." That change made what DSPy was doing much clearer. That terminology...

lebsral

Sub-module Compile With Different Settings

1

Sub-modules might need different compile settings. For example, suppose we have a sub-module that summarizes a list of contexts, we may want to set `max_bootstrapped_demos` and `max_labeled_demos` of the main...

etwk

Dspy for multineedle in a haystack

2

I'm trying to use llama 3.1 70B to do "multi-needle in a haystack" search. Basically, I'm asking the model to use a text and search through a list of terms;...

joshpopelka20

How to make this reply only with the answer?

2

How to make this reply only with the answer? 

haalcala

Langtrace added support for native DSPy projects with support for tracing and experimentation. ### Instructions https://docs.langtrace.ai/supported-integrations/llm-frameworks/dspy#dspy ### Inference Metrics  ### Evaluation Scores 

karthikscale3

Cached bad response: ValueError: Expected dict_keys(['answer']) but got dict_keys([])

4

If running an evaluation but the model messes up the output formatting, litellm will still cache it because the response was valid, just not correct schema. If this happens enough...

isaacbmiller