From 1089d5ca282acfde87055353bd6006d0e5ac6057 Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 27 Oct 2025 08:12:30 +0000 Subject: [PATCH 1/2] Add missing explanation for SIMBA optimizer MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Added descriptive text for SIMBA in the Automatic Instruction Optimization section to match the style and detail of other optimizer explanations (COPRO, MIPROv2, GEPA). 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude --- docs/docs/learn/optimization/optimizers.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/docs/learn/optimization/optimizers.md b/docs/docs/learn/optimization/optimizers.md index a684f6f5b4..979d7efc0e 100644 --- a/docs/docs/learn/optimization/optimizers.md +++ b/docs/docs/learn/optimization/optimizers.md @@ -58,7 +58,7 @@ These optimizers produce optimal instructions for the prompt and, in the case of 6. [**`MIPROv2`**](../../api/optimizers/MIPROv2.md): Generates instructions *and* few-shot examples in each step. The instruction generation is data-aware and demonstration-aware. Uses Bayesian Optimization to effectively search over the space of generation instructions/demonstrations across your modules. -7. [**`SIMBA`**](../../api/optimizers/SIMBA.md) +7. [**`SIMBA`**](../../api/optimizers/SIMBA.md): Uses the LLM to analyze its own performance, identifies challenging examples with high output variability, and generates improvement rules by either creating self-reflective rules or adding successful examples as demonstrations. 8. [**`GEPA`**](../../api/optimizers/GEPA/overview.md): Uses LM's to reflect on the DSPy program's trajectory, to identify what worked, what didn't and propose prompts addressing the gaps. Additionally, GEPA can leverage domain-specific textual feedback to rapidly improve the DSPy program. Detailed tutorials on using GEPA are available at [dspy.GEPA Tutorials](../../tutorials/gepa_ai_program/index.md). From bece4c19ac06d10a688898b2dc568c2759e23c27 Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 27 Oct 2025 08:16:53 +0000 Subject: [PATCH 2/2] Improve SIMBA optimizer description based on code analysis MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Refined the SIMBA description to more accurately reflect the implementation: - Clarifies it samples mini-batches (stochastic process) - More precisely describes the two strategies: adding demonstrations vs. generating instructions - Emphasizes the comparison of good vs. bad trajectories for instruction generation 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude --- docs/docs/learn/optimization/optimizers.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/docs/learn/optimization/optimizers.md b/docs/docs/learn/optimization/optimizers.md index 979d7efc0e..1ed0b08bba 100644 --- a/docs/docs/learn/optimization/optimizers.md +++ b/docs/docs/learn/optimization/optimizers.md @@ -58,7 +58,7 @@ These optimizers produce optimal instructions for the prompt and, in the case of 6. [**`MIPROv2`**](../../api/optimizers/MIPROv2.md): Generates instructions *and* few-shot examples in each step. The instruction generation is data-aware and demonstration-aware. Uses Bayesian Optimization to effectively search over the space of generation instructions/demonstrations across your modules. -7. [**`SIMBA`**](../../api/optimizers/SIMBA.md): Uses the LLM to analyze its own performance, identifies challenging examples with high output variability, and generates improvement rules by either creating self-reflective rules or adding successful examples as demonstrations. +7. [**`SIMBA`**](../../api/optimizers/SIMBA.md): Samples mini-batches, identifies challenging examples with high output variability, and improves the program by either adding successful examples as demonstrations or using the LLM to compare good vs. bad trajectories and generate improvement instructions. 8. [**`GEPA`**](../../api/optimizers/GEPA/overview.md): Uses LM's to reflect on the DSPy program's trajectory, to identify what worked, what didn't and propose prompts addressing the gaps. Additionally, GEPA can leverage domain-specific textual feedback to rapidly improve the DSPy program. Detailed tutorials on using GEPA are available at [dspy.GEPA Tutorials](../../tutorials/gepa_ai_program/index.md).