Prompting technique

Context Repetition (CoRe)

Unleashing Multi-Hop Reasoning Potential in Large Language Models through Repetition of Misordered Context, arXiv:2410.07103 (2024) · read the paper

What it does

Context Repetition (CoRe) method involves two or three-time repetition of the entire source data block within a single prompt. This allows the model to discover connections between facts when re-reading context that were missed due to unfavorable information order, increasing accuracy by 30-70% in multi-step analysis tasks.

Simple two or three-time repetition of source data in a prompt can improve answer accuracy in tasks requiring multi-step analysis by 30-70%.checked against the paper
Task typemulti step
Shapesingle shot
Effortsingle step
Addressesthe model losing what was agreed earlier; answers that are close but wrong; the same prompt giving different answers
Published2024-10-09
AuthorsSangwon Yu, Ik-hwan Kim, Jongyoon Song et al.

When it helps

Use CoRe when the model needs to match facts from different parts of a long context for multi-step reasoning. Particularly effective in route planning, analyzing documents with fragmented information, solving tasks requiring data synthesis from multiple sources. The method helps when information presentation order might be suboptimal, and the model misses important fact connections. Ideal for tasks where accuracy is critical and prompt length increase by 2-3 times is acceptable.

A prompt you can paste

Generic by design: it applies the technique without knowing your task. Adapt the marked parts.

Prompt
# Role
You are an AI assistant tasked with analyzing customer feedback to identify key issues and propose solutions.

# Task
Analyze the provided customer feedback to identify the primary problem and suggest actionable improvements.

# Context (Feedback Data)
[INSERT CONTEXT HERE - e.g., customer reviews, survey responses, support tickets]

# REPEAT CONTEXT FOR ENHANCED REASONING:

# Context (Feedback Data)
[INSERT CONTEXT HERE - e.g., customer reviews, survey responses, support tickets]

# Output Format
1.  **Primary Problem:** State the most critical issue identified from the feedback in a single, concise sentence.
2.  **Supporting Evidence:** Provide 2-3 direct quotes from the feedback that strongly support the identified primary problem.
3.  **Actionable Recommendation:** Propose one immediate, actionable step to address the primary problem.

If this one does not fit, the two closest alternatives in the corpus are Debugging Prompts Framework and MTR (Multi-Turn Reasoning) Framework, which target the same failure from a different angle.

Worked example

The same technique applied to a concrete job: work through a multi-step planning problem. Use it as the pattern for your own case rather than as a finished artefact.

Worked example
**Role:** You are an AI assistant tasked with planning a multi-step process.

**Task:** Plan a 7-day trip to Italy for a family of three, ensuring all their diverse requirements are met.

**Context (Family Requirements):**
- **Husband (Andrey):**
    - Desires: Visit at least 2 historical cities (e.g., Rome, Florence).
    - Constraints: Hotel budget not exceeding 150 euros per night. Dislikes beach holidays.
- **Wife (Maria):**
    - Desires: Spend at least 3 days by the sea on a sandy beach. Wants to attend a culinary masterclass.
    - Passion: Loves Italian cuisine.
- **Son (10 years old, Lev):**
    - Desires: Visit an amusement park or a large water park.
    - Constraints: Cannot endure long walking tours.

**CoRe (Context Repetition) for Multi-Hop Reasoning:**

**First Pass Context:**
- **Husband (Andrey):** Visit 2+ historical cities (Rome, Florence). Hotels <= 150€/night. No beaches.
- **Wife (Maria):** 3+ days at a sandy beach. Culinary masterclass. Loves Italian food.
- **Son (Lev, 10):** Amusement/water park. Avoids long walking tours.

**Second Pass Context (for enhanced reasoning):**
- **Husband (Andrey):** Visit 2+ historical cities (Rome, Florence). Hotels <= 150€/night. No beaches.
- **Wife (Maria):** 3+ days at a sandy beach. Culinary masterclass. Loves Italian food.
- **Son (Lev, 10):** Amusement/water park. Avoids long walking tours.

**Execution Plan:**
1.  **Synthesize Constraints:** Identify potential conflicts and synergies between Andrey's, Maria's, and Lev's requirements.
2.  **Geographic Planning:** Propose a logical 7-day itinerary in Italy that balances historical city visits, beach time, and proximity to parks/activities. Consider travel time between locations.
3.  **Accommodation Selection:** For each location, suggest hotel options that fit Andrey's budget (<= 150€/night) and Maria's preference for beach proximity where applicable.
4.  **Activity Integration:** Schedule specific activities, including historical tours (adapted for Lev's limitations), beach days, a culinary masterclass, and an amusement/water park visit.
5.  **Final Itinerary:** Present a day-by-day plan detailing locations, accommodation type, and key activities.

**Output Format:**
- A structured 7-day itinerary.
- For each day: Location, Accommodation type/suggestion, Key Activities.
- A brief explanation of how the itinerary addresses each family member's core needs and constraints.

Get this written for your actual task

Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.

Free · no signup · ~10s
0.00match confidence
single retrieval pass
Prompt for your task

      

That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.

6,235techniques in the corpus
one of which is this page

Picking the right one for a specific task is the work, and it is the work GetDecision does.

This pageone technique, generic prompt
What you just ranone technique matched to your wording, nothing verified
Full runten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history

See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.

Run the full analysis

Related techniques

Debugging Prompts FrameworkA systematic framework for diagnosing and preventing typical errors of language models in instruction understa…MTR (Multi-Turn Reasoning) FrameworkA method of structuring multi-step interaction with an LLM through a "game with rules" format. Includes four b…StateAct (Self-Prompting + Chain-of-States)StateAct improves LLM agent performance through two mechanisms: continuous reminder of the end goal (self-prom…Dynamic Enhancement Chain (DEC)A method that breaks down a complex question into a chain of simple sub-questions solved sequentially. Each st…

All techniques · Failure modes and fixes