langchain-ai/docs
1
0
Fork 0
mirror of https://github.com/langchain-ai/docs.git synced 2026-07-21 11:55:23 -04:00
Code Issues 490 Packages Projects Releases Wiki Activity

[PR #3554] docs: add Sulci Cache — context-aware semantic LLM cache integration #3596

New Issue
Open
opened 2026-06-05 18:23:37 -04:00 by yindo · 0 comments
yindo commented 2026-06-05 18:23:37 -04:00
Owner
Copy Link
Copy Source

📋 Pull Request Information

Original PR: https://github.com/langchain-ai/docs/pull/3554
Author: @sulci-cache
Created: 4/13/2026
Status: 🔄 Open

Base: main ← Head: docs/add-sulci-cache-integration

📝 Commits (5)

  • 9da7e47 docs: add Sulci Cache context-aware semantic LLM cache integration
  • f7fb31b docs: fix Vale DashesSpaces lint error — remove spaces around em dashes
  • 7e2c5df Merge branch 'main' into docs/add-sulci-cache-integration
  • a3595aa Merge branch 'main' into docs/add-sulci-cache-integration
  • 8365b9a Merge branch 'main' into docs/add-sulci-cache-integration

📊 Changes

1 file changed (+194 additions, -0 deletions)

View changed files

➕ src/oss/python/integrations/caches/sulci_llm_caching.mdx (+194 -0)

📄 Description

Summary

Adds documentation for Sulci Cache, a context-aware semantic cache for
LLM apps, to the Model caches integrations section.

What is Sulci Cache

Unlike stateless semantic caches (GPTCache, Redis SemanticCache), Sulci
Cache blends prior conversation turns into the lookup vector at query time:

lookup_vec = α · embed(query) + (1−α) · Σ(decay^i · turn_i)

This solves the multi-turn ambiguity problem — follow-up queries like
"What are the main differences?" resolve correctly based on conversation
history rather than matching against the entire cache blindly.

Benchmark results (800 multi-turn conversation pairs)

Domain Stateless Context-aware Δ
Customer support 32% 88% +56pp
Developer Q&A 80% 96% +16pp
Medical 40% 60% +20pp
Overall 64.0% 81.6% +17.6pp

Resolution accuracy: +20.8pp. Hit latency: 0.74ms p50 (2,486× faster
than a live LLM call).

Integration details

  • No changes to langchain-community required
  • Integration ships inside the sulci package
  • Install: pip install "sulci[sqlite,langchain]"
  • PyPI: https://pypi.org/project/sulci/ (v0.5.6, ~3,000 bot-filtered human installs/month — BigQuery bigquery-public-data.pypi.file_downloads, last 30 days)
  • GitHub: https://github.com/sulci-io/sulci-oss
  • Apache 2.0 licensed

Checklist

  • Follows existing .mdx format (matches redis_llm_caching.mdx structure)
  • Benchmark numbers sourced from published benchmark suite
  • Code examples are runnable
  • No changes to docs.json required (caches folder is auto-discovered)

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/langchain-ai/docs/pull/3554 **Author:** [@sulci-cache](https://github.com/sulci-cache) **Created:** 4/13/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `docs/add-sulci-cache-integration` --- ### 📝 Commits (5) - [`9da7e47`](https://github.com/langchain-ai/docs/commit/9da7e478f51b0fd962c2401309fb271834f82d74) docs: add Sulci Cache context-aware semantic LLM cache integration - [`f7fb31b`](https://github.com/langchain-ai/docs/commit/f7fb31b6f60b4fd1fd003ff5694fdbf23f568f01) docs: fix Vale DashesSpaces lint error — remove spaces around em dashes - [`7e2c5df`](https://github.com/langchain-ai/docs/commit/7e2c5df4047a5d0d87a23c23fb68cd949489a01d) Merge branch 'main' into docs/add-sulci-cache-integration - [`a3595aa`](https://github.com/langchain-ai/docs/commit/a3595aa7ff70604c03216a7d730a293892a42b92) Merge branch 'main' into docs/add-sulci-cache-integration - [`8365b9a`](https://github.com/langchain-ai/docs/commit/8365b9a2ef406bd2f98efa2e547baf525e44e671) Merge branch 'main' into docs/add-sulci-cache-integration ### 📊 Changes **1 file changed** (+194 additions, -0 deletions) <details> <summary>View changed files</summary> ➕ `src/oss/python/integrations/caches/sulci_llm_caching.mdx` (+194 -0) </details> ### 📄 Description ## Summary Adds documentation for Sulci Cache, a context-aware semantic cache for LLM apps, to the Model caches integrations section. ## What is Sulci Cache Unlike stateless semantic caches (GPTCache, Redis SemanticCache), Sulci Cache blends prior conversation turns into the lookup vector at query time: lookup_vec = α · embed(query) + (1−α) · Σ(decay^i · turn_i) This solves the multi-turn ambiguity problem — follow-up queries like "What are the main differences?" resolve correctly based on conversation history rather than matching against the entire cache blindly. ## Benchmark results (800 multi-turn conversation pairs) | Domain | Stateless | Context-aware | Δ | |---|---|---|---| | Customer support | 32% | 88% | +56pp | | Developer Q&A | 80% | 96% | +16pp | | Medical | 40% | 60% | +20pp | | Overall | 64.0% | 81.6% | +17.6pp | Resolution accuracy: +20.8pp. Hit latency: 0.74ms p50 (2,486× faster than a live LLM call). ## Integration details - No changes to langchain-community required - Integration ships inside the `sulci` package - Install: `pip install "sulci[sqlite,langchain]"` - PyPI: https://pypi.org/project/sulci/ (v0.5.6, ~3,000 bot-filtered human installs/month — BigQuery `bigquery-public-data.pypi.file_downloads`, last 30 days) - GitHub: https://github.com/sulci-io/sulci-oss - Apache 2.0 licensed ## Checklist - [x] Follows existing .mdx format (matches redis_llm_caching.mdx structure) - [x] Benchmark numbers sourced from published benchmark suite - [x] Code examples are runnable - [x] No changes to docs.json required (caches folder is auto-discovered) --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
yindo added the pull-request label 2026-06-05 18:23:37 -04:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
deepagents
docs-infra
external
external
help wanted
integration
internal
internal
javascript
js/ts
langchain
langgraph
langsmith
needs-title
pull-request
Mirrored from GitHub Pull Request
 python
reference
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/docs#3596
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?