mudler / mudler/LocalRecall

search results on default mode missing embedding

Open
#27 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Go
Stars
974
Forks
121
Avg merge
5h 52m
Merged PRs (30d)
1

Description

Using the default configuration when setting up the LocalAIG system with the default confirm from the example: https://github.com/mudler/LocalAGI/blob/main/docker-compose.yaml

  • Entered to local recall interface
  • Created a new collection
  • Added a 4k txt file into a collection (SecurityAuditCheatSheet.txt)
  • Search from the locall recall search interface on an npm vendor name that lives inside the txt file and have only one occurrence in the whole file.

All 5 search similarities have the same value and the embedded are empty.

Result #1
{
  "ID": "14",
  "Metadata": {
    "source": "SecurityAuditCheatSheet.txt",
    "type": "file"
  },
  "Embedding": null,
  "Content": ".......",
  "Similarity": 0.99999964
}
Result #2
{
  "ID": "16",
  "Metadata": {
    "source": "SecurityAuditCheatSheet.txt",
    "type": "file"
  },
  "Embedding": null,
  "Content": "........",
  "Similarity": 0.99999964
}
Result #3
{
  "ID": "15",
  "Metadata": {
    "source": "SecurityAuditCheatSheet.txt",
    "type": "file"
  },
  "Embedding": null,
  "Content": ".......",
  "Similarity": 0.99999964
}
Result #4
{
  "ID": "13",
  "Metadata": {
    "source": "SecurityAuditCheatSheet.txt",
    "type": "file"
  },
  "Embedding": null,
  "Content": "......",
  "Similarity": 0.99999964
}
Result #5
{
  "ID": "12",
  "Metadata": {
    "source": "SecurityAuditCheatSheet.txt",
    "type": "file"
  },
  "Embedding": null,
  "Content": "......",
  "Similarity": 0.99999964
}

Content returned does not belongs to the section of context where the word id located, nor have the reference for the package searched.

I see the request for the embedded been triggered on on localai docker logs, but is not stored?

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with docker-compose.yaml and reproduce the reported workflow through the local recall search interface while checking the LocalAI Docker logs. Trace why search results have identical similarities and null embeddings; done means the searched package returns relevant content and stored results include usable embeddings.

Written by the indexing model from the issue text.

Assessment

Tech stack
docker, go
Domain
backend, search
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.