#How do I set embedding provide to I’ll studio

11 messages · Page 1 of 1 (latest)

eternal spindle
#

How do I set my embedding model to this

lmstudio/google/embeddinggemma-300m-qat

sturdy slateBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

ebon ridgeBOT
eternal spindle
#

lmstudio/google/embeddinggemma-300m-qat

#

@shut zephyr

#

How do I make it so the embedding uses this model

lmstudio/google/embeddinggemma-300m-qat

shut zephyrBOT
#

I’m checking the actual docs/source so I don’t hand you a fake config key.

#

Set it under agents.defaults.memorySearch, not your primary chat model.

{
  agents: {
    defaults: {
      memorySearch: {
        provider: "lmstudio",
        model: "lmstudio/google/embeddinggemma-300m-qat",
      },
    },
  },
}

A couple notes:

#
  • provider should be "lmstudio"
  • model can be either:
    • lmstudio/google/embeddinggemma-300m-qat
    • or just google/embeddinggemma-300m-qat

OpenClaw accepts the lmstudio/ prefix for embeddings and normalizes it.

If LM Studio itself is not configured yet, make sure models.providers.lmstudio.baseUrl points to your LM Studio server (usually http://localhost:1234/v1).

After changing it, restart the gateway. Since the embedding model changed, I’d also run a reindex:

openclaw memory index --force

Docs:

#

If you want, paste your current memorySearch block and I can sanity check it.