From 01cf2fed170de0f7ceec134709aa2b87c3bc8135 Mon Sep 17 00:00:00 2001
From: Timothy Carambat <rambat1010@gmail.com>
Date: Thu, 16 May 2024 17:25:05 -0700
Subject: [PATCH] Make native embedder the fallback for all LLMs (#1427)
---
server/utils/AiProviders/anthropic/index.js | 7 ++-----
server/utils/AiProviders/azureOpenAi/index.js | 8 ++------
server/utils/AiProviders/cohere/index.js | 3 ++-
server/utils/AiProviders/gemini/index.js | 7 ++-----
server/utils/AiProviders/genericOpenAi/index.js | 6 +-----
server/utils/AiProviders/groq/index.js | 2 +-
server/utils/AiProviders/huggingface/index.js | 7 +------
server/utils/AiProviders/koboldCPP/index.js | 6 +-----
server/utils/AiProviders/liteLLM/index.js | 6 +-----
server/utils/AiProviders/lmStudio/index.js | 7 ++-----
server/utils/AiProviders/localAi/index.js | 7 ++-----
server/utils/AiProviders/mistral/index.js | 7 ++-----
server/utils/AiProviders/native/index.js | 2 +-
server/utils/AiProviders/ollama/index.js | 7 ++-----
server/utils/AiProviders/openAi/index.js | 8 ++------
server/utils/AiProviders/openRouter/index.js | 2 +-
server/utils/AiProviders/perplexity/index.js | 2 +-
server/utils/AiProviders/textGenWebUI/index.js | 2 +-
server/utils/AiProviders/togetherAi/index.js | 7 ++-----
19 files changed, 29 insertions(+), 74 deletions(-)
diff --git a/server/utils/AiProviders/anthropic/index.js b/server/utils/AiProviders/anthropic/index.js
index d5ee1f9d3..215fa5fbb 100644
--- a/server/utils/AiProviders/anthropic/index.js
+++ b/server/utils/AiProviders/anthropic/index.js
@@ -3,6 +3,7 @@ const {
writeResponseChunk,
clientAbortedHandler,
} = require("../../helpers/chat/responses");
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
class AnthropicLLM {
constructor(embedder = null, modelPreference = null) {
@@ -23,11 +24,7 @@ class AnthropicLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- throw new Error(
- "INVALID ANTHROPIC SETUP. No embedding engine has been set. Go to instance settings and set up an embedding interface to use Anthropic as your LLM."
- );
- this.embedder = embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/azureOpenAi/index.js b/server/utils/AiProviders/azureOpenAi/index.js
index a2ab556db..3678567e3 100644
--- a/server/utils/AiProviders/azureOpenAi/index.js
+++ b/server/utils/AiProviders/azureOpenAi/index.js
@@ -1,4 +1,4 @@
-const { AzureOpenAiEmbedder } = require("../../EmbeddingEngines/azureOpenAi");
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
writeResponseChunk,
clientAbortedHandler,
@@ -23,11 +23,7 @@ class AzureOpenAiLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for AzureOpenAiLLM - falling back to AzureOpenAiEmbedder for embedding!"
- );
- this.embedder = !embedder ? new AzureOpenAiEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/cohere/index.js b/server/utils/AiProviders/cohere/index.js
index a97a15fca..6357d503e 100644
--- a/server/utils/AiProviders/cohere/index.js
+++ b/server/utils/AiProviders/cohere/index.js
@@ -19,7 +19,8 @@ class CohereLLM {
system: this.promptWindowLimit() * 0.15,
user: this.promptWindowLimit() * 0.7,
};
- this.embedder = !!embedder ? embedder : new NativeEmbedder();
+
+ this.embedder = embedder ?? new NativeEmbedder();
}
#appendContext(contextTexts = []) {
diff --git a/server/utils/AiProviders/gemini/index.js b/server/utils/AiProviders/gemini/index.js
index 962b12013..3dd307aff 100644
--- a/server/utils/AiProviders/gemini/index.js
+++ b/server/utils/AiProviders/gemini/index.js
@@ -1,3 +1,4 @@
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
writeResponseChunk,
clientAbortedHandler,
@@ -26,11 +27,7 @@ class GeminiLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- throw new Error(
- "INVALID GEMINI LLM SETUP. No embedding engine has been set. Go to instance settings and set up an embedding interface to use Gemini as your LLM."
- );
- this.embedder = embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7; // not used for Gemini
}
diff --git a/server/utils/AiProviders/genericOpenAi/index.js b/server/utils/AiProviders/genericOpenAi/index.js
index 686d4c677..dc0264e48 100644
--- a/server/utils/AiProviders/genericOpenAi/index.js
+++ b/server/utils/AiProviders/genericOpenAi/index.js
@@ -27,11 +27,7 @@ class GenericOpenAiLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for GenericOpenAiLLM - falling back to NativeEmbedder for embedding!"
- );
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
this.log(`Inference API: ${this.basePath} Model: ${this.model}`);
}
diff --git a/server/utils/AiProviders/groq/index.js b/server/utils/AiProviders/groq/index.js
index 01d92f006..24a430e62 100644
--- a/server/utils/AiProviders/groq/index.js
+++ b/server/utils/AiProviders/groq/index.js
@@ -20,7 +20,7 @@ class GroqLLM {
user: this.promptWindowLimit() * 0.7,
};
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/huggingface/index.js b/server/utils/AiProviders/huggingface/index.js
index 6a79880c8..d25725c2a 100644
--- a/server/utils/AiProviders/huggingface/index.js
+++ b/server/utils/AiProviders/huggingface/index.js
@@ -1,5 +1,4 @@
const { NativeEmbedder } = require("../../EmbeddingEngines/native");
-const { OpenAiEmbedder } = require("../../EmbeddingEngines/openAi");
const {
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
@@ -26,11 +25,7 @@ class HuggingFaceLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for HuggingFaceLLM - falling back to Native for embedding!"
- );
- this.embedder = !embedder ? new OpenAiEmbedder() : new NativeEmbedder();
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.2;
}
diff --git a/server/utils/AiProviders/koboldCPP/index.js b/server/utils/AiProviders/koboldCPP/index.js
index 4b1ff3f61..90186aabf 100644
--- a/server/utils/AiProviders/koboldCPP/index.js
+++ b/server/utils/AiProviders/koboldCPP/index.js
@@ -26,11 +26,7 @@ class KoboldCPPLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for KoboldCPPLLM - falling back to NativeEmbedder for embedding!"
- );
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
this.log(`Inference API: ${this.basePath} Model: ${this.model}`);
}
diff --git a/server/utils/AiProviders/liteLLM/index.js b/server/utils/AiProviders/liteLLM/index.js
index 5973826cc..6152ea296 100644
--- a/server/utils/AiProviders/liteLLM/index.js
+++ b/server/utils/AiProviders/liteLLM/index.js
@@ -26,11 +26,7 @@ class LiteLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for LiteLLM - falling back to NativeEmbedder for embedding!"
- );
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
this.log(`Inference API: ${this.basePath} Model: ${this.model}`);
}
diff --git a/server/utils/AiProviders/lmStudio/index.js b/server/utils/AiProviders/lmStudio/index.js
index 48f689fbc..6345fb04e 100644
--- a/server/utils/AiProviders/lmStudio/index.js
+++ b/server/utils/AiProviders/lmStudio/index.js
@@ -1,3 +1,4 @@
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
@@ -27,11 +28,7 @@ class LMStudioLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- throw new Error(
- "INVALID LM STUDIO SETUP. No embedding engine has been set. Go to instance settings and set up an embedding interface to use LMStudio as your LLM."
- );
- this.embedder = embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/localAi/index.js b/server/utils/AiProviders/localAi/index.js
index 504775285..8bc4d047d 100644
--- a/server/utils/AiProviders/localAi/index.js
+++ b/server/utils/AiProviders/localAi/index.js
@@ -1,3 +1,4 @@
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
@@ -19,11 +20,7 @@ class LocalAiLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- throw new Error(
- "INVALID LOCAL AI SETUP. No embedding engine has been set. Go to instance settings and set up an embedding interface to use LocalAI as your LLM."
- );
- this.embedder = embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/mistral/index.js b/server/utils/AiProviders/mistral/index.js
index 8410d4cb6..c2546299f 100644
--- a/server/utils/AiProviders/mistral/index.js
+++ b/server/utils/AiProviders/mistral/index.js
@@ -1,3 +1,4 @@
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
@@ -20,11 +21,7 @@ class MistralLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for MistralLLM - falling back to OpenAiEmbedder for embedding!"
- );
- this.embedder = embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.0;
}
diff --git a/server/utils/AiProviders/native/index.js b/server/utils/AiProviders/native/index.js
index e13b68a2f..0cbd87476 100644
--- a/server/utils/AiProviders/native/index.js
+++ b/server/utils/AiProviders/native/index.js
@@ -23,7 +23,7 @@ class NativeLLM {
system: this.promptWindowLimit() * 0.15,
user: this.promptWindowLimit() * 0.7,
};
- this.embedder = embedder || new NativeEmbedder();
+ this.embedder = embedder ?? new NativeEmbedder();
this.cacheDir = path.resolve(
process.env.STORAGE_DIR
? path.resolve(process.env.STORAGE_DIR, "models", "downloaded")
diff --git a/server/utils/AiProviders/ollama/index.js b/server/utils/AiProviders/ollama/index.js
index 73269d6d2..609a7601b 100644
--- a/server/utils/AiProviders/ollama/index.js
+++ b/server/utils/AiProviders/ollama/index.js
@@ -3,6 +3,7 @@ const {
writeResponseChunk,
clientAbortedHandler,
} = require("../../helpers/chat/responses");
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
// Docs: https://github.com/jmorganca/ollama/blob/main/docs/api.md
class OllamaAILLM {
@@ -18,11 +19,7 @@ class OllamaAILLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- throw new Error(
- "INVALID OLLAMA SETUP. No embedding engine has been set. Go to instance settings and set up an embedding interface to use Ollama as your LLM."
- );
- this.embedder = embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/openAi/index.js b/server/utils/AiProviders/openAi/index.js
index 43d7c2874..8037d23dc 100644
--- a/server/utils/AiProviders/openAi/index.js
+++ b/server/utils/AiProviders/openAi/index.js
@@ -1,4 +1,4 @@
-const { OpenAiEmbedder } = require("../../EmbeddingEngines/openAi");
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
@@ -18,11 +18,7 @@ class OpenAiLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- console.warn(
- "No embedding provider defined for OpenAiLLM - falling back to OpenAiEmbedder for embedding!"
- );
- this.embedder = !embedder ? new OpenAiEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/openRouter/index.js b/server/utils/AiProviders/openRouter/index.js
index a83010835..bb8cf4471 100644
--- a/server/utils/AiProviders/openRouter/index.js
+++ b/server/utils/AiProviders/openRouter/index.js
@@ -36,7 +36,7 @@ class OpenRouterLLM {
user: this.promptWindowLimit() * 0.7,
};
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
if (!fs.existsSync(cacheFolder))
diff --git a/server/utils/AiProviders/perplexity/index.js b/server/utils/AiProviders/perplexity/index.js
index a17ec43f5..d3f50de75 100644
--- a/server/utils/AiProviders/perplexity/index.js
+++ b/server/utils/AiProviders/perplexity/index.js
@@ -28,7 +28,7 @@ class PerplexityLLM {
user: this.promptWindowLimit() * 0.7,
};
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
}
diff --git a/server/utils/AiProviders/textGenWebUI/index.js b/server/utils/AiProviders/textGenWebUI/index.js
index dfce76a5a..72f116f85 100644
--- a/server/utils/AiProviders/textGenWebUI/index.js
+++ b/server/utils/AiProviders/textGenWebUI/index.js
@@ -23,7 +23,7 @@ class TextGenWebUILLM {
user: this.promptWindowLimit() * 0.7,
};
- this.embedder = !embedder ? new NativeEmbedder() : embedder;
+ this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
this.log(`Inference API: ${this.basePath} Model: ${this.model}`);
}
diff --git a/server/utils/AiProviders/togetherAi/index.js b/server/utils/AiProviders/togetherAi/index.js
index 577a4b742..cdfef3397 100644
--- a/server/utils/AiProviders/togetherAi/index.js
+++ b/server/utils/AiProviders/togetherAi/index.js
@@ -1,3 +1,4 @@
+const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
@@ -23,11 +24,7 @@ class TogetherAiLLM {
user: this.promptWindowLimit() * 0.7,
};
- if (!embedder)
- throw new Error(
- "INVALID TOGETHER AI SETUP. No embedding engine has been set. Go to instance settings and set up an embedding interface to use Together AI as your LLM."
- );
- this.embedder = embedder;
+ this.embedder = !embedder ? new NativeEmbedder() : embedder;
this.defaultTemp = 0.7;
}
--
GitLab