ben/coze-studio
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: intergrate gemini embedding (#783)

Browse Source
This commit is contained in:
mhowto
2025-08-18 16:44:19 +08:00
committed by GitHub
parent a9b87c188b
commit 85e6926a14
6 changed files with 115 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
docker/.env.debug.example
View File

@@ -101,7 +101,7 @@ export VIKING_DB_MODEL_NAME="" # if vikingdb model name is not set, you need to
# The Embedding model relied on by knowledge base vectorization does not need to be configured
# if the vector database comes with built-in Embedding functionality (such as VikingDB). Currently,
# Coze Studio supports four access methods: openai, ark, ollama, and custom http. Users can simply choose one of them when using
# embedding type: openai / ark / ollama / http
# embedding type: openai / ark / ollama / http / gemini
export EMBEDDING_TYPE="ark"
export EMBEDDING_MAX_BATCH_SIZE=100
@@ -126,6 +126,15 @@ export OLLAMA_EMBEDDING_BASE_URL="" # (string, required) Ollama embedding base_u
export OLLAMA_EMBEDDING_MODEL="" # (string, required) Ollama embedding model
export OLLAMA_EMBEDDING_DIMS="" # (int, required) Ollama embedding dimensions
# gemini embedding
export GEMINI_EMBEDDING_BASE_URL="" # (string, required) Gemini embedding base_url
export GEMINI_EMBEDDING_MODEL="gemini-embedding-001" # (string, required) Gemini embedding model.
export GEMINI_EMBEDDING_API_KEY="" # (string, required) Gemini embedding api_key
export GEMINI_EMBEDDING_DIMS=2048 # (int, required) Gemini embedding dimensions
export GEMINI_EMBEDDING_BACKEND="1" # (string, required) Gemini embedding backend, should be "1" for BackendGeminiAPI / "2" for BackendVertexAI.
export GEMINI_EMBEDDING_PROJECT="" # (string, optional) Gemini embedding project
export GEMINI_EMBEDDING_LOCATION="" # (string, optional) Gemini embedding location
# http embedding
export HTTP_EMBEDDING_ADDR="" # (string, required) http embedding address
export HTTP_EMBEDDING_DIMS=1024 # (string, required) http embedding dimensions

11
docker/.env.example
View File

@@ -99,7 +99,7 @@ export VIKING_DB_MODEL_NAME="" # if vikingdb model name is not set, you need to
# The Embedding model relied on by knowledge base vectorization does not need to be configured
# if the vector database comes with built-in Embedding functionality (such as VikingDB). Currently,
# Coze Studio supports four access methods: openai, ark, ollama, and custom http. Users can simply choose one of them when using
# embedding type: ark / openai / ollama / http
# embedding type: ark / openai / ollama / gemini / http
export EMBEDDING_TYPE="ark"
export EMBEDDING_MAX_BATCH_SIZE=100
@@ -124,6 +124,15 @@ export OLLAMA_EMBEDDING_BASE_URL="" # (string, required) Ollama embedding base_u
export OLLAMA_EMBEDDING_MODEL="" # (string, required) Ollama embedding model
export OLLAMA_EMBEDDING_DIMS="" # (int, required) Ollama embedding dimensions
# gemini embedding
export GEMINI_EMBEDDING_BASE_URL="" # (string, required) Gemini embedding base_url
export GEMINI_EMBEDDING_MODEL="gemini-embedding-001" # (string, required) Gemini embedding model.
export GEMINI_EMBEDDING_API_KEY="" # (string, required) Gemini embedding api_key
export GEMINI_EMBEDDING_DIMS=2048 # (int, required) Gemini embedding dimensions
export GEMINI_EMBEDDING_BACKEND="1" # (string, required) Gemini embedding backend, should be "1" for BackendGeminiAPI / "2" for BackendVertexAI.
export GEMINI_EMBEDDING_PROJECT="" # (string, optional) Gemini embedding project
export GEMINI_EMBEDDING_LOCATION="" # (string, optional) Gemini embedding location
# http embedding
export HTTP_EMBEDDING_ADDR="" # (string, required) http embedding address
export HTTP_EMBEDDING_DIMS=1024 # (string, required) http embedding dimensions