Gradient Embeddings

Gradient offers embeddings model that can be easily integrated with LlamaIndex. Below is an example of how to use it with LlamaIndex.

%pip install llama-index-llms-gradient
%pip install llama-index-embeddings-gradient

# Install the required packages

%pip install llama-index --quiet
%pip install gradientai --quiet

Gradient needs an access token and workspaces id for authorization. They can be obtained from:

• Gradient UI, or

• Gradient CLI with gradient env command.

import os

os.environ["GRADIENT_ACCESS_TOKEN"] = "{GRADIENT_ACCESS_TOKEN}"
os.environ["GRADIENT_WORKSPACE_ID"] = "{GRADIENT_WORKSPACE_ID}"

from llama_index.llms.gradient import GradientBaseModelLLM

# NOTE: we use a base model here, you can as well insert your fine-tuned model.
llm = GradientBaseModelLLM(
    base_model_slug="llama2-7b-chat",
    max_tokens=400,
)

Download Data

!mkdir -p 'data/paul_graham/'
!wget 'https://raw.githubusercontent.com/run-llama/llama_index/main/docs/examples/data/paul_graham/paul_graham_essay.txt' -O 'data/paul_graham/paul_graham_essay.txt'

Load Documents

from llama_index.core import SimpleDirectoryReader

documents = SimpleDirectoryReader("./data/paul_graham").load_data()
print(f"Loaded {len(documents)} document(s).")

Configure Gradient embeddings

from llama_index.embeddings.gradient import GradientEmbedding
from llama_index.core import Settings

embed_model = GradientEmbedding(
    gradient_access_token=os.environ["GRADIENT_ACCESS_TOKEN"],
    gradient_workspace_id=os.environ["GRADIENT_WORKSPACE_ID"],
    gradient_model_slug="bge-large",
)

Settings.embed_model = embed_model
Settings.llm = llm
Settings.chunk_size = 1024

Setup and Query Index

from llama_index.core import VectorStoreIndex

index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()

response = query_engine.query(
    "What did the author do after his time at Y Combinator?"
)
print(response)