SciPhi Quickstart#

Welcome to the SciPhi quickstart guide! SciPhi, or ΨΦ, is your portal to using large language models (LLMs) like OpenAI’s models, Anthropic, HuggingFace, and vLLM, combined with the power of Retriever-Augmented Generation (RAG).

This guide will introduce you to: - Generating data tailored to your needs. - Using the RAG provider interface. - Creating RAG-enhanced textbooks. - Evaluating your RAG pipeline.

Let’s get started!

Setting Up Your Environment#

Before you start, ensure you’ve installed SciPhi:

pip install sciphi

For additional details, refer to the installation guide.

Instantiate Your LLM and RAG Provider#

Here’s a simple example of how you can utilize SciPhi to work with your own LLM and RAG provider:

from sciphi.core import LLMProviderName, RAGProviderName
from sciphi.interface import LLMInterfaceManager, RAGInterfaceManager
from sciphi.llm import GenerationConfig

# Define your parameters here...

# RAG Provider Settings
rag_interface = (
    RAGInterfaceManager.get_interface_from_args(
        RAGProviderName(rag_provider_name),
        api_base=rag_api_base or llm_api_base,
        api_key=rag_api_key or llm_api_key,
        top_k=rag_top_k,
    )
    if rag_enabled
    else None
)

# LLM Provider Settings
llm_interface = LLMInterfaceManager.get_interface_from_args(
    LLMProviderName(llm_provider_name),
    api_key=llm_api_key,
    api_base=llm_api_base,
    rag_interface=rag_interface,
    model_name=llm_model_name,
)

# Set up typical LLM generation settings
completion_config = GenerationConfig(
    temperature=llm_temperature,
    top_k=llm_top_k,
    max_tokens_to_sample=llm_max_tokens_to_sample,
    model_name=llm_model_name,
    skip_special_tokens=llm_skip_special_tokens,
    stop_token=SciPhiFormatter.INIT_PARAGRAPH_TOKEN,
)

# Get the completion for a prompt
completion = llm_interface.get_completion(prompt, generation_config)

# Continue with your process...

This example showcases the flexibility and power of SciPhi, allowing you to seamlessly integrate various LLM and RAG providers into your applications.

Generating Completions with SciPhi#

SciPhi supports multiple LLM providers (e.g. OpenAI, Anthropic, HuggingFace, and vLLM) and RAG providers (e.g. SciPhi). To run an example completion with SciPhi the code shown above, execute:

`bash python -m sciphi.scripts.sciphi_gen_completion -llm_provider_name=sciphi --llm_api_key=YOUR_SCIPHI_API_KEY --llm_api_base=https://api.sciphi.ai/v1 --rag_api_base=https://api.sciphi.ai --llm_model_name=SciPhi/SciPhi-Self-RAG-Mistral-7B-32k --query="Write a few paragraphs on general relativity. Include the mathematical definition of Einsteins field equation in your writeup." `

Generating Data with SciPhi#

To generate data tailored to your specifications, you can use the provided scripts. For instance, to generate a dataset with a desired number of samples:

python -m sciphi.scripts.data_augmenter --config-path=$PWD/sciphi/config/prompts/question_and_answer.yaml --config_name=None --n_samples=1

Inspecting the output:

{"question": "What is the reaction called when alcohol and carboxylic acids react?", "answer": "Fischer esterification"}
...
{"question": "Are tertiary alcohols resistant to oxidation?", "answer": "Yes"}

This command can be readily expanded to other configurations.

RAG-Enhanced Textbooks#

With SciPhi, you can generate textbooks with the assistance of RAG. To perform a dry-run:

python -m sciphi.scripts.textbook_generator dry_run --toc_dir=sciphi/data/sample/table_of_contents --rag-enabled=False

To generate a textbook:

python -m sciphi.scripts.textbook_generator run --toc_dir=sciphi/data/sample/table_of_contents --rag-enabled=False --filter_existing_books=False

You can also use a custom table of contents:

python -m sciphi.scripts.textbook_generator run --toc_dir=toc --output_dir=books --data_dir=$PWD

RAG Evaluation#

Measure the efficacy of your RAG pipeline using SciPhi’s evaluation harness:

python -m sciphi.scripts.rag_harness --n-samples=100 --rag-enabled=True --evals_to_run="science_multiple_choice"

This will evaluate your RAG over a set of questions and report the final accuracy.

Wrapping Up#

Congratulations! You’ve now been introduced to the core functionalities of SciPhi. This is just the beginning; delve deeper into the documentation, explore the community on Discord, or reach out for tailored inquiries. Happy modeling!

SciPhi Quickstart

Contents