🚀 FloTorch-core
FloTorch-core is a modular and extensible Python framework for building LLM-powered RAG (Retrieval-Augmented Generation) pipelines. It offers plug-and-play components for embeddings, chunking, retrieval, gateway-based LLM calls, and RAG evaluation.
✨ Features
- 🧩 Text Chunking (Fixed-size, Hierarchical)
- 🧠 Embedding Models (Titan, Cohere, Bedrock)
- 🔍 Document Retrieval (OpenSearch + Vector Storage)
- 💻 Bedrock/sagemaker/gateway inferencer
- 🔌 Unified LLM Gateway (OpenAI, Bedrock, Ollama, etc.)
- 📏 RAG Evaluation (RAGAS Metrics)
- ☁️ AWS Integration (S3, DynamoDB, Lambda)
- 🧢 Built-in Testing Support
📆 Installation
pip install FloTorch-core
To install development dependencies:
pip install FloTorch-core[dev]
📂 Project Structure
flotorch/
├── inferencer/ # LLM gateway/bedrock/sagemaker interface
├── embedding/ # Embedding models
├── chunking/ # Text chunking logic
├── evaluator/ # RAG evaluation (RAGAS)
├── storage/ # Vector DB, S3, DynamoDB
├── util/ # Utilities and helpers
├── rerank/ # Ranking documents
├── guardrails/ # Enabling guardrails
├── reader/ # reader for json/pdf
📖 Usage Example
Reader
from flotorch_core.reader.json_reader import JSONReader
from flotorch_core.storage.s3_storage import S3StorageProvider
json_reader = JSONReader(S3StorageProvider(<S3 bucket>))
json_reader.read(<path>)
Embedding
from flotorch_core.embedding.embedding_registry import embedding_registry
embedding_class = embedding_registry.get_model(<model id>)
# model id example: amazon.titan-text-express-v1, amazon.titan-embed-text-v2:0, cohere.embed-multilingual-v3
Vector storage (opensearch)
from flotorch_core.storage.db.vector.open_search import OpenSearchClient
vector_storage_object = OpenSearchClient(
<opensearch_host>,
<opensearch_port>,
<opensearch_username>,
<opensearch_password>,
<index_id>,
<embedding object>
)
Vector storage (bedrock knowledgebase)
from flotorch_core.storage.db.vector.bedrock_knowledgebase_storage import BedrockKnowledgeBaseStorage
vector_storage_object = BedrockKnowledgeBaseStorage(
knowledge_base_id=<knowledge_base_id>,
region=<aws_region>
)
Guardrails over vector storage
from flotorch_core.storage.db.vector.guardrails_vector_storage import GuardRailsVectorStorage
base_guardrails = BedrockGuardrail(<guardrail_id>, <guardrail_version>, <aws_region>)
vector_storage_object = GuardRailsVectorStorage(
vector_storage_object,
base_guardrails,
<enable_prompt_guardrails(True/False)>,
<enable_context_guardrails(True/False)>
)
Inferencer
from flotorch_core.inferencer.bedrock_inferencer import BedrockInferencer
from flotorch_core.inferencer.gateway_inferencer import GatewayInferencer
from flotorch_core.inferencer.sagemaker_inferencer import SageMakerInferencer
inferencer = BedrockInferencer(
<model_id>,
<region>,
<number of n_shot_prompts>,
<temperature>,
<n_shot_prompt_guide_obj>
)
inferencer = GatewayInferencer(
model_id=<model_id>,
api_key=<api_key>,
base_url=<base_url>,
n_shot_prompts=<n_shot_prompts>,
n_shot_prompt_guide_obj=<n_shot_prompt_guide_obj>
)
inferencer = SageMakerInferencer(
<model_id>,
<region>,
<arn_role>,
<n_shot_prompts>,
<temperature>,
<n_shot_prompt_guide_obj>
)
GuardRail over inferencer
from flotorch_core.inferencer.guardrails.guardrails_inferencer import GuardRailsInferencer
inferencer = GuardRailsInferencer(inferencer, base_guardrails)
📬 Maintainer
Shiva Krishna
📧 Email: shiva.krishnaah@gmail.com
Adil Raza
📧 Email: adilraza.9752@gmail.com
📄 License
This project is licensed under the MIT License.
🌐 Links
Metadata
Release files for FloTorch-core 2.10.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| flotorch_core-2.10.0.tar.gz | 65.4 kB | Without provenance | Details for flotorch_core-2.10.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| flotorch_core-2.10.0-py3-none-any.whl | Python 3 | none | any | Details for flotorch_core-2.10.0-py3-none-any.whl |
Total release size: 163.3 kB
Release files / flotorch_core-2.10.0.tar.gz
| Download URL | flotorch_core-2.10.0.tar.gz |
|---|---|
| Size | 65.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
d329082ac923e70c54c3e691bc255758a9e567cb60c46f4f44286174e4f4bd67
|
|
BLAKE2b-256 checksum How to use checksums |
ebe5f8ca052ba371e25337a3853e266e0db4b4d782c77400f01f99467ac2f785
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.10
|
Release files / flotorch_core-2.10.0-py3-none-any.whl
| Download URL | flotorch_core-2.10.0-py3-none-any.whl |
|---|---|
| Size | 97.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
a0443125b8585a5ceada49a8411d6833e8375b66de1d8d82003887bce516f90f
|
|
BLAKE2b-256 checksum How to use checksums |
9b13b4cc1c06910d8863cee9c1dbc23706738e04ef06125530dd0904909a0d62
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.10
|
Security for FloTorch-core 2.10.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.