Open Source Python Large Language Models (LLM) - Page 8
Sort By:
Python Large Language Models (LLM)
View 367 business solutionsBrowse free open source Python Large Language Models (LLM) and projects below. Use the toggles on the left to filter open source Python Large Language Models (LLM) by OS, license, language, programming language, and project status.
-
Earn up to 16% annual interest with Nexo.Generate interest, access liquidity without selling, and execute trades seamlessly. All in one platform. Geographic restrictions, eligibility, and terms apply.
-
AI-powered service management for IT and enterprise teamsGive your IT, operations, and business teams the ability to deliver exceptional services—without the complexity. Maximize operational efficiency with refreshingly simple, AI-powered Freshservice.
-
1
LLaMA-MoE
Building Mixture-of-Experts from LLaMA with Continual Pre-training
LLaMA-MoE is an open-source project that builds mixture-of-experts language models from LLaMA through expert partitioning and continual pre-training. The repository is centered on making MoE research more accessible by offering smaller and more affordable models with only about 3.0 to 3.5 billion activated parameters, which helps reduce deployment and experimentation costs. Its architecture works by splitting LLaMA feed-forward networks into sparse experts and adding gating mechanisms so that only selected experts are activated during inference and training. The project is not just a model release, but also a research framework that includes multiple expert construction methods, several gating strategies, and tooling for continual pre-training on filtered SlimPajama-based datasets. It also emphasizes training efficiency through features such as FlashAttention-v2 integration and fast streaming dataset loading, which are important for large-scale experimentation. -
2
MathModelAgent
An Agent Designed for Mathematical Modeling
MathModelAgent is an AI agent system designed specifically for assisting with mathematical modeling tasks and academic problem solving. The platform automates the process of analyzing mathematical problems, constructing models, generating code for simulations or computations, and producing a complete research-style report. The project uses a multi-agent architecture where different specialized agents handle tasks such as problem interpretation, modeling design, programming implementation, and paper writing. Through integration with multiple large language models, the system can coordinate these components to generate structured modeling solutions and formatted research papers suitable for submission. The platform also includes a code execution environment that allows generated programs to be tested, corrected, and refined during the modeling workflow. -
3
MiniMind
Train a 26M-parameter GPT from scratch in just 2h
minimind is a framework that enables users to train a 26-million-parameter GPT (Generative Pre-trained Transformer) model from scratch in approximately two hours. It provides a streamlined process for data preparation, model training, and evaluation, making it accessible for individuals and organizations to develop their own language models without extensive computational resources. -
4
MobileLLM
MobileLLM Optimizing Sub-billion Parameter Language Models
MobileLLM is a lightweight large language model (LLM) framework developed by Facebook Research, optimized for on-device deployment where computational and memory efficiency are critical. Introduced in the ICML 2024 paper “MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases”, it focuses on delivering strong reasoning and generalization capabilities in models under one billion parameters. The framework integrates several architectural innovations—SwiGLU activation, deep and thin network design, embedding sharing, and grouped-query attention (GQA)—to achieve a superior trade-off between model size, inference speed, and accuracy. MobileLLM demonstrates remarkable performance, with the 125M and 350M variants outperforming previous state-of-the-art models of the same scale by up to 4.3% on zero-shot commonsense reasoning tasks. -
Train ML Models With SQL You Already KnowBuild and deploy ML models using familiar SQL. Automate data prep with built-in Gemini. Query 1 TB and store 10 GB free monthly.
-
5
ModernBERT
Bringing BERT into modernity via both architecture changes and scaling
ModernBERT is an open-source research project that modernizes the classic BERT encoder architecture by incorporating recent advances in transformer design, training techniques, and efficiency improvements. The goal of the project is to bring BERT-style models up to date with the capabilities of modern large language models while preserving the strengths of bidirectional encoder architectures used for tasks such as classification, retrieval, and semantic search. ModernBERT introduces architectural improvements that enhance both training efficiency and inference performance, making the model more suitable for modern large-scale machine learning pipelines. The repository also includes FlexBERT, a modular framework that allows developers to experiment with different encoder building blocks and configurations when constructing new models. -
6
Mosec
A high-performance ML model serving framework, offers dynamic batching
Mosec is a high-performance and flexible model-serving framework for building ML model-enabled backend and microservices. It bridges the gap between any machine learning models you just trained and the efficient online service API. -
7
NBA Sports Betting Machine Learning
NBA sports betting using machine learning
NBA-Machine-Learning-Sports-Betting is an open-source Python project that applies machine learning techniques to predict outcomes of National Basketball Association games for analytical and betting-related research. The system gathers historical team statistics and game data spanning multiple seasons, beginning with the 2007–2008 NBA season and continuing through the present. Using this dataset, the project constructs matchup features that represent team performance trends and contextual information about each game. Machine learning models are then trained to estimate the probability that a team will win a game as well as whether the total score will fall above or below the sportsbook’s predicted total. In addition to predicting outcomes, the project evaluates expected value to determine whether a potential bet offers a statistical advantage compared with sportsbook odds. -
8
NarratoAI
Using AI models to automatically provide commentary and edit videos
NarratoAI is an open-source platform designed to automate the generation of narrative content using artificial intelligence. The system combines large language models with media processing capabilities to create scripts, stories, and structured narrative outputs from user inputs. NarratoAI supports workflows where users provide prompts, themes, or source materials, and the software organizes them into coherent narrative structures suitable for articles, scripts, or multimedia storytelling. The project integrates multiple AI components such as text generation models, content structuring pipelines, and automated editing tools to streamline content creation. It is particularly useful for developers and creators building automated storytelling systems, AI-generated videos, or long-form written content. The architecture allows integration with external APIs and generative models so users can customize how narratives are generated and refined. -
9
PRIME
Scalable RL solution for advanced reasoning of language models
PRIME is an open-source reinforcement learning framework designed to improve the reasoning capabilities of large language models through process-level rewards rather than relying only on final outputs. The system introduces the concept of process reinforcement through implicit rewards, allowing models to receive feedback on intermediate reasoning steps instead of evaluating only the final answer. This approach helps models learn better reasoning strategies and encourages them to generate more reliable multi-step solutions to complex tasks. PRIME provides training pipelines, datasets, and experimental infrastructure that allow researchers to train models with reinforcement learning tailored for reasoning improvement. The framework also includes data preprocessing utilities and example datasets such as mathematical reasoning tasks that are well suited for process-based reward signals. -
Forever Free Full-Stack Observability | Grafana CloudBuilt on open standards like Prometheus and OpenTelemetry, Grafana Cloud includes Kubernetes Monitoring, Application Observability, Incident Response, plus the AI-powered Grafana Assistant. Get started with our generous free tier today.
-
10
Paper2Slides
From Paper to Presentation in One Click
Paper2Slides is an automation tool that converts research papers, reports, and other documents into polished slide decks and posters with minimal manual effort. It is designed to replace the repetitive work of turning dense technical documents into presentation-friendly structure by extracting key points, figures, and data into a coherent visual narrative. The system supports multiple input formats, so you can process PDFs and common office documents rather than being locked to a single file type. It uses an extraction approach intended to capture critical insights comprehensively, including important visuals and data points that often get missed in naive summarization. A major focus is traceability: generated slide content is designed to remain linked back to the source material so you can verify accuracy and reduce information drift. It also offers styling flexibility, letting you use built-in themes or describe a custom design direction in natural language for themed outputs. -
11
Qwen-VL
Chat & pretrained large vision language model
Qwen-VL is Alibaba Cloud’s vision-language large model family, designed to integrate visual and linguistic modalities. It accepts image inputs (with optional bounding boxes) and text, and produces text (and sometimes bounding boxes) as output. The model variants (VL-Plus, VL-Max, etc.) have been upgraded for better visual reasoning, text recognition from images, fine-grained understanding, and support for high image resolutions / extreme aspect ratios. Qwen-VL supports multilingual inputs and conversation (e.g. Chinese, English), and is aimed at tasks like image captioning, question answering on images (VQA, DocVQA), grounding (detecting objects or regions from textual queries), etc. -
12
Qwen2.5-Math
A series of math-specific large language models of our Qwen2 series
Qwen2.5-Math is a series of mathematics-specialized large language models in the Qwen2 family, released by Alibaba’s QwenLM. It includes base models (1.5B / 7B / 72B parameters), instruction-tuned versions, and a reward model (RM) to improve alignment. Unlike its predecessor Qwen2-Math, Qwen2.5-Math supports both Chain-of-Thought (CoT) reasoning and Tool-Integrated Reasoning (TIR) for solving math problems, and works in both Chinese and English. It is optimized for solving mathematical benchmarks and exams; the 72B-Instruct model achieves state-of-the-art results among open source models on many English and Chinese math tasks. -
13
Qwen2.5-Omni
Capable of understanding text, audio, vision, video
Qwen2.5-Omni is an end-to-end multimodal flagship model in the Qwen series by Alibaba Cloud, designed to process multiple modalities (text, images, audio, video) and generate responses both as text and natural speech in streaming real-time. It supports “Thinker-Talker” architecture, and introduces innovations for aligning modalities over time (for example synchronizing video/audio), robust speech generation, and low-VRAM/quantized versions to make usage more accessible. It holds state-of-the-art performance in many multimodal benchmarks, particularly spoken language understanding, audio reasoning, image/video understanding, etc. Very strong benchmark performance across modalities (audio understanding, speech recognition, image/video reasoning) and often outperforming or matching single-modality models at a similar scale. Real-time streaming responses, including natural speech synthesis (text-to-speech) and chunked inputs for low latency interaction. -
14
RAGxplorer
Open-source tool to visualise your RAG
RAGxplorer is an open-source visualization tool designed to help developers analyze and understand Retrieval-Augmented Generation (RAG) pipelines. Retrieval-augmented generation combines language models with external document retrieval systems in order to produce more accurate and grounded responses. However, RAG systems can be complex because they involve multiple components such as embedding models, vector databases, and retrieval algorithms. RAGxplorer provides visual tools that allow developers to inspect how documents are embedded, retrieved, and used to answer queries. The software can load documents, generate embeddings, and project them into reduced vector spaces so that users can visually explore relationships between queries and retrieved documents. It also includes interactive interfaces that show how retrieval affects the final output of the language model. -
15
Reader 3
Quick illustration of how one can easily read books together with LLMs
This project is a minimalist, self-hosted EPUB reader designed to help users browse and read EPUB books one chapter at a time through a lightweight local server, making it especially easy to extract or work with chapters in external tools like large language models. It was created primarily as a simple demonstration of how to combine local book reading with LLM workflows without heavy dependencies or complicated setup, and it runs with just a small Python script and a basic HTTP server. The interface focuses on clarity and ease of use, offering straightforward navigation of book chapters rather than full-featured e-reading capabilities. While it lacks advanced features like built-in annotations or rich media support, its simplicity is intentional, enabling users to quickly load EPUBs, view them in a browser, and even repurpose text for downstream tasks. -
16
Text-to-LoRA (T2L)
Hypernetworks that adapt LLMs for specific benchmark tasks
Text-to-LoRA is a research project that introduces a method for dynamically adapting large language models using hypernetworks that generate LoRA parameters directly from textual descriptions. Instead of training a new LoRA adapter for every task or dataset, the system can produce task-specific adaptations based solely on a text description of the desired capability. This approach enables models to rapidly internalize new contextual knowledge without performing traditional fine-tuning steps. The project provides a reference implementation of the Doc-to-LoRA method, which allows language models to quickly encode factual information or contextual constraints into lightweight LoRA modules. Developers and researchers can experiment with how textual task descriptions can generate LoRA weights that modify model behavior in real time. -
17
Unstructured.IO
Open source libraries and APIs to build custom preprocessing pipelines
The unstructured library provides open-source components for ingesting and pre-processing images and text documents, such as PDFs, HTML, Word docs, and many more. The use cases of unstructured revolve around streamlining and optimizing the data processing workflow for LLMs. unstructured modular bricks and connectors form a cohesive system that simplifies data ingestion and pre-processing, making it adaptable to different platforms and is efficient in transforming unstructured data into structured outputs. -
18
VibeThinker
Diversity-driven optimization and large-model reasoning ability
VibeThinker is a compact but high-capability open-source language model released by WeiboAI (Sina AI Lab). It contains about 1.5 billion parameters, far smaller than many “frontier” models, yet it is explicitly optimized for reasoning, mathematics, and code generation tasks rather than general open-domain chat. The innovation lies in its training methodology: the team uses what they call the Spectrum-to-Signal Principle (SSP), where a first stage emphasizes diversity of reasoning paths (the “spectrum” phase) and a second stage uses reinforcement techniques (the “signal” phase) to refine toward correctness and strong reasoning. The result is a model that outpaces many much larger models on domain-specific benchmarks, demonstrating that smaller models, if trained carefully and with the right objectives, can achieve high performance in reasoning-centric tasks. -
19
bitsandbytes
Accessible large language models via k-bit quantization for PyTorch
bitsandbytes is an open-source library designed to make training and inference of large neural networks more efficient by dramatically reducing memory usage. Built primarily for the PyTorch ecosystem, the library introduces advanced quantization techniques that allow models to operate using reduced numerical precision while maintaining high accuracy. These optimizations enable large language models and other deep learning architectures to run on hardware with limited memory resources, including consumer-grade GPUs. The project includes specialized optimizers and quantized matrix operations that significantly reduce the memory footprint of training and inference workloads. By lowering the hardware requirements needed to work with large models, bitsandbytes helps make modern AI development more accessible to researchers and engineers. The library has become widely used in machine learning pipelines that rely on parameter-efficient training techniques and low-precision inference. -
20
llm.c
LLM training in simple, raw C/CUDA
llm.c is a minimalist, systems-level implementation of a small transformer-based language model in C that prioritizes clarity and educational value. By stripping away heavy frameworks, it exposes the core math and memory flows of embeddings, attention, and feed-forward layers. The code illustrates how to wire forward passes, losses, and simple training or inference loops with direct control over arrays and buffers. Its compact design makes it easy to trace execution, profile hotspots, and understand the cost of each operation. Portability is a goal: it aims to compile with common toolchains and run on modest hardware for small experiments. Rather than delivering a production-grade stack, it serves as a reference and learning scaffold for people who want to “see the metal” behind LLMs. -
21
llmware
Unified framework for building enterprise RAG pipelines
llmware is an open source framework designed to simplify the creation of enterprise-grade applications powered by large language models. The platform focuses on building secure and private AI workflows that can run locally on laptops, edge devices, or self-hosted servers without relying exclusively on cloud APIs. It provides a unified interface for constructing retrieval-augmented generation pipelines, agent workflows, and document intelligence applications. One of the framework’s defining characteristics is its collection of small specialized language models optimized for specific tasks such as summarization, classification, and document analysis. The system supports a wide range of inference backends including PyTorch, OpenVINO, ONNX Runtime, and other optimized runtimes, allowing developers to choose the most efficient execution environment for their hardware. -
22
mindflow
AI-powered CLI git wrapper, boilerplate code generator, chat history
I-powered CLI git wrapper, boilerplate code generator, chat history manager, and code search engine to streamline your dev workflow. The ChatGPT-powered swiss army knife for the modern developer! We provide an AI-powered CLI git wrapper, boilerplate code generator, code search engine, a conversation history manager, and much more! Configure the model used for generating responses by running mf config and selecting either GPT 3.5 Turbo (default) or GPT 4. In order to use GPT 4, you'll need to have special access to the API. If you have access, you can run mf config and select GPT 4. If you don't have access, you'll get an error message. Interact with chatGPT directly just like on the chatGPT website. We also have chat persistence, so it will remember the previous chat messages. You can provide single or multi-file context to chatGPT by passing in any number of files as a separate argument in the mf chat call. -
23
runprompt
Run LLM prompts from your shell
runprompt is an interactive command launcher and prompt utility that lets users bind shell commands, scripts, and workflows to quick keyboard shortcuts or natural-language queries, helping streamline repetitive terminal tasks and boost developer productivity. It functions as a lightweight, launcher-centric interface where you can type a phrase, partial command, or alias and have RunPrompt suggest or execute relevant actions instantly, reducing the need to memorize long commands or navigate complex directory structures. The project emphasizes extensibility, letting users define custom actions, integrate with existing shell environments, and even leverage fuzzy matching or contextual prompts to narrow down options as you type. Designed to be cross-platform, RunPrompt works with standard shells on Windows, macOS, and Linux while honoring the user’s preferred environment and configurations. -
24
text-extract-api
Document (PDF, Word, PPTX ...) extraction and parse API
text-extract-api is an open-source service designed to extract readable text from a wide variety of document formats through a simple API interface. The project focuses on converting complex files such as PDFs, images, scanned documents, and office files into structured plain text that can be processed by downstream applications or language models. Instead of requiring developers to integrate multiple document parsing libraries individually, the system centralizes text extraction capabilities into a unified API that standardizes the output. The platform supports automated processing pipelines that detect file types and apply the appropriate extraction method to obtain the most accurate text representation possible. It can be integrated into document analysis systems, knowledge retrieval tools, and AI pipelines that rely on clean textual data. The architecture is designed to be lightweight and easily deployable, making it suitable for both local installations and cloud environments. -
25
tldw Server
Your Personal Research Multi-Tool
tldw-server (mirror) is a mirrored distribution of an open-source backend service designed to store, process, and serve summarized information extracted from long pieces of content. The name “tldw” reflects the phrase “too long; didn’t watch,” which refers to tools that condense lengthy videos, articles, or documents into concise summaries. The server component typically acts as the core infrastructure that manages summaries, metadata, and retrieval operations for client applications or user interfaces. In practical deployments, a system like this can support AI-powered summarization pipelines that process transcripts, articles, or other long-form material and store condensed versions for easier consumption. The mirrored project hosted on SourceForge exists to preserve the availability of the code and provide an alternative download location for developers and researchers. Such servers are commonly integrated with AI models that generate summaries and tag content automatically.
