Search Results for "reinforcement learning"
Sort By:
Showing 25 open source projects for "reinforcement learning"
View related business solutions-
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.
-
Fully Managed MySQL, PostgreSQL, and SQL ServerCloud SQL handles your database ops end to end, so you can focus on your app.
-
1
MedicalGPT
MedicalGPT: Training Your Own Medical GPT Model with ChatGPT Training
MedicalGPT training medical GPT model with ChatGPT training pipeline, implementation of Pretraining, Supervised Finetuning, Reward Modeling and Reinforcement Learning. MedicalGPT trains large medical models, including secondary pre-training, supervised fine-tuning, reward modeling, and reinforcement learning training. -
2
DeepSeek R1
Open-source, high-performance AI model with advanced reasoning
...The model employs a Mixture of Experts (MoE) architecture, comprising 671 billion total parameters with 37 billion active parameters per token, and supports a context length of up to 128,000 tokens. DeepSeek-R1's training regimen uniquely integrates large-scale reinforcement learning (RL) without relying on supervised fine-tuning, enabling the model to develop advanced reasoning capabilities. This approach has resulted in performance comparable to leading models like OpenAI's o1, while maintaining cost-efficiency. To further support the research community, DeepSeek has released distilled versions of the model based on architectures such as LLaMA and Qwen. -
3
DeepSeek-V3
Powerful AI language model (MoE) optimized for efficiency/performance
...The model introduces an auxiliary-loss-free load balancing strategy and a multi-token prediction training objective to boost performance. Trained on 14.8 trillion diverse, high-quality tokens, DeepSeek-V3 underwent supervised fine-tuning and reinforcement learning to fully realize its capabilities. Evaluations indicate that it outperforms other open-source models and rivals leading closed-source models, achieving this with a training duration of 55 days on 2,048 Nvidia H800 GPUs, costing approximately $5.58 million. -
4
OpenTinker
OpenTinker is an RL-as-a-Service infrastructure for foundation models
OpenTinker is an open-source Reinforcement Learning-as-a-Service (RLaaS) infrastructure intended to democratize reinforcement learning for large language model (LLM) agents. Traditional RL setups can be monolithic and difficult to configure, but OpenTinker separates concerns across agent definition, environment interaction, and execution, which lets developers focus on defining the logic of agents and environments separately from how training and inference are run. ... -
AI-generated apps that pass security reviewRetool lets you generate dashboards, admin panels, and workflows directly on your data. Type something like “Build me a revenue dashboard on my Stripe data” and get a working app with security, permissions, and compliance built in from day one. Whether on our cloud or self-hosted, create the internal software your team needs without compromising enterprise standards or control.
-
5
Ring
Ring is a reasoning MoE LLM provided and open-sourced by InclusionAI
...Its design emphasizes reasoning, efficiency, and modular expert activation. In its “flash” variant (Ring-flash-2.0), it optimizes inference by activating only a subset of experts. It applies reinforcement learning/reasoning optimization techniques. Its architectures and training approaches are tuned to enable efficient and capable reasoning performance. Reasoning-optimized model with reinforcement learning enhancements. Efficient architecture and memory design for large-scale reasoning. If you are located in mainland China, we also provide the model on ModelScope.cn to speed up the download process. -
6
Sapiens
High-resolution models for human tasks
...It includes simulation environments, datasets, and benchmarks for testing grounded understanding, imitation learning, and decision-making. The system’s modular pipeline supports both imitation-based and reinforcement-based training strategies, allowing flexible experimentation with different embodiments and tasks. -
7
Tongyi DeepResearch
Tongyi Deep Research, the Leading Open-source Deep Research Agent
...The model is about 30.5 billion parameters in size, though at any given token only ~3.3B parameters are active. It uses a mix of synthetic data generation, fine-tuning and reinforcement learning; supports benchmarks like web search, document understanding, question answering, “agentic” tasks; provides inference tools, evaluation scripts, and “web agent” style interfaces. The aim is to enable more autonomous, agentic models that can perform sustained knowledge gathering, reasoning, and synthesis across multiple modalities (web, files, etc.). -
8
Pearl
A Production-ready Reinforcement Learning AI Agent Library
Pearl is a production-ready reinforcement learning and contextual bandit agent library built for real-world sequential decision making. It is organized around modular components—policy learners, replay buffers, exploration strategies, safety modules, and history summarizers—that snap together to form reliable agents with clear boundaries and strong defaults. The library implements classic and modern algorithms across two regimes: contextual bandits (e.g., LinUCB, LinTS, SquareCB, neural bandits) and fully sequential RL (e.g., DQN, PPO-style policy optimization), with attention to practical concerns like nonstationarity and dynamic action spaces. ... -
9
GLM-TTS
Controllable & emotion-expressive zero-shot TTS
...It uses a two-stage architecture where a generative LLM first converts text into intermediate speech token sequences and then a Flow-based neural model converts those tokens into natural audio waveforms, enabling rich prosody and voice character even for unseen speakers. The system introduces a multi-reward reinforcement learning framework that jointly optimizes for voice similarity, emotional expressiveness, pronunciation, and intelligibility, yielding output that can rival commercial options in naturalness and expressiveness. GLM-TTS also supports phoneme-level control and hybrid text + phoneme input, giving developers precise control over pronunciation critical for multilingual or polyphone-rich languages. -
Gemini 3 and 200+ AI Models on One PlatformBuild generative AI apps with Vertex AI. Switch between models without switching platforms.
-
10
GLM-4
GLM-4 series: Open Multilingual Multimodal Chat LMs
...The GLM-4-32B-0414 models are trained on ~15T high-quality data (including substantial synthetic reasoning data), then post-trained with preference alignment, rejection sampling, and reinforcement learning to improve instruction following, coding, function calling, and agent-style behaviors. The GLM-Z1-32B-0414 line adds deeper mathematical, coding, and logical reasoning via extended reinforcement learning and pairwise ranking feedback, while GLM-Z1-Rumination-32B-0414 introduces a “rumination” mode that performs longer, tool-using deep research for complex, open-ended tasks. ... -
11
GLM-4.5V
GLM-4.6V/4.5V/4.1V-Thinking, towards versatile multimodal reasoning
...It embodies the design philosophy of mixing visual and textual modalities into a unified model capable of general-purpose reasoning, content understanding, and generation, while already supporting a wide variety of tasks: from image captioning and visual question answering to content recognition, GUI-based agents, video understanding, and long-document interpretation. GLM-4.5V emerged from a training framework that leverages scalable reinforcement learning (with curriculum sampling) to boost performance across tasks ranging from STEM problem solving to long-context reasoning, giving it broad applicability beyond narrow benchmarks. When it was released, it achieved state-of-the-art results on a large collection of public multimodal benchmarks for open-source models. -
12
HY-Motion 1.0
HY-Motion model for 3D character animation generation
HY-Motion 1.0 is an open-source, large-scale AI model suite developed by Tencent’s Hunyuan team that generates high-quality 3D human motion from simple text prompts, enabling the automatic production of fluid, diverse, and semantically accurate animations without manual keyframing or rigging. Built on advanced deep learning architectures that combine Diffusion Transformer (DiT) and flow matching techniques, HY-Motion scales these approaches to the billion-parameter level, resulting in strong instruction-following capabilities and richer motion outputs compared to existing open-source models. The training strategy for the HY-Motion series includes extensive pre-training on thousands of hours of varied motion data, fine-tuning on curated high-quality datasets, and reinforcement learning with human feedback, which improves both the plausibility and adaptability of generated motion sequences. -
13
MiniMax-M1
Open-weight, large-scale hybrid-attention reasoning model
...The team emphasizes efficient scaling of test-time compute: at 100K-token generation lengths, M1 reportedly uses only about 25 percent of the FLOPs of some competing models, making extended “think step” traces more feasible. M1 is further trained with large-scale reinforcement learning over diverse tasks. -
14
GLM-4.5
GLM-4.5: Open-source LLM for intelligent agents by Z.ai
GLM-4.5 is a cutting-edge open-source large language model designed by Z.ai for intelligent agent applications. The flagship GLM-4.5 model has 355 billion total parameters with 32 billion active parameters, while the compact GLM-4.5-Air version offers 106 billion total parameters and 12 billion active parameters. Both models unify reasoning, coding, and intelligent agent capabilities, providing two modes: a thinking mode for complex reasoning and tool usage, and a non-thinking mode for... -
15
FireRed-Image-Edit
General-purpose image editing model that delivers high-fidelity
FireRed-Image-Edit is an open-source general-purpose image editing model and toolset designed to deliver high-fidelity, visually coherent edits across a wide range of editing tasks, from simple object modifications to complex enhancements like restoration and style preservation. It is built on a flexible text-to-image foundation model that has been extended with training paradigms including pretraining, supervised fine-tuning, and reinforcement learning to imbue the system with strong instruction following and editing consistency. The model excels in maintaining visual and text stylistic fidelity, allowing users to preserve the original artistic qualities of an image while applying creative changes according to natural language instructions. In addition to editing single images, FireRed supports multi-image editing scenarios such as virtual try-on or batch transformations, making it suitable for both creative and practical workflows. -
16
fairseq2
FAIR Sequence Modeling Toolkit 2
...Built from the ground up for scalability, composability, and research flexibility, fairseq2 supports a broad range of language, speech, and multimodal content generation tasks, including instruction fine-tuning, reinforcement learning from human feedback (RLHF), and large-scale multilingual modeling. Unlike the original fairseq—which evolved into a large, monolithic codebase—fairseq2 introduces a clean, plugin-oriented architecture designed for long-term maintainability and rapid experimentation. It supports multi-GPU and multi-node distributed training using DDP, FSDP, and tensor parallelism, capable of scaling up to 70B+ parameter models. ... -
17
Step-Audio-EditX
LLM-based Reinforcement Learning audio edit model
Step-Audio-EditX is an open-source, 3 billion-parameter audio model from StepFun AI designed to make expressive and precise editing of speech and audio as easy as text editing. Rather than treating audio editing as low-level waveform manipulation, this model converts speech into a sequence of discrete “audio tokens” (via a dual-codebook tokenizer) — combining a linguistic token stream and a semantic (prosody/emotion/style) token stream — thereby abstracting audio editing into high-level... -
18
GLM-V
GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning
...It introduces hybrid training for broad-spectrum reasoning and a Thinking Mode switch to balance speed and depth of reasoning. GLM-4.1V-9B-Thinking incorporates reinforcement learning with curriculum sampling (RLCS) and Chain-of-Thought reasoning, outperforming models much larger in scale (e.g., Qwen-2.5-VL-72B) across many benchmarks. -
19
Ling-V2
Ling-V2 is a MoE LLM provided and open-sourced by InclusionAI
...It introduces highly sparse architectures where only a fraction of the model’s parameters are activated per input token, enabling models like Ling-mini-2.0 to achieve reasoning and instruction-following capabilities on par with much larger dense models while remaining significantly more computationally efficient. Trained on more than 20 trillion tokens of high-quality data and enhanced through multi-stage supervised fine-tuning and reinforcement learning, Ling-V2’s models demonstrate strong general reasoning, mathematical problem-solving, coding understanding, and knowledge-intensive task performance. -
20
Step-Audio 2
Multi-modal large language model designed for audio understanding
Step-Audio2 is an advanced, end-to-end multimodal large language model designed for high-fidelity audio understanding and natural speech conversation: unlike many pipelines that separate speech recognition, processing, and synthesis, Step-Audio2 processes raw audio, reasons about semantic and paralinguistic content (like emotion, speaker characteristics, non-verbal cues), and can generate contextually appropriate responses — including potentially generating or transforming audio output. It integrates a latent-space audio encoder, discrete acoustic tokens, and reinforcement-learning–based training (CoT + RL) to enhance its ability to capture and reproduce voice styles, intonations, and subtle vocal cues. Moreover, Step-Audio2 supports tool-calling and retrieval-augmented generation (RAG), allowing it to access external knowledge sources or audio/text databases, thus reducing hallucinations and improving coherence in complex dialogues. -
21
GLM-4-32B-0414
Open Multilingual Multimodal Chat LMs
...It supports multilingual and multimodal chat capabilities with an extensive 32K token context length, making it ideal for dialogue, reasoning, and complex task completion. The model is pre-trained on 15 trillion tokens of high-quality data, including substantial synthetic reasoning datasets, and further enhanced with reinforcement learning and human preference alignment for improved instruction-following and function calling. Variants like GLM-Z1-32B-0414 offer deep reasoning and advanced mathematical problem-solving, while GLM-Z1-Rumination-32B-0414 specializes in long-form, complex research-style writing using scaled reinforcement learning and external search tools. ... -
22
Multi-Agent Emergence Environments
Environment generation code for the paper "Emergent Tool Use"
multi-agent-emergence-environments is an open source research environment framework developed by OpenAI for the study of emergent behaviors in multi-agent systems. It was designed for the experiments described in the paper and blog post “Emergent Tool Use from Multi-Agent Autocurricula”, which investigated how complex cooperative and competitive behaviors can evolve through self-play. The repository provides environment generation code that builds on the mujoco-worldgen package, enabling... -
23
DeepSeek-V3.2-Speciale
High-compute ultra-reasoning model surpassing model surpassing GPT-5
...Unlike the standard version, Speciale is tuned exclusively for deep reasoning and therefore does not support tool-calling, focusing its full capacity on pure cognitive performance. The model uses a scaled reinforcement learning framework that allows it to surpass GPT-5 in several evaluations and reach reasoning performance comparable to Gemini-3.0-Pro. DeepSeek-V3.2-Speciale contributed to gold-medal solutions in the 2025 IMO, IOI, ICPC World Finals, and CMO, demonstrating its ability to handle elite-level problem solving. It is released under the MIT license and includes curated benchmark solutions for community verification and analysis. -
24
DeepSeek-V3.2
High-efficiency reasoning and agentic intelligence model
...It introduces DeepSeek Sparse Attention (DSA), a new attention mechanism that dramatically reduces computational overhead while maintaining strong long-context performance. Built with a scalable reinforcement learning framework, it reaches near-GPT-5 levels of reasoning and outperforms comparable models like DeepSeek-V3.1 and Gemini-3.0-Pro in advanced benchmarks. The model was notably used in competitive AI challenges such as the 2025 International Mathematical Olympiad (IMO) and IOI, achieving top-tier results. DeepSeek-V3.2 also features a large-scale agentic task synthesis pipeline, which generates training data to enhance tool-use intelligence and multi-step reasoning. ... -
25
Mellum-4b-base
JetBrains’ 4B parameter code model for completions
...The model is efficient for both cloud inference with vLLM and local deployment using llama.cpp or Ollama, thanks to its bf16 precision and AMP training. While the base model is not fine-tuned for downstream tasks, it is designed to be easily adapted through supervised fine-tuning (SFT) or reinforcement learning (RL). Benchmarks on RepoBench, SAFIM, and HumanEval demonstrate its competitive performance, with specialized fine-tuned versions for Python already showing strong improvements.
- Previous
- You're on page 1
- Next
