Search Results for "expert"
Sort By:
Showing 21 open source projects for "expert"
View related business solutions-
$300 Free Credits to Build on Google CloudPut your $300 in credit toward real workloads, then keep building with free monthly usage for 20+ products. No commitment and no charge until you upgrade.
-
Build Securely on AWS with Proven FrameworksMoving to the cloud brings new challenges. How can you manage a larger attack surface while ensuring great network performance? Turn to Fortinet’s Tested Reference Architectures, blueprints for designing and securing cloud environments built by cybersecurity experts. Learn more and explore use cases in this white paper.
-
1
Ring
Ring is a reasoning MoE LLM provided and open-sourced by InclusionAI
Ring is a reasoning Mixture-of-Experts (MoE) large language model (LLM) developed by inclusionAI. It is built from or derived from Ling. 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. ... -
2
Colibrì
Run GLM-5.2 (744B MoE) on a 25GB-RAM consumer machine
...It keeps the dense portion of the quantized model in memory while streaming routed experts from a large disk-based store as they are needed. The runtime is implemented in pure C, requires no Python or BLAS during inference, and can operate without a GPU. Compressed attention caches, expert caching, optional hot tiers, and speculative decoding reduce memory pressure and improve repeated use. A planning tool calculates safe disk, RAM, and VRAM placement before loading the model, while a diagnostic command checks system readiness. Colibri includes terminal chat, an OpenAI-compatible text API, and a browser client, but disk-bound generation can be slow on cold caches. -
3
Tencent-Hunyuan-Large
Open-source large language model family from Tencent Hunyuan
Tencent-Hunyuan-Large is the flagship open-source large language model family from Tencent Hunyuan, offering both pre-trained and instruct (fine-tuned) variants. It is designed with long-context capabilities, quantization support, and high performance on benchmarks across general reasoning, mathematics, language understanding, and Chinese / multilingual tasks. It aims to provide competitive capability with efficient deployment and inference. FP8 quantization support to reduce memory usage... -
4
Ling
Ling is a MoE LLM provided and open-sourced by InclusionAI
Ling is a Mixture-of-Experts (MoE) large language model (LLM) provided and open-sourced by inclusionAI. The project offers different sizes (Ling-lite, Ling-plus) and emphasizes flexibility and efficiency: being able to scale, adapt expert activation, and perform across a range of natural language/reasoning tasks. Example scripts, inference pipelines, and documentation. The codebase includes inference, examples, models, documentation, and model download infrastructure. As more developers and researchers engage with the platform, we can expect rapid advancements and improvements, leading to even more sophisticated applications. ... -
Ship Agents FasterGemini Enterprise Agent Platform lets you rapidly build, scale, govern and optimize production-ready agents grounded in your organization's data. The platform enables developers to build custom or pre-built agents for virtually any use case. New customers get $300 in free credits.
-
5
Wan2.2
Wan2.2: Open and Advanced Large-Scale Video Generative Model
Wan2.2 is a major upgrade to the Wan series of open and advanced large-scale video generative models, incorporating cutting-edge innovations to boost video generation quality and efficiency. It introduces a Mixture-of-Experts (MoE) architecture that splits the denoising process across specialized expert models, increasing total model capacity without raising computational costs. Wan2.2 integrates meticulously curated cinematic aesthetic data, enabling precise control over lighting, composition, color tone, and more, for high-quality, customizable video styles. The model is trained on significantly larger datasets than its predecessor, greatly enhancing motion complexity, semantic understanding, and aesthetic diversity. ... -
6
Kimi K2
Kimi K2 is the large language model series developed by Moonshot AI
...The model family includes variants like a foundational base model that researchers can fine-tune for specific use cases and an instruct-optimized variant primed for general-purpose chat and agent-style interactions, offering flexibility for both experimentation and deployment. With its high-dimensional attention mechanisms and expert routing, Kimi-K2 excels across benchmarks in live coding, math reasoning, and problem solving. -
7
Kimi K3 in C
A 2.78-trillion-parameter Kimi K3 running inference on a single CPU
...The runtime streams model trunk layers and routed experts from disk instead of keeping all weights resident in memory. Memory presets balance pinned layers and an expert LRU cache for laptops, desktops, workstations, and servers. Incremental generation preserves KV cache and recurrent state between tokens. The repository also includes tokenization, safetensors loading, diagnostics, benchmarks, trace replay, and comparisons against a PyTorch reference. -
8
HunyuanImage-3.0
A Powerful Native Multimodal Model for Image Generation
...It unifies multimodal understanding and generation in a single autoregressive framework, combining text and image modalities seamlessly rather than relying on separate image-only diffusion components. It uses a Mixture-of-Experts (MoE) architecture with many expert subnetworks to scale efficiently, deploying only a subset of experts per token, which allows large parameter counts without linear inference cost explosion. The model is intended to be competitive with closed-source image generation systems, aiming for high fidelity, prompt adherence, fine detail, and even “world knowledge” reasoning (i.e. leveraging context, semantics, or common sense in generation). ... -
9
TurboFieldfare
Gemma 4 26B-A4B inference in ~2 GB of RAM on any M-series MacBook
...The project includes a native Mac application, command-line tools, a streaming installer, a Swift library, and an experimental OpenAI-compatible local server. Quantized weights, custom Metal kernels, chunked prefill, and a bounded expert cache improve efficiency. The installer downloads and repacks model ranges without staging a second full checkpoint. Its current scope is text-only Gemma inference on macOS 26 and Apple Silicon hardware. -
Custom VMs From 1 to 96 vCPUs With 99.95% UptimeLive migration and automatic failover keep workloads online through maintenance. One free e2-micro VM every month.
-
10
WASTE
Run the full 2.78-trillion-parameter Kimi K3 model
WASTE is an embeddable C inference engine for running extremely large mixture-of-experts models when the weights exceed available RAM. It keeps the shared model trunk in memory and streams only the experts selected for each token from fast NVMe storage. A bounded cache reuses recently needed experts, while lookahead routing begins disk reads before the next layer requires them. Its main target is the full 2.78-trillion-parameter Kimi K3 model, including multimodal image input. The engine has... -
11
SAM 3D Body
Code for running inference with the SAM 3D Body Model 3DB
...There are Jupyter notebooks that walk you through setting up the model, running it on example images, and visualizing outputs in 3D, making it approachable even if you are not a 3D expert. -
12
DeepGEMM
Clean and efficient FP8 GEMM kernels with fine-grained scaling
DeepGEMM is a specialized CUDA library for efficient, high-performance general matrix multiplication (GEMM) operations, with particular focus on low-precision formats such as FP8 (and experimental support for BF16). The library is designed to work cleanly and simply, avoiding overly templated or heavily abstracted code, while still delivering performance that rivals expert-tuned libraries. It supports both standard and “grouped” GEMMs, which is useful for architectures like Mixture of Experts (MoE) that require segmented matrix multiplications. One distinguishing aspect is that DeepGEMM compiles its kernels at runtime (via a lightweight Just-In-Time (JIT) module), so users don’t need to precompile CUDA kernels before installation. ... -
13
HunyuanOCR
OCR expert VLM powered by Hunyuan's native multimodal architecture
HunyuanOCR is an open-source, end-to-end OCR (optical character recognition) Vision-Language Model (VLM) developed by Tencent‑Hunyuan. It’s designed to unify the entire OCR pipeline, detection, recognition, layout parsing, information extraction, translation, and even subtitle or structured output generation, into a single model inference instead of a cascade of separate tools. Despite being fairly lightweight (about 1 billion parameters), it delivers state-of-the-art performance across a... -
14
MiniMax-01
Large-language-model & vision-language-model based on Linear Attention
...It has 456 billion total parameters with 45.9 billion activated per token and is trained with advanced parallel strategies such as LASP+, varlen ring attention, and Expert Tensor Parallelism, enabling a training context of 1 million tokens and up to 4 million tokens at inference. MiniMax-VL-01 extends this core by adding a 303M-parameter Vision Transformer and a two-layer MLP projector in a ViT–MLP–LLM framework, allowing the model to process images at dynamic resolutions up to 2016×2016. -
15
DeepSeek MoE
Towards Ultimate Expert Specialization in Mixture-of-Experts Language
DeepSeek-MoE (“DeepSeek MoE”) is the DeepSeek open implementation of a Mixture-of-Experts (MoE) model architecture meant to increase parameter efficiency by activating only a subset of “expert” submodules per input. The repository introduces fine-grained expert segmentation and shared expert isolation to improve specialization while controlling compute cost. For example, their MoE variant with 16.4B parameters claims comparable or better performance to standard dense models like DeepSeek 7B or LLaMA2 7B using about 40% of the total compute. ... -
16
ZAYA1-8B
Efficient MoE reasoning model for coding and math workloads
...ZAYA1-8B is optimized for long-form reasoning and test-time compute workflows, making it particularly effective for mathematical problem solving, coding tasks, and advanced reasoning chains. It introduces architectural innovations such as Compressed Convolutional Attention, a novel MLP-based expert router, and learned residual scaling to improve routing stability and inference efficiency. The model was trained entirely on AMD infrastructure and refined through supervised fine-tuning and multi-stage reinforcement learning focused on reasoning and coding. -
17
Solar Open 2
Efficient 250B MoE model for agents, coding, and long-context work
...The model supports a native 1M-token context window and uses NoPE instead of rotary positional encoding, reducing long-context KV-cache requirements. Solar Open 2 includes 321 experts, with eight routed experts plus one shared expert activated per token. It was pretrained on roughly 12 trillion tokens and supports English, Korean, and Japanese. Agent capabilities include multi-step reasoning, tool calling, MCP tools, and end-to-end task execution. It also offers direct-response and high-reasoning modes and supports deployment through Transformers, vLLM, SGLang, and quantized variants. -
18
Hy3 preview
Efficient MoE model for reasoning, coding, and AI agent workflows
Hy3 preview is Tencent Hunyuan’s latest open-weight Mixture-of-Experts language model, designed for advanced reasoning, coding, instruction following, and autonomous agent workflows. It is the first model built on Tencent’s rebuilt training infrastructure and introduces significant improvements in context learning, software engineering, and tool-based task execution. The model features 295B total parameters with only 21B activated during inference, plus a dedicated 3.8B Multi-Token... -
19
Qwen3.8-2.4T-A95B
Massive 2.4T MoE model for coding, agents, research, and reasoning
Qwen3.8-2.4T-A95B is Qwen’s largest open-weight model and the first Qwen-Max-class model released openly, targeting advanced coding, professional work, research, and long-horizon agentic tasks. It uses a massive Mixture-of-Experts architecture with 2.4 trillion total parameters while activating 95B per token, combining Gated DeltaNet and attention layers across 512 experts. The model emphasizes reliable autonomous execution, including stronger planning, environment feedback handling, and... -
20
Ling 3.0 Tiny
Lightweight MoE model for local reasoning, coding, and AI agents
...It contains 7.9B total parameters while activating only 1.3B per token, using a hybrid architecture that alternates Kimi Delta Attention and Multi-Head Latent Attention with a sparse 128-expert MoE. The model supports both fast responses and configurable multi-step thinking, covering general agents, coding, mathematics, scientific reasoning, and instruction following. It is specifically optimized for local and resource-constrained deployment and has been validated on NVIDIA DGX Spark, Apple Silicon MacBooks, and Mac mini systems. ... -
21
DeepSeek-V4-Flash
Efficient MoE model for million-token reasoning and coding
...The model uses a hybrid attention architecture that combines Compressed Sparse Attention and Heavily Compressed Attention to improve long-context efficiency, while Manifold-Constrained Hyper-Connections strengthen signal stability across layers. It is trained on more than 32T tokens and refined through a post-training pipeline that includes supervised fine-tuning, reinforcement learning, domain-specific expert cultivation, and on-policy distillation. DeepSeek-V4-Flash supports non-think, think, and think-max reasoning modes, allowing users to balance speed and depth. It is smaller than DeepSeek-V4-Pro but can approach Pro-level reasoning.
- Previous
- You're on page 1
- Next