保存済みツールを推薦
FOXAI SEARCH

サイト内検索

AI モデル 2119

  • Carnice-Qwen3.6-MoE-35B-A3B-APEX-MTP-GGUF↗

    mudler

    ⚡ Each donation = another big MoE quantized I host 30+ free APEX MoE quantizations as independent research. My only local hardware is an NVIDIA DGX Spark (122 GB unified memory), enough for ~30-50B-class MoEs, but bigger ones (200B+) require rented compute on H100/H200/Blackwell, typically $20-100 per quant. If APEX quants are useful to you, your support directly funds those bigger runs. **APEX (Adaptive Precision for EXpert Models)** quantizations of samuelcardillo/Carnice-Qwen3.6-MoE-35B-A3B, with the **MTP (multi-token prediction) head bundled** for in-the-box self-speculative decoding. These GGUFs bundle the model's **MTP (multi-token prediction) head** alongside the trunk in a single file, courtesy of llama.cpp PR #22673. With a recent llama.cpp (>= commit 255582687) you can enable self-speculative decoding using just this one file — no separate draft model needed:

  • UVDoc↗

    PaddlePaddle

    The main purpose of text image correction is to carry out geometric transformation on the image to correct the document distortion, inclination, perspective deformation and other problems in the image, so that the subsequent text recognition can be more accurate. For details about PaddlePaddle installation, please refer to the PaddlePaddle official website. You can also integrate the model inference of the TextImageUnwarping module into your project. Before running the following code, please download the sample image to your local machine.

  • wav2vec2-large-xlsr-marathi↗

    sumedh

    Fine-tuned facebook/wav2vec2-large-xlsr-53 on Marathi using the Open SLR64 dataset. When using this model, make sure that your speech input is sampled at 16kHz. This data contains only female voices but the model works well for male voices too. Trained on Google Colab Pro on Tesla P100 16GB GPU. **WER (Word Error Rate) on the Test Set**: 12.70 % The model can be used directly without a language model as follows, given that your dataset has Marathi `actual_text` and `path_in_folder` columns: Train-Test ratio was 90:10. The training notebook Colab link here. weights-and-biases run summary here

  • DeepSeek-V4-Flash-DSpark-draft-GGUF↗

    YanissAmz

  • MiniMax-H3-GGUF↗

    Abiray

    This repository (`Abiray/MiniMax-H3-GGUF`) provides GGUF quantized versions and necessary component files for the MiniMax H3 model. MiniMax H3 is a general-purpose, omni-modal generative system that supports unified understanding of multimodal contexts composed of text, images, video, and audio. It can generate video with native stereo audio at resolutions up to 2K and durations of up to 15 seconds. --- If you are looking for a smaller model with the same great quality that fits better on consumer-tier GPUs, please check out the **MiniMax-H3-Pruned-GGUF** repository. The pruned architecture is compressed down to 8.9 GB – 21.6 GB, bringing MiniMax H3 execution directly to consumer hardware. **Available Pruned Quants:** | Bit Size | Quantization Type | File Size | | :--- | :--- | :--- | | **3-bit** | `Q3_K_M` | 8.9 GB | | **4-bit** | `Q4_K_S`, `Q4_K_M` | 11.6 GB | | **5-bit** | `Q5_K_S`, `Q5_K_M` | 14.1 GB | | **6-bit** | `Q6_K` | 16.7 GB | | **8-bit** | `Q8_0` | 21.6 GB | --- This repository includes quantized versions of both the `FL2VA` (First-and-last-frame mode) and `Ref2VA` (Omni-reference mode) base models. * **FL2VA Models:** * `MiniMax-H3-FL2VA-Q3_K_M.gguf` (15.6 GB) * `MiniMax-H3-FL2VA-Q3_K_S.gguf` (15.6 GB) * `MiniMax-H3-FL2VA-Q4_0.gguf` (18.6 GB) * `MiniMax-H3-FL2VA-Q4_K_M.gguf` (19.9 GB) * `MiniMax-H3-FL2VA-Q4_K_S.gguf` (19.9 GB) * `MiniMax-H3-FL2VA-Q5_0.gguf` (22.8 GB) * `MiniMax-H3-FL2VA-Q5_K_M.gguf` (23.9 GB) * `MiniMax-H3-FL2VA-Q5_K_S.gguf` (23.9 GB) * `MiniMax-H3-FL2VA-Q6_K.gguf` (28.2 GB) * `MiniMax-H3-FL2VA-Q8_0.gguf` (36 GB) * **Ref2VA Models:** * `MiniMax-H3-Ref2VA-Q3_K_M.gguf` (15.6 GB) * `MiniMax-H3-Ref2VA-Q3_K_S.gguf` (15.6 GB) * `MiniMax-H3-Ref2VA-Q4_0.gguf` (18.6 GB) * `MiniMax-H3-Ref2VA-Q4_K_M.gguf` (19.9 GB) * `MiniMax-H3-Ref2VA-Q4_K_S.gguf` (19.9 GB) * `MiniMax-H3-Ref2VA-Q5_0.gguf` (22.8 GB) * `MiniMax-H3-Ref2VA-Q5_K_M.gguf` (23.9 GB) * `MiniMax-H3-Ref2VA-Q5_K_S.gguf` (23.9 GB) * `MiniMax-H3-Ref2VA-Q6_K.gguf` (28.2 GB) * `MiniMax-H3-Ref2VA-Q8_0.gguf` (36 GB) * `qwen3vl_32b_minimax_h3_int4_convrot.safetensors` (15 GB) * `qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors` (27.1 GB) * `qwen3vl_32b_minimax_h3-Q4_K_M.gguf` (14.6 GB) * `minimax_h3_audio_vae_fp32.safetensors` (605 MB) * `minimax_h3_video_vae_fp16.safetensors` (5.21 GB) * **H3-Base-FL2VA:** First-and-last-frame mode. Supports zero, one, or two input images. * No image input: T

  • chronos-bolt-base↗

    amazon

    🚀 **Update Feb 14, 2025**: Chronos-Bolt models are now available on Amazon SageMaker JumpStart! Check out the tutorial notebook to learn how to deploy Chronos endpoints for production use in a few lines of code. Chronos-Bolt is a family of pretrained time series forecasting models which can be used for zero-shot forecasting. It is based on the T5 encoder-decoder architecture and has been trained on nearly 100 billion time series observations. It chunks the historical time series context into patches of multiple observations, which are then input into the encoder. The decoder then uses these representations to directly generate quantile forecasts across multiple future steps—a method known as direct multi-step forecasting. Chronos-Bolt models are up to 250 times faster and 20 times more memory-efficient than the original Chronos models of the same size. The following plot compares the inference time of Chronos-Bolt against the original Chronos models for forecasting 1024 time series with a context length of 512 observations and a prediction horizon of 64 steps.

  • Ovis2.6-30B-A3B↗

    ATH-MaaS

    We introduce **Ovis2.6-30B-A3B**, the latest advancement in the Ovis series of Multimodal Large Language Models (MLLMs). Building on the strong foundation of Ovis2.5, Ovis2.6 upgrades the LLM backbone to a **Mixture-of-Experts (MoE)** architecture, delivering superior multimodal performance at a fraction of the serving cost. It also brings major improvements in long-context and high-resolution understanding, visual reasoning with active image analysis, and information-dense document comprehension. The following table presents a detailed performance comparison. Please note that superscripted results are sourced from external technical reports, and Qwen scores represent the highest value between its Think and Instruct versions. For quick reference, the best results are highlighted in red, and the second-best results are underlined. All values are rounded to one decimal place. The thinking and thinking budget logic can be applied in the same way for multi-image, video and pure text scenarios.

  • GLM-5.3-Flash-abliterated-Maya-M-GGUF↗

    alesha-pro

    Project Maya's Maya-M quant of GLM-5.3-Flash plus a 2.2 MB vector that removes refusals while the model runs, and a small patch for the Maya engine that applies it. On 4x RTX 3090 it decodes 56 tokens/s at a 256K context with the vector on. The model weights in this repo are not mine. They are Maya-M and the vision files of Project Maya, copied byte for byte from peasantsmith/GLM-5.3-Flash-Maya-GGUF at revision `3e9f3bb20aa191e01c6fb850faf132d8149da98a`. The SHA256 of all 5 files match theirs at that revision. See `SHA256SUMS`. The authors changed the first shard after I downloaded it. Their commit `1084950a` renames the architecture in the GGUF header to `glm5-next`, and the file size is the same. This repo has the earlier first shard, the one all my runs used with Maya v1.0.27. For the current files go to their repo.

  • Bonsai-Image-ternary-4B↗

    litert-community

    LiteRT is Google's on-device runtime, the new name for TensorFlow Lite (Android: `com.google.ai.edge.litert:litert`), and `litert-torch`, the renamed `ai-edge-torch`, is its PyTorch converter: a PyTorch model converted unmodified with `litert_torch.convert` matched the original to 4e-7 on a Galaxy S26 (measured, LiteRT 2.2.0, Android 16, 2026-09-05). PrismML Bonsai Image 4B — a ternary-weight diffusion transformer built on the FLUX.2-klein-4B architecture — converted to LiteRT for on-device text-to-image. The whole pipeline runs in three fixed-shape `.tflite` graphs; the host does only tokenization, the FlowMatch-Euler loop, and latent unpatchify (`generate.py` in this repo, ~150 lines, no torch and no diffusers). Smallest working set: **3.97 GiB** (int4 text encoder). Output is 512×512 with the default pair of graphs, or 256×256 with the `dit_256` / `vae_dec_256` pair (`generate.py --size 256`; `variants.256` in `pipeline_meta.json`) — the text encoder, tokenizer and latent constants are shared by both sizes; 256 prompt tokens and 4 sampling steps by default (the model is step-distilled; more steps also work).

  • InternVL3-14B↗

    OpenGVLab

    We introduce InternVL3, an advanced multimodal large language model (MLLM) series that demonstrates superior overall performance. Compared to InternVL 2.5, InternVL3 exhibits superior multimodal perception and reasoning capabilities, while further extending its multimodal capabilities to encompass tool usage, GUI agents, industrial image analysis, 3D vision perception, and more. Additionally, we compare InternVL3 with Qwen2.5 Chat models, whose corresponding pre-trained base models are employed as the initialization of the langauge component in InternVL3. Benefitting from Native Multimodal Pre-Training, the InternVL3 series achieves even better overall text performance than the Qwen2.5 series. As shown in the following figure, InternVL3 retains the same model architecture as InternVL 2.5 and its predecessors, InternVL 1.5 and 2.0, following the "ViT-MLP-LLM" paradigm. In this new version, we integrate a newly incrementally pre-trained InternViT with various pre-trained LLMs, including InternLM 3 and Qwen 2.5, using a randomly initialized MLP projector. As in the previous version, we applied a pixel unshuffle operation, reducing the number of visual tokens to one-quarter of the original. Besides, we adopted a similar dynamic resolution strategy as InternVL 1.5, dividing images into tiles of 448×448 pixels. The key difference, starting from InternVL 2.0, is that we additionally introduced support for multi-image and video data.