Saved toolsSubmit a tool
FOXAI SEARCH

Search FoxAI

AI Models 2119

  • MOSS-Audio-Tokenizer-Nano↗

    OpenMOSS-Team

    This repository contains the Hugging Face remote-code implementation and weights for **MOSS-Audio-Tokenizer-Nano**, the lightweight audio tokenizer used by **MOSS-TTS-Nano**. MOSS-Audio-Tokenizer-Nano is a compact discrete audio tokenizer based on the **Cat** (**C**ausal **A**udio **T**okenizer with **T**ransformer) architecture from MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models. The checkpoint in this repository has **21,969,664 parameters** (approximately **22M**), making it much smaller than the full-size MOSS-Audio-Tokenizer while preserving the 48 kHz stereo tokenizer interface used by the MOSS-TTS family. - **Small model size**: approximately **22M parameters**, including about 10.45M encoder parameters, 10.45M decoder parameters, and 1.07M quantizer parameters. - **Native high-resolution audio**: supports **48 kHz** input and output with **2-channel stereo** audio, helping reduce compression loss and improve listening quality. - **Low-frame-rate discrete codes**: compresses 48 kHz stereo audio into a **12.5 Hz** token stream with a downsample rate of 7,680 samples. - **Variable bitrate reconstruction**: uses a residual quantizer stack with **16 codebooks** and 1,024 entries per codebook. Each codebook contributes about **0.125 kbps**, for an inference range from **0.125 kbps to 2 kbps**. - **Transformer-based tokenizer**: uses causal Transformer blocks and supports low-latency streaming encode/decode. - **MOSS-TTS family interface**: designed as the audio tokenizer backbone for MOSS-TTS-Nano and compatible MOSS-TTS-family workflows.

  • PP-OCRv6_small_det_onnx↗

    PaddlePaddle

    PP-OCRv6: From 1.5M to 34.5M Parameters, Surpassing Billion-Scale VLMs on OCR Tasks **🔥 Official Website** **📝 Technical Report** PP-OCRv6 is a lightweight OCR system that combines architectural innovation with data-centric optimization. It redesigns the backbone, detection neck, and recognition neck around a unified MetaFormer-style building block with structural reparameterization. Three model tiers (medium, small, tiny) share the same block primitives, covering deployment scenarios from server to edge.

  • arc-agi2-joint-loopvit-epochs-trainium-v3↗

    ritwika96

    Fresh joint-context ARC model: all demonstrations and the query share attention. Not an upstream checkpoint or paper reproduction. Check metrics.jsonl for completed optimizer updates; compilation is not training. Checkpoints are saved on the remote Trainium host and uploaded here, not downloaded to the workstation. Dataset policy and sources are in manifest.json. Spatial ConvGLU: False. Dataset policy: all-public-kaggle-only. With all-public training, local metrics are training diagnostics, not a benchmark. Snapshots are 300-DPI PNGs plus vector SVGs. Epochs: 100; sampling: shuffled without replacement. Epoch checkpoints preserve the permutation and cursor.

  • LynnReal-Onmi-light-vae↗

    stdstu123

    A complete codec with the unchanged MiniMax H3 encoder and a distilled 26-block video decoder. Latent channels, normalization, spatial scale and temporal compression match the official VAE. Deployment uses the decoder EMA parameters. `config.json` and the indexed Safetensors shards follow the Diffusers component layout. `decode_config.json` records the release decoder settings. Load with `model.light_vae.LightVAE.from_pretrained("weight/light-vae")`, or pass `--light-vae weight/light-vae` to `script/sample.py`. Native decoding tiles are the default; precision and tile geometry must accompany speed/quality comparisons. The codec changes reconstruction, not the denoiser latent interface. The same latent need not decode to identical RGB under the official and lightweight VAE. See `LICENSE` and the release `NOTICE` for MiniMax H3 terms and attribution.

  • RouWei-0.8↗

    Minthy

    Dataset of 13M unique pictures (~4M with natural text captions) picked and balanced from over 25M of anime art, covers, digital illustrations, western media and other sources, including private datasets. More detailed description on Civitai Vpred version is now available. It works flawlessly out of box without any burning or related issues. Consider to use lower CFG (3..5), other generation settings are same. Some exotic and experimental samplers/schedulers untested. When you are prompting artist styles, especially mixing several, their tags **MUST BE in a separate CLIP chunk**. Add `BREAK` after it (for A1111 and derivatives), use conditioning concat node (for Comfy) or **at least put them in the very end**. Otherwise, **significant degradation of results** is likely.

  • Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2↗

    nerkyor

    **GGUF repository: nerkyor/Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2-GGUF** The main and standalone GGUF repositories provide all six tiers. These are formal full-suite frozen scores, with every non-passing sample retained in the denominator. Main-repository paths add the `GGUF/` prefix: > **Q2-LynnStyle uses GSQ-RCO mixed-precision quantization with IQ numerical refinement.** Its exact 12,999,977,600-byte build has no frozen TPS result. > > Q2 LCB disclosure: the 75/100 final view preserves 99 original rows and uses one Lynn-authorized exact retry for a streaming-JSON failure; the retry again reached 32K with empty code.

  • korean_PP-OCRv5_mobile_rec↗

    PaddlePaddle

    korean_PP-OCRv5_mobile_rec is one of the PP-OCRv5_rec that are the latest generation text line recognition models developed by PaddleOCR team. It aims to efficiently and accurately support the recognition of Korean. The key accuracy metrics are as follow: **Note**: If any character (including punctuation) in a line was incorrect, the entire line was marked as wrong. This ensures higher accuracy in practical applications. You can also integrate the model inference of the text recognition module into your project. Before running the following code, please download the sample image to your local machine.

  • Gestura-FastH3-GGUF↗

    ZawShiShawn

    App-level measurements on Samsung Galaxy S26 (SM-S9420), **256 × 256, 5-second output option (124 frames), 24 fps**: Gestura is an Android app for on-device AI generation, including text-to-video, first-frame image-to-video and reference-conditioned video with audio. The models in these repositories were converted or packaged during its development. Q4_K_M GGUF conversion of the FastH3 Dense four-step diffusion model. It provides quantized denoiser weights for implementations of the FastH3 architecture, reducing model storage compared with full-precision weights.

  • LeVJEPA-VideoMix-Large↗

    galilai-group

    A ViT-L/16 video encoder trained with LeV-JEPA, a self-supervised objective combining a multi-crop prediction loss with SIGReg. No labels are used at any point. `trust_remote_code=True` is required: this is a custom architecture (RoPE + block-causal attention) rather than a stock `transformers` model, so the modeling code ships with the weights. Normalise with the ImageNet statistics used in training — `mean=[0.485, 0.456, 0.406]`, `std=[0.229, 0.224, 0.225]` — and resize/crop to 224. Frames are sampled at roughly 7.5 fps in training, so a 16-frame clip covers about two seconds.

  • creative-writing-control-vectors-v3.0↗

    jukofyork

    This repo contains pre-generated control vectors in GGUF format for use with llama.cpp: - **IMPORTANT**: These **new control vectors** must use their **respective de-bias control vector(s)**. - The code used to generate these can now be found at github.com/jukofyork/control-vectors. - All were generated with `'--num_prompt_samples'` set to the model's hidden state dimension. Control vectors allow fine-tuned control over LLMs, enabling more precise/targeted text generation.