全站搜索
AI 模型 2119
- NPO-SAM-WMDP-llama3-8b-instruct-i1-GGUF↗
mradermacher
- Qwen3.8-27B-QUASAR-NVFP4↗
QUASAR-QAT
**19.7 GB · 496/496 transformer linears in NVFP4 (W4A4) · 0.909 GPQA-Diamond vs. 0.914 BF16 · serves directly with vLLM.** **Among the public NVFP4 builds compared below, QUASAR is both the smallest and the highest-scoring: 90.9 GPQA-Diamond and 100% AIME'26.** **Independent evaluation:** A third-party NVFP4 shootout on the official Qwen3.8-27B repo compares QUASAR against other public NVFP4 checkpoints under the same evaluation and serving setup.
- qwen3.8-flash-next-gguf↗
antirez
Self-contained GGUFs for DwarfStar, with original BF16 n-grams read directly from disk. Keep the file on a local SSD. No n-gram sidecar is needed. MTP weights are included; enable them with `--mtp`. Requires DwarfStar's native BF16 n-gram reader. Older versions expecting `--ple` cannot load these files. Each file contains the same 95.37 GiB BF16 n-gram table. It is not made resident or quantized. Context and runtime buffers require additional RAM. The main/MTP tensor bytes come unchanged from Ivan Fioravanti's Q2 and Q4 releases. Q2 has imatrix-calibrated IQ2_XXS gate/up and Q2_K down routed experts, with the down rows padded from 640 to 768 inputs. Q4 has calibrated Q4_K gate/up and MXFP4 down. Other tensor formats and MTP weights are unchanged.
- mms-tts-kaz↗
facebook
This model is part of Facebook's Massively Multilingual Speech project, aiming to provide speech technology across a diverse range of languages. You can find more details about the supported languages and their ISO 639-3 codes in the MMS Language Coverage Overview, and see all MMS-TTS checkpoints on the Hugging Face Hub: facebook/mms-tts. VITS (**V**ariational **I**nference with adversarial learning for end-to-end **T**ext-to-**S**peech) is an end-to-end speech synthesis model that predicts a speech waveform conditional on an input text sequence. It is a conditional variational autoencoder (VAE) comprised of a posterior encoder, decoder, and conditional prior. A set of spectrogram-based acoustic features are predicted by the flow-based module, which is formed of a Transformer-based text encoder and multiple coupling layers. The spectrogram is decoded using a stack of transposed convolutional layers, much in the same style as the HiFi-GAN vocoder. Motivated by the one-to-many nature of the TTS problem, where the same text input can be spoken in multiple ways, the model also includes a stochastic duration predictor, which allows the model to synthesise speech with different rhythms from the same input text.
- NexusRaven-V2-13B-GGUF↗
TheBloke
GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp. The creator of the source model has listed its license as `other`, and this quantization has therefore used that same license. As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
- normistral-7b-warm-instruct↗
ltg
This is a model instruction-tuned on open datasets released under the most permissive apache-2.0 licence (in other words, we don't use any datasets generated by ChatGPT) — thus we can release this model under the same license and make it openly available for commercial applications. The model has been finetuned on 4096 context length, twice as many tokens as the base model. The released weights are still a work in progress and they might change in the future. This is the first iteration of instruction-tuning our NorMistral models and it currently uses only the SFT phase without any preference optimization. Please let us know your feedback to improve the model in future releases. 1. We gathered all openly available datasets: Aya, OASST 1, OASST 2, OIG-small-chip2, No Robots, Dolly and Glaive code assistant. 2. These were first manually inspected and filtered, and then automatically filtered with Mixtral-8x7B to remove incorrect, offensive, non-English and American-centric responses. 3. The responses were augmented to be more descriptive by Mixtral-8x7B. 4. Since most of that dataset contains only a single dialogue turn, we generated more turns using Mixtral-8x7B. 5. Finally, we translated the resulting dataset into Bokmål and Nynorsk using NorMistral-7b-warm.
- LaMini-GPT-124M↗
MBZUAI
This model is one of our LaMini-LM model series in paper "LaMini-LM: A Diverse Herd of Distilled Models from Large-Scale Instructions". This model is a fine-tuned version of gpt2 on LaMini-instruction dataset that contains 2.58M samples for instruction fine-tuning. For more information about our dataset, please refer to our project repository. You can view other models of LaMini-LM series as follows. Models with ✩ are those with the best overall performance given their size/architecture, hence we recommend using them. More details can be seen in our paper. We recommend using the model to respond to human instructions written in natural language. Since this decoder-only model is fine-tuned with wrapper text, we suggest using the same wrapper text to achieve the best performance. See the example on the right or the code below. We initialize with gpt2 and fine-tune it on our LaMini-instruction dataset. Its total number of parameters is 124M.
- Wan-Move-14B-480P↗
Ruihang
**Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance [Paper]** Ruihang Chu, Yefei He, Zhekai Chen, Shiwei Zhang, Xiaogang Xu, Bin Xia, Dingdong Wang, Hongwei Yi, Xihui Liu, Hengshuang Zhao, Yu Liu, Yingya Zhang, Yujiu Yang We present our NeurIPS 2025 paper Wan-Move, a simple and scalable motion-control framework for video generation. Wan-Move offers the following key features: - 🎯 **High-Quality 5s 480p Motion Control**: Through scaled training, Wan-Move can generate 5-second, 480p videos with SOTA motion controllability on par with commercial systems such as Kling 1.5 Pro’s Motion Brush, as verified via user studies. - 🧩 **Novel latent Trajectory Guidance**: Our core idea is to represent the motion condition by propagating the first frame’s features along the trajectory, which can be seamlessly integrated into off-the-shelf image-to-video models (e.g., Wan-I2V-14B) without any architecture change or extra motion modules. Wan-Move spports diverse motion control applications in image-to-video generation. The generated samples (832×480p, 5s) exhibits high visual fidelity and accurate motion.
- GLM-5.3-Flash-Maya-GGUF↗
peasantsmith
**Maya-L: 99.2% of the full FP8 model's accuracy** on zero-shot tasks (ARC, HellaSwag, WinoGrande, PIQA), in 156 GB. **Maya-S: 97.9%** in a 96 GB file. **Maya-S24: 97.7%, and up to 14% faster decode on 24 GB cards**, in 94.7 GB. **Maya-M: 97.9%, and closer to the full model token by token**, in 116 GB. GGUF quantizations of zai-org/GLM-5.3-Flash (321 B parameters, a mixture of experts with about 18 B active per token), made for **Project Maya**. Project Maya keeps the most-used experts on the GPU, the next in RAM and the rest on the SSD, so all four run well below the model's size (two GPUs with 30 GB of RAM in the measurements below). More memory is faster, and every machine is different: try it on yours. All four are made from Z.ai's own FP8 release - the precision the model is served at - not from a re-quantized file. - **Maya-S** (96.5 GB) is the compact one, made for PCs with a smaller memory pool across RAM and VRAM: its routed experts - most of the model - in about 2 bits (IQ2_XXS / IQ2_S), its attention and shared experts in 6-bit (Q6_K). - **Maya-S24** (94.7 GB) is Maya-S for 24 GB cards (RTX 3090 / 4090): the same 2-bit routed experts, with only the small part every token runs through - the attention and shared experts - in 4-bit (Q4_K) instead of Maya-S's 6-bit. That leaves about 1.5 GB more room for experts on the GPU. - **Maya-M** (116 GB) is closer to the full model, for PCs with a bigger memory pool: built with the same FP8-statistics recipe and more bits where they count (2- to 3-bit experts), calibrated toward tool calls and front-end code. - **Maya-L** (156.3 GB) is the closest, for PCs with the biggest memory pool: Maya-M's recipe one step up (3- to 4-bit experts: IQ3_S / IQ4_XS, Q5_K in the most sensitive layers).
- occamy-1.0-abliterated-FIT-GGUF↗
SC117
FIT-GGUF v0.4.0 GATE-VERIFIED TIERS 5 FIT + 4 APEX-I- + BF16 ABLITERATED (INPUT-GATED) MEASURED KL + SAME-TOP VISION mmproj APACHE-2.0 Occamy-1.0-abliterated · FIT-GGUF Five fidelity tiers of a 256K-context 35B MoE with vision — the smallest GGUF that provably meets each KL gate, re-verified on its own bytes. Plus an independent APEX-I- family and the BF16 reference, all measured by the same protocol. 🧭 About FIT-GGUF — the tool behind these files Every FIT tier here was planned, executed and verified by FIT-GGUF , an open-source, deterministic tensor-level planning layer on top of llama.cpp quantization. Standard GGUF quantization asks you to pick one of a handful of presets; FIT-GGUF instead asks what quality do you want , then finds and verifies the smallest GGUF that demonstrably meets it. Traditional GGUF gives you presets. FIT gives you a fidelity contract: macro KL ≤ tier anchor, measured against this model's own BF16 on five fixed domains. Deterministic size prediction & byte-exact delivery ✅ Validated (G2 gate, delta = 0) Universally optimal tensor allocation ⚠️ Not established — FIT claims verified fidelity contracts, not a universal quality optimum Method, pre-registered research record and the fit CLI are all open source: github.com/Scorp1o117/FIT-GGUF ⚠️ Safety notice / 安全提示 The source model is an abliterated, refusal-removed model with no meaningful built-in guardrails, and may comply with harmful, illegal or unsafe requests. Use it only where you can provide appropriate moderation, access control and legal review. Do not deploy it to end users without your own safety layer. 源模型经过拒答方向消融，不具备可靠的内置安全护栏；请仅在合法、受控、具备审核与滥用防护的环境中使用，使用者自行承担部署责任。