Saved toolsSubmit a tool
FOXAI SEARCH

Search FoxAI

AI Models 2119

  • MioCodec-25Hz-44.1kHz-v2↗

    Aratako

    **MioCodec-25Hz-44.1kHz-v2** is an upsampled, high-fidelity version of the MioCodec-25Hz-24kHz model. By integrating an **UpsamplerBlock** inspired by Inworld TTS-1 into the decoder, this model reconstructs 44.1 kHz audio from the standard 25 Hz token stream. This model is a fine-tuned version of `MioCodec-25Hz-24kHz` with the following architectural enhancements:

  • Qwen2.5-7B-Instruct-heretic↗

    saidutta69

    A decensored variant of Qwen/Qwen2.5-7B-Instruct, produced with Heretic v1.4.0 (directional ablation / "abliteration"). Refusal behavior is suppressed via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the base model's knowledge and instruction-following are left largely intact. **Who this is for:** developers who want a mid-size, locally-runnable Qwen2.5 model that answers directly instead of refusing or lecturing - for local agents, roleplay, research on alignment/refusal mechanics, or any use case blocked by RLHF-era over-refusal. Not a general capability upgrade over base Qwen2.5-7B-Instruct - treat it as the same model with refusal-shaped guardrails removed. Weights only, at this model's 7.6B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up.

  • mms-tts-vie↗

    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.

  • Parable-Granite-4.1-8B-Claude-Fable-5-GGUF↗

    AnkitAI

    > **~6 GB of RAM is all you need.** The Q4 build fits comfortably on an ordinary laptop or a mid-range GPU. > One command and you have a private, offline reasoning model on your machine: > > Parable-Granite-4.1-8B is an ibm-granite/granite-4.1-8b fine-tune trained on real multi-step agent sessions: planning, tool use, and ` ` reasoning captured from actual Claude Fable 5 and GPT-5.5 agent work, not synthetic Q&A. It is the largest release in the Parable series. **🔮 v2 is coming.** The 3B just got the v2 treatment (13× corpus, rebuilt recipe) — the same upgrade lands here next. Same links, in-place.

  • Qwen3.8-27B-Uncensored-Heretic-MTP-UD-GGUF↗

    codavidgarcia

    GGUF quants of llmfan46/Qwen3.8-27B-Uncensored-Heretic-Native-MTP-Preserved, built so the model **and its native MTP head** fit fully in VRAM on common consumer cards. Each file applies Unsloth's per-tensor "UD" recipe for the base Qwen3.8-27B, copied tensor by tensor, to llmfan46's weights. The MTP block (`blk.64`) stays at q6_K / q8_0 in every tier. Mean KL divergence against a Q8_0 reference of the same BF16 (lower is better), with the share of tokens whose top-1 prediction matches the reference. ¹ Better than UD-IQ4_XS, but at 14.26 GiB it does not leave room for MTP plus a useful context on a 16 GB card.

  • wav2vec2-BERT-cantonese↗

    alvanlii

    This model is a fine-tuned version of facebook/w2v-bert-2.0. This has a CER of 10.27 on Common Voice 16 (yue) test set (without punctuations). For training, three datasets were used: - Common Voice 16 `zh-HK` and `yue` Train Set - CantoMap: Winterstein, Grégoire, Tang, Carmen and Lai, Regine (2020) "CantoMap: a Hong Kong Cantonese MapTask Corpus", in Proceedings of The 12th Language Resources and Evaluation Conference, Marseille: European Language Resources Association, p. 2899-2906. - Cantonse-ASR: Yu, Tiezheng, Frieske, Rita, Xu, Peng, Cahyawijaya, Samuel, Yiu, Cheuk Tung, Lovenia, Holy, Dai, Wenliang, Barezi, Elham, Chen, Qifeng, Ma, Xiaojuan, Shi, Bertram, Fung, Pascale (2022) "Automatic Speech Recognition Datasets in Cantonese: A Survey and New Dataset", 2022. Link: https://arxiv.org/pdf/2201.02419.pdf

  • Agnes-AI_Agnes-3.0-Qwen-GGUF↗

    bartowski

    **Model details:** - Parameter count: 180B (source checkpoint) - Input support: text, image (with mmproj file) - details - Speculative decoding: yes (MTP) - details - imatrix: yes - details **Don't know which to choose?** Grab Q4_K_M (127.09GB) - usually a good mix of size and performance. Download instructions available here **About these sizes:** 51.7 % of this model's weights (`per_layer_token_embd` (160 wide), `ffn_down_exps` (640 wide), `hc_attn_up` (320 wide), `hc_ffn_up` (320 wide) and 2 more) have row widths that llama.cpp's 256-wide K and IQ block types cannot use, so each size gives them one of the 32-wide types instead (Q4_0 to Q8_0, or IQ4_NL), whatever type its name asks for, and the files above are closer together in size than usual. These sizes were not made, each within 1 % of the size of another file of the same kind: Q4_K_L (0.9 % from Q5_K_S), IQ2_XXS (0.7 % from IQ2_XS).

  • sam-3d-body-vith↗

    facebook

  • whistle↗

    Cactus-Compute

    A speech-to-text model for mobiles, wearables, robots, smart home, automotive and microcontrollers. The whole model is a single 16.9 MB file, and it runs on the same CPU engine as Needle, from the same container and the same quantisation, with no dependencies and no GPU. - **Transcription**: 16 kHz mono audio of any length, one pass up to 30 seconds and windowed beyond it, in English, German, French, Spanish, Italian, Dutch and Polish. The language is detected unless you name it, and silence and steady noise return an empty transcript rather than an invented sentence; other sounds, music especially, can still come out as text, so put a voice activity detector or a wake word in front of an always-on microphone. Live audio streams for as long as you like, committing the words two passes agree on. - **Word timestamps**: every word with its start, end and probability, aligned from the decoder's own attention, so an app can highlight, seek or cut on a word. - **Speech embedding**: the encoder output, one row per 80 ms frame, for matching and retrieval without decoding a transcript at all. Keyword biasing takes the names, places, product words and jargon your users actually say and favours them during the search, so the ones that matter survive.

  • distilbert-NER↗

    dslim

    If my open source models have been useful to you, please consider supporting me in building small, useful AI models for everyone (and help me afford med school / help out my parents financially). Thanks! **distilbert-NER** is the fine-tuned version of **DistilBERT**, which is a distilled variant of the BERT model. DistilBERT has fewer parameters than BERT, making it smaller, faster, and more efficient. distilbert-NER is specifically fine-tuned for the task of **Named Entity Recognition (NER)**. This model accurately identifies the same four types of entities as its BERT counterparts: location (LOC), organizations (ORG), person (PER), and Miscellaneous (MISC). Although it is a more compact model, distilbert-NER demonstrates a robust performance in NER tasks, balancing between size, speed, and accuracy.