サイト内検索

AI モデル 2119

• LTX2.3_comfy↗

  Kijai

  The fp8 quantizations were done with the basic static weight scales and are set to not run with fp8 matmuls, the models marked `input_scaled` additionally have activation scaling, and are set to run with fp8 matmuls on supported hardware (roughly 40xx and later Nvidia GPUs). As this is first time I'm attempting to do calibrate input scales, these are pretty experimental, but result wise seems to work, this is a test on a 4090, 8 steps with distill: **update:** input_scaled_v3 follows the same pattern as the official one where blocks 0-1 and 46-47 (first two and last two) are kept in bf16, as well as better calibrated input scales, this does fix some of the issues in v2, especially when using input audio.

• smollm2-1.7b-gsm8k↗

  EiadEssam

• seresnext50_32x4d.racm_in1k↗

  timm

  A SE-ResNeXt-B image classification model with Squeeze-and-Excitation channel attention. This model features: * ReLU activations * single layer 7x7 convolution with pooling * 1x1 convolution shortcut downsample * grouped 3x3 bottleneck convolutions * Squeeze-and-Excitation channel attention Recipe details: * RandAugment `RACM` recipe. Inspired by and evolved from EfficientNet RandAugment recipes. Published as `B` recipe in ResNet Strikes Back. * RMSProp (TF 1.0 behaviour) optimizer, EMA weight averaging * Step (exponential decay w/ staircase) LR schedule with warmup

• Qwen3.8-Flash-Next-Uncensored-MLX-Serve-4bit↗

  ARC4NUM

  The **uncensored** Qwen3.8-Flash-Next (125B total / 6B active MoE) packed for mlx-serve, so it actually fits and flies on a 128 GB Apple Silicon Mac: > The other ways to run this model uncensored on a Mac either don't fit 128 GB > (the original MLX repo is 163 GiB) or run slower (GGUF IQ4_XS ≈ 26 tok/s on > the same hardware; generic Python MLX paths ≈ 27–36 tok/s). This is a 4-bit MLX-Serve pack of **orcarouter/Qwen3.8-Flash-Next-Uncensored** — the full-precision **BF16 abliterated** (refusal-removed) build of Qwen's Qwen3.8-Flash-Next. See orcarouter's card for the full abliteration methodology and evaluation; this section summarizes it.

• Bonsai-2-27B-Ternary-CRACK-GGUF↗

  dealignai

  **Abliterated · No guardrails · PQ2_0 ternary 2.13 bpw · 7.2 GB · Runs on a laptop / single GPU · Vision-capable** > ⚠️ **Re-download notice (2026-09-17 20:44 PDT / 2026-09-18 03:44 UTC)** — an earlier build of this model had a coherence bug in reasoning modes (`low`/`xhigh`) that could cause token loops on some prompts. This version fixes it. If you downloaded before this timestamp, please pull the latest .gguf. **Proprietary weight-level abliteration** by the dealignai research team. Byte-identical to the base ternary quant everywhere except a small set of tensors that carry the refusal circuit. Drop-in replacement for the ternary base at inference — same tokenizer, same chat template, same reasoning modes, same vision projector interface.

• mask2former-swin-large-ade-panoptic↗

  facebook

  Mask2Former model trained on ADE20k panoptic segmentation (large-sized version, Swin backbone). It was introduced in the paper Masked-attention Mask Transformer for Universal Image Segmentation and first released in this repository. Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. Mask2Former addresses instance, semantic and panoptic segmentation with the same paradigm: by predicting a set of masks and corresponding labels. Hence, all 3 tasks are treated as if they were instance segmentation. Mask2Former outperforms the previous SOTA, MaskFormer both in terms of performance an efficiency by (i) replacing the pixel decoder with a more advanced multi-scale deformable attention Transformer, (ii) adopting a Transformer decoder with masked attention to boost performance without without introducing additional computation and (iii) improving training efficiency by calculating the loss on subsampled points instead of whole masks.

• Reason-ModernColBERT↗

  lightonai

  Reason-ModernColBERT is a late interaction model trained on the reasonir-hq dataset. It achieves extremely competitive performance on the BRIGHT benchmark aimed at evaluating reasoning-intensive retrieval performance, outperforming all existing models up to 7B (more than 45 times its size) and even surprisingly improving performance of ReasonIR-8B (a 8B model trained on the same data) by more than 2.5 NDCG@10 on average on Stack Exchange splits. We attribute such strong results to late-interaction, see evaluation section. Unfortunately, since the ReasonIR data has been released under a cc-by-nc-4.0 license, we cannot release this model under an Apache 2.0 license. However, the authors of ReasonIR released code to generate the data. Anyone willing to reproduce the data could then easily reproduce this model under an Apache 2.0 license by running a fine-tuning lasting lower than 2 hours using this boilerplate. This is a multi-vector (ColBERT-style late interaction) embedding model finetuned from lightonai/GTE-ModernColBERT-v1 on the reasonir-hq dataset. It maps sentences & paragraphs to sequences of 128-dimensional dense vectors and can be used for semantic textual similarity using the MaxSim operator.

• CognitiveFusion2-4x7B-BF16↗

  Kquant03

  This is an update to the original Cognitive Fusion. We intend to perform a fine-tune on it in order to increase its performance. The scale of a model is one of the most important axes for better model quality. Given a fixed computing budget, training a larger model for fewer steps is better than training a smaller model for more steps. Mixture of Experts enable models to be pretrained with far less compute, which means you can dramatically scale up the model or dataset size with the same compute budget as a dense model. In particular, a MoE model should achieve the same quality as its dense counterpart much faster during pretraining.

• tabpfn_3_5↗

  Prior-Labs

  TabPFN-3.5 is a transformer-based foundation model that uses in-context learning to solve tabular prediction problems in a forward pass. One checkpoint serves both classification and regression. Inference code can be found at https://github.com/PriorLabs/TabPFN. `TabPFNRegressor` works the same way, off the same checkpoint. For more examples, see the github repo: https://github.com/PriorLabs/TabPFN. Every checkpoint ships with basic text and datetime column handling enabled (`TRANSFORM_TEXT`, `TRANSFORM_DATES`) and a feature limit of 20,000.

• Ingot-III-27B-Q2↗

  dendriteholdings

  A public mirror of a model that won on Ingot, a Bittensor subnet where miners quantize a fixed teacher model and the validator crowns the copy that stays closest to it. It is Prism ML's Ternary Bonsai 2 27B, refined: the same ternary codes, with the per-block scales and the norm weights re-tuned by the miner to bring it closer to Qwen3.8-27B. The weights are the miner's upload, byte for byte. Credit belongs to the miner, and to Prism ML for Bonsai 2.