spooknik

This repository contains Nunchaku-quantized (SVDQ) versions of Fluxmania Legacy, a text-to-image model based on Flux.1 Dev by AdelAI If you like the model, please consider liking, reviewing and tipping the creator. Model Files - svdq-int4r32-fluxmania-legacy.safetensors: SVDQuant INT4 (rank 32) Fluxmania Legacy model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-fluxmania-legacy.safetensors: SVDQuant NVFP4 (rank 32) Fluxmania Legacy model. For users with Blackwell GPUs (50-series). Below is the quality and similarity evaluated with 256 samples from MJHQ-30K dataset. (BF16 is the unqauntized model. INT W4A4 is INT4 and NVFP4 is FP4) | Model | Precision | Method | FID | IR | LPIPS | PSNR | |----------------------------|-----------|-----------|--------------------|-----------------|----------------------|-------------------| | Fluxmania Legacy (25 step) | BF16 | -- | 175.08 | 0.895 | -- | -- | | | INT W4A4 | SVDQ | 174.84 | 0.795 | 0.222 | 22.84 | | | NVFP4 | SVDQ | 173.10 | 0.884 | 0.192 | 23.38 |

This repository contains Nunchaku-quantized (SVDQ) versions of CenKreChro, a text-to-image model based on Chroma and Flux Krea merged by TiwazM If you like the model, please consider liking, reviewing and tipping the creator. Model Files - svdq-int4r32-CenKreChro.safetensors: SVDQuant INT4 (rank 32) CenKreChro 1.0 model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-CenKreChro.safetensors: SVDQuant NVFP4 (rank 32) CenKreChro 1.0 model. For users with Blackwell GPUs (50-series). - svdq-int4r32-CenKreChro-V2.3.safetensors: SVDQuant INT4 (rank 32) CenKreChro 2.3 model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-CenKreChro-V2.3.safetensors: SVDQuant NVFP4 (rank 32) CenKreChro 2.3 model. For users with Blackwell GPUs (50-series). Below is the quality and similarity evaluated with 256 samples from MJHQ-30K dataset. (BF16 is the unqauntized model. INT W4A4 is INT4 and NVFP4 is FP4) | Model | Precision | Method | FID | IR | LPIPS | PSNR | |----------------------------|-----------|-----------|--------------------|-----------------|----------------------|-------------------| | CenKreChro (25 Steps) | BF16 | -- | 135.52 | 0.842 | -- | -- | | | INT W4A4 | SVDQ | 135.50 | 0.795 | 0.255 | 20.64 | | | NVFP4 | SVDQ | 134.19 | 0.849 | 0.219 | 21.30 |

This repository contains Nunchaku-quantized (SVDQ) versions of CyberRealistic Flux, a text-to-image model based on Flux.1 Dev by Cyberdelia If you like the model, please consider liking, reviewing and tipping the creator. Model Files - svdq-int4r32-CyberRealistic-Flux-V2.5.safetensors: SVDQuant INT4 (rank 32) CyberRealistic Flux V2.5 model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-CyberRealistic-Flux-V2.5.safetensors: SVDQuant NVFP4 (rank 32) CyberRealistic Flux V2.5. For users with Blackwell GPUs (50-series). Below is the quality and similarity evaluated with 256 samples from MJHQ-30K dataset. (BF16 is the unqauntized model. INT W4A4 is INT4 and NVFP4 is FP4) | Model | Precision | Method | FID | IR | LPIPS | PSNR | |----------------------------|-----------|-----------|--------------------|-----------------|----------------------|-------------------| | CyberRealistic-Flux-V2.5 | BF16 | -- | 173.06 | 0.909 | -- | -- | | (30 Steps) | INT W4A4 | SVDQ | 173.07 | 0.902 | 0.215 | 21.04 | | | NVFP4 | SVDQ | 173.54 | 0.901 | 0.222 | 20.55 |

This repository contains Nunchaku-quantized (SVDQ) versions of Jib Mix Flux, a text-to-image model based on Flux.1-Dev by J1B If you like the model, please consider liking, reviewing and tipping the creator. Model Files - svdq-int4r32-Jib-Mix-Flux-V12.safetensors: SVDQuant INT4 (rank 32) Jib Mix Flux V12 model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-Jib-Mix-Flux-V12.safetensors: SVDQuant NVFP4 (rank 32) Jib Mix Flux V12 model. For users with Blackwell GPUs (50-series). Below is the quality and similarity evaluated with 256 samples from MJHQ-30K dataset. (BF16 is the unqauntized model. INT W4A4 is INT4 and NVFP4 is FP4) | Model | Precision | Method | FID | IR | LPIPS | PSNR | |----------------------------|-----------|-----------|--------------------|-----------------|----------------------|-------------------| | Jib Flux Mix V12 (25 steps)| BF16 | -- | 170.37 | 0.946 | -- | -- | | | INT W4A4 | SVDQ | 171.64 | 0.907 | 0.253 | 20.31 | | | NVFP4 | SVDQ | 170.99 | 0.955 | 0.220 | 20.71 |

This repository contains Nunchaku-quantized (SVDQ) versions of PixelWave, a text-to-image model based on Flux.1 Dev by humblemikey Model Files - svdq-int4r32-pixelwave-schnell-04.safetensors: SVDQuant INT4 (rank 32) PixelWave FLUX.1-schnell 04. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-pixelwave-schnell-04.safetensors: SVDQuant NVFP4 (rank 32) PixelWave FLUX.1-schnell 04. For users with Blackwell GPUs (50-series). - svdq-int4r32-pixelwave-dev-0.3.safetensors: SVDQuant INT4 (rank 32) PixelWave FLUX.1-dev 03. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-pixelwave-dev-0.3.safetensors: SVDQuant NVFP4 (rank 32) PixelWave FLUX.1-dev 03. For users with Blackwell GPUs (50-series). Below is the quality and similarity evaluated with 256 samples from MJHQ-30K dataset. (BF16 is the unqauntized model. INT W4A4 is INT4 and NVFP4 is FP4) | Model | Precision | Method | FID | IR | LPIPS | PSNR | |----------------------------|-----------|-----------|--------------------|-----------------|----------------------|-------------------| | PixelWave schnell 04 | BF16 | -- | 176.37 | 0.813 | -- | -- | | (8 step) | INT W4A4 | SVDQ | 176.68 | 0.820 | 0.322 | 17.12 | | | NVFP4 | SVDQ | 176.91 | 0.839 | 0.298 | 17.70 |

This repository contains Nunchaku-quantized (SVDQ) versions of project0, a text-to-image model based on Flux Dev and Flux Krea created by speach1sdef178 If you like the model, please consider liking, reviewing and tipping the creator. Model Files - svdq-int4r32-Project0-REAL1SMV3.safetensors: SVDQuant INT4 (rank 32) Project0-REAL1SMV3 model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-Project0-REAL1SMV3.safetensors: SVDQuant NVFP4 (rank 32) Project0-REAL1SMV3 model. For users with Blackwell GPUs (50-series). - svdq-int4r32-Project0-KREA.safetensors: SVDQuant INT4 (rank 32) Project0-KREA model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-Project0-KREA.safetensors: SVDQuant NVFP4 (rank 32) Project0-KREA model. For users with Blackwell GPUs (50-series).

This repository contains Nunchaku-quantized (SVDQ) versions of Nepotism, a text-to-image model based on Flux by BobsBlazed If you like the model, please consider liking, reviewing and tipping the creator. Model Files - svdq-int4r32-Nepotism-XI.safetensors: SVDQuant INT4 (rank 32) Nepotism XI model. For users with non-Blackwell GPUs (pre-50-series). - svdq-fp4r32-Nepotism-XI.safetensors: SVDQuant NVFP4 (rank 32) Nepotism XI model. For users with Blackwell GPUs (50-series).

0. TL;DR 1. Model Details 2. Usage 3. Uses 4. Bias, Risks, and Limitations 5. Training Details 6. Evaluation 7. Environmental Impact 8. Citation If you already know T5, FLAN-T5 is just better at everything. For the same number of parameters, these models have been fine-tuned on more than 1000 additional tasks covering also more languages. As mentioned in the first few lines of the abstract : > Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks, such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints,1 which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models. Disclaimer: Content from this model card has been written by the Hugging Face team, and parts of it were copy pasted from the T5 model card. - Model type: Language model - Language(s) (NLP): English, German, French - License: Apache 2.0 - Related Models: All FLAN-T5 Checkpoints - Original Checkpoints: All Original FLAN-T5 Checkpoints - Resources for more information: - Research paper - GitHub Repo - Hugging Face FLAN-T5 Docs (Similar to T5) Find below some example scripts on how to use the model in `transformers`: Running the model on a GPU using different precisions The authors write in the original paper's model card that: > The primary use is research on language models, including: research on zero-shot NLP tasks and in-context few-shot learning NLP tasks, such as reasoning, and question answering; advancing fairness and safety research, and understanding limitations of current large language models The information below in this section are copied from the model's official model card: > Language models, including Flan-T5, can potentially be used for language generation in a harmful way, according to Rae et al. (2021). Flan-T5 should not be used directly in any application, without a prior assessment of safety and fairness concerns specific to the application. > Flan-T5 is fine-tuned on a large corpus of text data that was not filtered for explicit content or assessed for existing biases. As a result the model itself is potentially vulnerable to generating equivalently inappropriate content or replicating inherent biases in the underlying data. > Flan-T5 has not been tested in real world applications. > Flan-T5 should not be applied for any unacceptable use cases, e.g., generation of abusive speech. The model was trained on a mixture of tasks, that includes the tasks described in the table below (from the original paper, figure 2): According to the model card from the original paper: > These models are based on pretrained T5 (Raffel et al., 2020) and fine-tuned with instructions for better zero-shot and few-shot performance. There is one fine-tuned Flan model per T5 model size. The model has been trained on TPU v3 or TPU v4 pods, using `t5x` codebase together with `jax`. The authors evaluated the model on various tasks covering several languages (1836 in total). See the table below for some quantitative evaluation: For full results for FLAN-T5-XXL, see the research paper, Table 3. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019). - Hardware Type: Google Cloud TPU Pods - TPU v3 or TPU v4 | Number of chips ≥ 4. - Hours used: More information needed - Cloud Provider: GCP - Compute Region: More information needed - Carbon Emitted: More information needed