Model URLs - React Native ExecuTorch

LLM Models

Llama 3.2

LLAMA3_2_3B

Llama 3.2 3B model (BF16 precision).

export const LLAMA3_2_3B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LLAMA3_2_3B_QLORA

Llama 3.2 3B model with QLoRA quantization.

export const LLAMA3_2_3B_QLORA = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LLAMA3_2_3B_SPINQUANT

Llama 3.2 3B model with SpinQuant quantization.

export const LLAMA3_2_3B_SPINQUANT = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LLAMA3_2_1B

Llama 3.2 1B model (BF16 precision).

export const LLAMA3_2_1B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LLAMA3_2_1B_QLORA

Llama 3.2 1B model with QLoRA quantization.

export const LLAMA3_2_1B_QLORA = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LLAMA3_2_1B_SPINQUANT

Llama 3.2 1B model with SpinQuant quantization.

export const LLAMA3_2_1B_SPINQUANT = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

Qwen 3

QWEN3_0_6B

Qwen 3 0.6B model (BF16 precision).

export const QWEN3_0_6B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN3_0_6B_QUANTIZED

Qwen 3 0.6B model with 8-bit quantization.

export const QWEN3_0_6B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN3_1_7B

Qwen 3 1.7B model (BF16 precision).

export const QWEN3_1_7B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN3_1_7B_QUANTIZED

Qwen 3 1.7B model with 8-bit quantization.

export const QWEN3_1_7B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN3_4B

Qwen 3 4B model (BF16 precision).

export const QWEN3_4B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN3_4B_QUANTIZED

Qwen 3 4B model with 8-bit quantization.

export const QWEN3_4B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

Hammer 2.1

HAMMER2_1_0_5B

Hammer 2.1 0.5B model (BF16 precision).

export const HAMMER2_1_0_5B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

HAMMER2_1_0_5B_QUANTIZED

Hammer 2.1 0.5B model with 8-bit quantization.

export const HAMMER2_1_0_5B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

HAMMER2_1_1_5B

Hammer 2.1 1.5B model (BF16 precision).

export const HAMMER2_1_1_5B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

HAMMER2_1_1_5B_QUANTIZED

Hammer 2.1 1.5B model with 8-bit quantization.

export const HAMMER2_1_1_5B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

HAMMER2_1_3B

Hammer 2.1 3B model (BF16 precision).

export const HAMMER2_1_3B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

HAMMER2_1_3B_QUANTIZED

Hammer 2.1 3B model with 8-bit quantization.

export const HAMMER2_1_3B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

SmolLM 2

SMOLLM2_1_135M

SmolLM 2 135M model (BF16 precision).

export const SMOLLM2_1_135M = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

SMOLLM2_1_135M_QUANTIZED

SmolLM 2 135M model with 8-bit quantization.

export const SMOLLM2_1_135M_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

SMOLLM2_1_360M

SmolLM 2 360M model (BF16 precision).

export const SMOLLM2_1_360M = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

SMOLLM2_1_360M_QUANTIZED

SmolLM 2 360M model with 8-bit quantization.

export const SMOLLM2_1_360M_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

SMOLLM2_1_1_7B

SmolLM 2 1.7B model (BF16 precision).

export const SMOLLM2_1_1_7B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

SMOLLM2_1_1_7B_QUANTIZED

SmolLM 2 1.7B model with 8-bit quantization.

export const SMOLLM2_1_1_7B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

Qwen 2.5

QWEN2_5_0_5B

Qwen 2.5 0.5B model (BF16 precision).

export const QWEN2_5_0_5B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN2_5_0_5B_QUANTIZED

Qwen 2.5 0.5B model with 8-bit quantization.

export const QWEN2_5_0_5B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN2_5_1_5B

Qwen 2.5 1.5B model (BF16 precision).

export const QWEN2_5_1_5B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN2_5_1_5B_QUANTIZED

Qwen 2.5 1.5B model with 8-bit quantization.

export const QWEN2_5_1_5B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN2_5_3B

Qwen 2.5 3B model (BF16 precision).

export const QWEN2_5_3B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

QWEN2_5_3B_QUANTIZED

Qwen 2.5 3B model with 8-bit quantization.

export const QWEN2_5_3B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

Phi 4

PHI_4_MINI_4B

Phi 4 Mini 4B model (BF16 precision).

export const PHI_4_MINI_4B = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

PHI_4_MINI_4B_QUANTIZED

Phi 4 Mini 4B model with 8-bit quantization.

export const PHI_4_MINI_4B_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LFM 2.5

LFM2_5_1_2B_INSTRUCT

LFM 2.5 1.2B Instruct model (FP16 precision).

export const LFM2_5_1_2B_INSTRUCT = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

LFM2_5_1_2B_INSTRUCT_QUANTIZED

LFM 2.5 1.2B Instruct model with 8-bit quantization.

export const LFM2_5_1_2B_INSTRUCT_QUANTIZED = {
  modelSource: string,
  tokenizerSource: string,
  tokenizerConfigSource: string,
};

Classification Models

EFFICIENTNET_V2_S

EfficientNet V2 Small classification model. Platform-optimized (CoreML for iOS, XNNPACK for Android).

export const EFFICIENTNET_V2_S = {
  modelSource: string,
};

Object Detection Models

SSDLITE_320_MOBILENET_V3_LARGE

SSDLite 320 MobileNet V3 Large object detection model.

export const SSDLITE_320_MOBILENET_V3_LARGE = {
  modelName: 'ssdlite-320-mobilenet-v3-large',
  modelSource: string,
} as const;

RF_DETR_NANO

RF-DETR Nano object detection model.

export const RF_DETR_NANO = {
  modelName: 'rf-detr-nano',
  modelSource: string,
} as const;

Style Transfer Models

STYLE_TRANSFER_CANDY

Candy style transfer model. Platform-optimized (CoreML for iOS, XNNPACK for Android).

export const STYLE_TRANSFER_CANDY = {
  modelSource: string,
};

STYLE_TRANSFER_MOSAIC

Mosaic style transfer model. Platform-optimized (CoreML for iOS, XNNPACK for Android).

export const STYLE_TRANSFER_MOSAIC = {
  modelSource: string,
};

STYLE_TRANSFER_RAIN_PRINCESS

Rain Princess style transfer model. Platform-optimized (CoreML for iOS, XNNPACK for Android).

export const STYLE_TRANSFER_RAIN_PRINCESS = {
  modelSource: string,
};

STYLE_TRANSFER_UDNIE

Udnie style transfer model. Platform-optimized (CoreML for iOS, XNNPACK for Android).

export const STYLE_TRANSFER_UDNIE = {
  modelSource: string,
};

Speech to Text Models

WHISPER_TINY_EN

Whisper Tiny English-only model.

export const WHISPER_TINY_EN = {
  isMultilingual: false,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

WHISPER_TINY_EN_QUANTIZED

Whisper Tiny English-only model with quantization.

export const WHISPER_TINY_EN_QUANTIZED = {
  isMultilingual: false,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

WHISPER_BASE_EN

Whisper Base English-only model.

export const WHISPER_BASE_EN = {
  isMultilingual: false,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

WHISPER_SMALL_EN

Whisper Small English-only model.

export const WHISPER_SMALL_EN = {
  isMultilingual: false,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

WHISPER_TINY

Whisper Tiny multilingual model.

export const WHISPER_TINY = {
  isMultilingual: true,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

WHISPER_BASE

Whisper Base multilingual model.

export const WHISPER_BASE = {
  isMultilingual: true,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

WHISPER_SMALL

Whisper Small multilingual model.

export const WHISPER_SMALL = {
  isMultilingual: true,
  encoderSource: string,
  decoderSource: string,
  tokenizerSource: string,
};

Semantic Segmentation Models

DEEPLAB_V3_RESNET50

DeepLab V3 with ResNet50 backbone.

export const DEEPLAB_V3_RESNET50 = {
  modelName: 'deeplab-v3-resnet50',
  modelSource: string,
} as const;

DEEPLAB_V3_RESNET101

DeepLab V3 with ResNet101 backbone.

export const DEEPLAB_V3_RESNET101 = {
  modelName: 'deeplab-v3-resnet101',
  modelSource: string,
} as const;

DEEPLAB_V3_MOBILENET_V3_LARGE

DeepLab V3 with MobileNet V3 Large backbone.

export const DEEPLAB_V3_MOBILENET_V3_LARGE = {
  modelName: 'deeplab-v3-mobilenet-v3-large',
  modelSource: string,
} as const;

LRASPP_MOBILENET_V3_LARGE

LRASPP with MobileNet V3 Large backbone.

export const LRASPP_MOBILENET_V3_LARGE = {
  modelName: 'lraspp-mobilenet-v3-large',
  modelSource: string,
} as const;

FCN_RESNET50

FCN with ResNet50 backbone.

export const FCN_RESNET50 = {
  modelName: 'fcn-resnet50',
  modelSource: string,
} as const;

FCN_RESNET101

FCN with ResNet101 backbone.

export const FCN_RESNET101 = {
  modelName: 'fcn-resnet101',
  modelSource: string,
} as const;

Quantized Variants

Quantized versions are also available:

DEEPLAB_V3_RESNET50_QUANTIZED
DEEPLAB_V3_RESNET101_QUANTIZED
DEEPLAB_V3_MOBILENET_V3_LARGE_QUANTIZED
LRASPP_MOBILENET_V3_LARGE_QUANTIZED
FCN_RESNET50_QUANTIZED
FCN_RESNET101_QUANTIZED

SELFIE_SEGMENTATION

Selfie segmentation model for person/background separation.

export const SELFIE_SEGMENTATION = {
  modelName: 'selfie-segmentation',
  modelSource: string,
} as const;

Image Embeddings Models

CLIP_VIT_BASE_PATCH32_IMAGE

CLIP ViT Base Patch32 image encoder.

export const CLIP_VIT_BASE_PATCH32_IMAGE = {
  modelSource: string,
};

Text Embeddings Models

ALL_MINILM_L6_V2

All-MiniLM-L6-v2 text embedding model.

export const ALL_MINILM_L6_V2 = {
  modelSource: string,
  tokenizerSource: string,
};

ALL_MPNET_BASE_V2

All-MPNet-base-v2 text embedding model.

export const ALL_MPNET_BASE_V2 = {
  modelSource: string,
  tokenizerSource: string,
};

MULTI_QA_MINILM_L6_COS_V1

Multi-QA-MiniLM-L6-cos-v1 text embedding model.

export const MULTI_QA_MINILM_L6_COS_V1 = {
  modelSource: string,
  tokenizerSource: string,
};

MULTI_QA_MPNET_BASE_DOT_V1

Multi-QA-MPNet-base-dot-v1 text embedding model.

export const MULTI_QA_MPNET_BASE_DOT_V1 = {
  modelSource: string,
  tokenizerSource: string,
};

CLIP_VIT_BASE_PATCH32_TEXT

CLIP ViT Base Patch32 text encoder.

export const CLIP_VIT_BASE_PATCH32_TEXT = {
  modelSource: string,
  tokenizerSource: string,
};

Image Generation Models

BK_SDM_TINY_VPRED_512

BK-SDM-Tiny V-Pred 512x512 diffusion model.

export const BK_SDM_TINY_VPRED_512 = {
  schedulerSource: string,
  tokenizerSource: string,
  encoderSource: string,
  unetSource: string,
  decoderSource: string,
};

BK_SDM_TINY_VPRED_256

BK-SDM-Tiny V-Pred 256x256 diffusion model.

export const BK_SDM_TINY_VPRED_256 = {
  schedulerSource: string,
  tokenizerSource: string,
  encoderSource: string,
  unetSource: string,
  decoderSource: string,
};

Voice Activity Detection Models

FSMN_VAD

FSMN Voice Activity Detection model.

export const FSMN_VAD = {
  modelSource: string,
};

Documentation Index

​LLM Models

​Llama 3.2

​LLAMA3_2_3B

​LLAMA3_2_3B_QLORA

​LLAMA3_2_3B_SPINQUANT

​LLAMA3_2_1B

​LLAMA3_2_1B_QLORA

​LLAMA3_2_1B_SPINQUANT

​Qwen 3

​QWEN3_0_6B

​QWEN3_0_6B_QUANTIZED

​QWEN3_1_7B

​QWEN3_1_7B_QUANTIZED

​QWEN3_4B

​QWEN3_4B_QUANTIZED

​Hammer 2.1

​HAMMER2_1_0_5B

​HAMMER2_1_0_5B_QUANTIZED

​HAMMER2_1_1_5B

​HAMMER2_1_1_5B_QUANTIZED

​HAMMER2_1_3B

​HAMMER2_1_3B_QUANTIZED

​SmolLM 2

​SMOLLM2_1_135M

​SMOLLM2_1_135M_QUANTIZED

​SMOLLM2_1_360M

​SMOLLM2_1_360M_QUANTIZED

​SMOLLM2_1_1_7B

​SMOLLM2_1_1_7B_QUANTIZED

​Qwen 2.5

​QWEN2_5_0_5B

​QWEN2_5_0_5B_QUANTIZED

​QWEN2_5_1_5B

​QWEN2_5_1_5B_QUANTIZED

​QWEN2_5_3B

​QWEN2_5_3B_QUANTIZED

​Phi 4

​PHI_4_MINI_4B

​PHI_4_MINI_4B_QUANTIZED

​LFM 2.5

​LFM2_5_1_2B_INSTRUCT

​LFM2_5_1_2B_INSTRUCT_QUANTIZED

​Classification Models

​EFFICIENTNET_V2_S

​Object Detection Models

​SSDLITE_320_MOBILENET_V3_LARGE

​RF_DETR_NANO

​Style Transfer Models

​STYLE_TRANSFER_CANDY

​STYLE_TRANSFER_MOSAIC

​STYLE_TRANSFER_RAIN_PRINCESS

​STYLE_TRANSFER_UDNIE

​Speech to Text Models

​WHISPER_TINY_EN

​WHISPER_TINY_EN_QUANTIZED

​WHISPER_BASE_EN

​WHISPER_SMALL_EN

​WHISPER_TINY

​WHISPER_BASE

​WHISPER_SMALL

​Semantic Segmentation Models

​DEEPLAB_V3_RESNET50

​DEEPLAB_V3_RESNET101

​DEEPLAB_V3_MOBILENET_V3_LARGE

​LRASPP_MOBILENET_V3_LARGE

​FCN_RESNET50

​FCN_RESNET101

​Quantized Variants

​SELFIE_SEGMENTATION

​Image Embeddings Models

​CLIP_VIT_BASE_PATCH32_IMAGE

​Text Embeddings Models

​ALL_MINILM_L6_V2

​ALL_MPNET_BASE_V2

​MULTI_QA_MINILM_L6_COS_V1

​MULTI_QA_MPNET_BASE_DOT_V1

​CLIP_VIT_BASE_PATCH32_TEXT

​Image Generation Models

​BK_SDM_TINY_VPRED_512