Modelpublic

RichardErkhov/LeroyDyer_-_SpydazWebAI_VisionEncoderDecoderModel_Mini548m-awq

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes2downloads

Model Card

Quantization made by Richard Erkhov.

Github

Discord

Request more models

SpydazWebAIVisionEncoderDecoderModelMini548m - AWQ

—Model creator: https://huggingface.co/LeroyDyer/
—Original model: https://huggingface.co/LeroyDyer/SpydazWebAIVisionEncoderDecoderModelMini548m/

Original model description: --- license: mit language:

—en library_name: transformers tags:
—vision
—VISION-ENCODER-DECODER-MODEL pipeline_tag: image-to-text ---

ADD HEAD


Mistral
VISION-ENCODER-DECODER-MODEL

print('Add Vision...')
# ADD HEAD
# Combine pre-trained encoder and pre-trained decoder to form a Seq2Seq model



Vmodel = VisionEncoderDecoderModel.from_encoder_decoder_pretrained(
    "google/vit-base-patch16-224-in21k", "LeroyDyer/Mixtral_AI_Tiny"
)
_Encoder_ImageProcessor = Vmodel.encoder
_Decoder_ImageTokenizer = Vmodel.decoder
_VisionEncoderDecoderModel = Vmodel
# Add Pad tokems
LM_MODEL.VisionEncoderDecoder = _VisionEncoderDecoderModel
# Add Sub Components
LM_MODEL.Encoder_ImageProcessor = _Encoder_ImageProcessor
LM_MODEL.Decoder_ImageTokenizer = _Decoder_ImageTokenizer
LM_MODEL