This model weight is identical to laion/CLIP-ViT-H-14-laion2B-s32B-b79K, but with the pytorch_model component only (without open_clip_pytorch_model.bin).

This is to support loading the model as a ClipModel, as I failed to load the original model using AutoModel (feedback appreciated)

With this distribution, I was finally able to load from AutoModel, and further support image classification tasks using my self-defined class CLIPViTForImageClassification listed below.

However, there is still a small issue that I cannot resolve, I can only load the model if I git clone this repo to local, if I load from web, the loading still fails.

from transformers.models.clip.modeling_clip import CLIPPreTrainedModel, CLIPConfig, CLIPVisionTransformer
from transformers.modeling_outputs import (
    BaseModelOutput,
    BaseModelOutputWithPooling,
    ImageClassifierOutput,
    MaskedImageModelingOutput,
)
from typing import Dict, List, Optional, Set, Tuple, Union
import torch
from torch import nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss


class CLIPViTForImageClassification(CLIPPreTrainedModel):
    def __init__(self, config: CLIPConfig) -> None:
        super().__init__(config)

        self.num_labels = config.num_labels
        vision_config = config.vision_config
        self.vision_model = CLIPVisionTransformer(vision_config)

        # Classifier head
        
        self.classifier = nn.Linear(vision_config.hidden_size, config.num_labels) if config.num_labels > 0 else nn.Identity()

        # Initialize weights and apply final processing
        self.post_init()

    def forward(
        self,
        pixel_values: Optional[torch.Tensor] = None,
        #head_mask: Optional[torch.Tensor] = None,
        labels: Optional[torch.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        #interpolate_pos_encoding: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[tuple, ImageClassifierOutput]:
        r"""
        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
            Labels for computing the image classification/regression loss. Indices should be in `[0, ...,
            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        outputs = self.vision_model(
            pixel_values,
            #head_mask=head_mask,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            #interpolate_pos_encoding=interpolate_pos_encoding,
            return_dict=return_dict,
        )

        sequence_output = outputs[0]

        logits = self.classifier(sequence_output[:, 0, :])

        loss = None
        if labels is not None:
            # move labels to correct device to enable model parallelism
            labels = labels.to(logits.device)
            if self.config.problem_type is None:
                if self.num_labels == 1:
                    self.config.problem_type = "regression"
                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
                    self.config.problem_type = "single_label_classification"
                else:
                    self.config.problem_type = "multi_label_classification"

            if self.config.problem_type == "regression":
                loss_fct = MSELoss()
                if self.num_labels == 1:
                    loss = loss_fct(logits.squeeze(), labels.squeeze())
                else:
                    loss = loss_fct(logits, labels)
            elif self.config.problem_type == "single_label_classification":
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
            elif self.config.problem_type == "multi_label_classification":
                loss_fct = BCEWithLogitsLoss()
                loss = loss_fct(logits, labels)

        if not return_dict:
            output = (logits,) + outputs[1:]
            return ((loss,) + output) if loss is not None else output

        return ImageClassifierOutput(
            loss=loss,
            logits=logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )