Initial commit

README.md

@@ -14,7 +14,7 @@ library_name: transformers
 pipeline_tag: feature-extraction
 ---
 
-📢 [2026-05-20] The pretrained SPECTRE can now be loaded directly from the `transformers` library. Check below for the details.
+📢 [2026-05-20] The pretrained SPECTRE model can now be loaded directly through the `transformers` library, no separate SPECTRE package installation required. Check below for details and usage examples.
 
 📢 [2026-04-10] SPECTRE is now an official baseline for the [**CVPR 2026 Workshop Competition: Foundation Models for General CT Image Diagnosis**](https://www.codabench.org/competitions/12650/)! See `experiments/cvpr26_fm_for_ct_diag_task_1` for scripts and additional details.  
 
@@ -30,8 +30,8 @@ pipeline_tag: feature-extraction
   <a href="https://pypi.org/project/spectre-fm/"><img alt="Python Versions" src="https://img.shields.io/pypi/pyversions/spectre-fm?style=flat-square&cacheSeconds=0" /></a>
   <a href="https://pypi.org/project/spectre-fm/"><img alt="Downloads per Month" src="https://img.shields.io/pypi/dm/spectre-fm?style=flat-square&label=downloads&cacheSeconds=0" /></a>
   <a href="https://github.com/cclaess/SPECTRE/blob/main/LICENSE"><img alt="License" src="https://img.shields.io/github/license/cclaess/SPECTRE?style=flat-square&cacheSeconds=0" /></a>
-  <a href="https://huggingface.co/cclaess/SPECTRE"><img alt="Model weights" src="https://img.shields.io/badge/models-Hugging%20Face-yellow?style=flat-square&cacheSeconds=0" /></a>
-  <a href="https://arxiv.org/abs/2511.17209"><img alt="Paper" src="https://img.shields.io/badge/paper-arXiv-b31b1b?style=flat-square&cacheSeconds=0" /></a>
+  <a href="https://huggingface.co/cclaess/SPECTRE-Large"><img alt="Model weights" src="https://img.shields.io/badge/model-Hugging%20Face-yellow?style=flat-square&cacheSeconds=0" /></a>
+  <a href="https://arxiv.org/abs/2511.17209"><img alt="Preprint" src="https://img.shields.io/badge/preprint-arXiv-b31b1b?style=flat-square&cacheSeconds=0" /></a>
 </p>
 
 <p align="center">
@@ -45,7 +45,7 @@ SPECTRE has been trained on a large cohort of **open-source CT scans** of the **
 This repository provides pretrained SPECTRE models together with tools for fine-tuning and evaluation.
 
 ## 🧠 Pretrained Models
-The pretrained SPECTRE model can easily be imported using `AutoModel` from the `transformers` library
+The pretrained SPECTRE model can easily be imported using the `transformers` library
 
 ```python
 from transformers import AutoModel

modeling_spectre.py

@@ -36,6 +36,6 @@ class SpectreModel(PreTrainedModel):
         outputs = self.model(pixel_values, grid_size=grid_size)
 
         if not return_dict:
-            return (outputs,)
+            return outputs
 
         return BaseModelOutput(last_hidden_state=outputs)

spectre/models/__init__.py

@@ -14,18 +14,6 @@ from .vision_transformer_features import (
     feat_vit_base,
     feat_vit_large,
 )
-from .resnet import (
-    ResNet,
-    resnet18,
-    resnet34, 
-    resnet50, 
-    resnet101, 
-    resnext50,
-    resnext101,
-)
-from .eomt import EoMT
-from .seomt import SEoMT
-from .upsample_anything import UPA
 
 __all__ = [
     'VisionTransformer',

spectre/utils/__init__.py

@@ -6,29 +6,6 @@ from ._utils import (
     to_3tuple,
     to_4tuple,
 )
-from .checkpointing import (
-    save_state,
-    load_state,
-    extract_model_from_checkpoint_dinov2,
-    extract_model_from_checkpoint_siglip,
-)
-from .collate import (
-    extended_collate_dino,
-    extended_collate_siglip,
-    collate_add_filenames,
-)
-from .config import setup
-from .dataloader import get_dataloader
-from .distributed import (
-    is_enabled,
-    get_global_size,
-    get_global_rank,
-    get_local_size,
-    get_local_rank,
-    init_distributed,
-)
-from .lora import add_lora_adapters
-from .masking import random_block_mask
 from .modeling import (
     deactivate_requires_grad_and_to_eval,
     activate_requires_grad_and_to_train,
@@ -54,13 +31,6 @@ from .modeling import (
     nchwd_to,
     nhwdc_to,
 )
-from .param_groups import get_param_groups_with_decay
-from .scheduler import (
-    linear_warmup_schedule,
-    cosine_schedule,
-    cosine_warmup_schedule,
-    CosineWarmupScheduler,
-)
 
 __all__ = [
     "fix_random_seeds",