Text Classification
Transformers
Safetensors
English
bert
nlp
routing
vision-task-classifier
text-embeddings-inference
Instructions to use beingamanforever/ICM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use beingamanforever/ICM with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="beingamanforever/ICM")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, TaskClassifier tokenizer = AutoTokenizer.from_pretrained("beingamanforever/ICM") model = TaskClassifier.from_pretrained("beingamanforever/ICM", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Commit History
-
Add dataset JSON files
Aman | अमनa286609
-
Update README.md
Verified signaturebb87f91 -
Update README.md
Verified signaturec482d55 -
Create README.md
Verified signature# Task Classification Model (ICM) ## Model Description A BERT-based sequence classification model that routes computer vision questions to appropriate specialized modules. Classifies questions into 4 task categories: VQA, Captioning, Grounding, and Geometry. - **Repository:** beingamanforever/ICM - **Base Model:** bert-base-uncased - **Task:** 4-way Sequence Classification ## Labels | ID | Label | Description | |---|---|---| | 0 | vqa | Visual Question Answering ("What color is the car?") | | 1 | captioning | Image Description ("Describe the sunset.") | | 2 | grounding | Object Localization ("Find the person in the image.") | | 3 | geometry | Spatial/Metric Queries ("Calculate the area of the red box.") | ## Architecture BERT-Base encoder + 3-layer MLP classifier on [CLS] token: - Layer 1: Linear(768 → 256) + ReLU + Dropout(0.1) - Layer 2: Linear(256 → 128) + ReLU + Dropout(0.1) - Layer 3: Linear(128 → 4) ## Training | Hyperparameter | Value | |---|---| | Samples | 1,600 (4…2dba87c -
Upload folder using huggingface_hub
Verified signature392638a -
initial commit
Verified signaturede54b8d