Open-Weight Models

Model

defect-aware-MAE

Shrey Patel

A Vision Transformer encoder pretrained without labels on steel-surface images with a masked autoencoder adapted to texture defects. Frozen features are clustered into defect categories. No class labels, bounding boxes or detection objective are used in pretraining, and no contrastive loss. Encoder: 224×224 grayscale input, 16×16 patches (196 tokens), 6 pre-norm Transformer layers, 12 heads, 768-d. The decoder in the MAE checkpoint has 2 layers and a 36-d output head: Decoder(depth=2) with dec.pred = torch.nn.Linear(256, 36). NEU-DET (hot-rolled steel strip, 6 defect classes, 1,800 images). Pretraining used only the 1,440-image training split. The 360-image validation split was held out…

Model

deit-classification

Fiona Reyes

A research-oriented DeiT prototype targeting Classification. The included large setup documents defaults and file formats without presenting unverified performance numbers. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses sgd with a linear warmup schedule. These are starting values in the script, not evidence of a completed run. For…

Model

deit-contrastive

Sophia Harris

This repository is a compact, custom PyTorch implementation of DeiT for Contrastive. The small configuration is intended for code review, smoke tests, and small controlled experiments rather than as a production-ready pretrained release. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses sgd with a exponential schedule. These are…

Model

deit-multitask-practice

Yui Nakamura

A research-oriented DeiT prototype targeting Multitask. The included xlarge setup documents defaults and file formats without presenting unverified performance numbers. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses novograd with a constant warmup schedule. These are starting values in the script, not evidence of a completed run.…

This repository is a compact, custom PyTorch implementation of DeiT for Retrieval. The xlarge configuration is intended for code review, smoke tests, and small controlled experiments rather than as a production-ready pretrained release. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses sgd with a cosine schedule. These are starting…

Model · Text generation

delta

Convergent Intelligence

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Model · Robotics

demodoctor-act-clean-s2

Akteruzzaman Raihan Sikder

Action Chunking with Transformers (ACT) is an imitation-learning method that predicts short action chunks instead of single steps. It learns from teleoperated data and often achieves high success rates. This policy has been trained and pushed to the Hub using LeRobot. Learn how to train and run it in the LeRobot act guide, or browse the full documentation. The policy consumes these observation features and produces these action features. Inputs Outputs New to LeRobot? These guides cover the full workflow: - Install LeRobot — set up the lerobot package. - Hardware setup — assemble, wire, and calibrate your robot and cameras. - Record data & train a policy — the end-to-end imitation-learning…

Model

denoising_autoencoder

George Perdrizet

A convolutional autoencoder trained to remove Gaussian noise from 256×256 RGB images. The model learns the manifold of clean images and projects noisy inputs back onto this learned space. This model removes Gaussian noise from corrupted images by learning to map noisy images back to their clean versions. During training, the autoencoder learns the distribution of clean images, enabling it to filter out noise while preserving image content. - σ=75 (on 0-255 scale) Gaussian noise - Approximately 0.294 on [0,1] normalized scale This model is designed for educational purposes to demonstrate how autoencoders can learn to denoise images by understanding the structure of clean data. - Exploring…

This repository contains the frozen baseline for the from-scratch Blender multi-object generation experiment. - 3,600 total scenes - 18 scene families - 2–6 objects - 2,400 train - 600 validation - 600 test - family-disjoint split - 50,020,480 parameters - dmodel: 640 - tied embeddings The model learned Blender/Python program grammar very well, but numeric attribute binding remains the primary failure mode. training/checkpoints/best.pt Design 3D-002-R2 will use a structured scene representation and a deterministic compiler to Blender Python, while preserving the same underlying dataset and family-disjoint evaluation protocol. Official Blender 3.0.1 is used. The Blender executable itself is…

Frozen conditional Blender scene-generation experiment. - 50,008,320 unique parameters - dmodel: 640 - tied token embeddings / LM head [BOS] instruction [SEP] target [EOS] target + EOS Instruction and SEP positions are masked. 4293b1f043ca6875a5bc927f4e7b297cbf5b5122983effbc8cfc202494dc190e Conditional training successfully removed the unconditional output collapse observed in R2. However, semantic generation remains limited by: - object under-generation - incomplete object state - duplicate field emission - weak relation endpoint binding - invalid/self relations - location numeric collapse - long-horizon state tracking Design 3D-002-R2D addresses these problems using: - explicit object…

Model

designgym-sft-qwen3.6-35b-a3b-r32

Adithya S K

This model is a fine-tuned version of Qwen/Qwen3.6-35B-A3B. It has been trained using TRL. This model was trained with SFT.

Dev is an open-source 399M parameter bidirectional decision model built on ModernBERT-large. It is purpose-built for unstructured-to-structured classification—including ticket routing, yes/no verification, and rating scales—executing in a single forward pass (~28ms on MPS) without token generation. Following Jev (TypeSafe) and Kev-0.5B (Jared Palmer), Dev tests a fundamental architectural question: What if decision models shouldn't be causal decoders at all, but native bidirectional cross-encoders? Evaluated against Jared Palmer's kev-0.5b (built on a frozen Qwen-2.5-0.5B causal backbone + 9.3M LoRA pointer head): Inference Latency: Dev-0.4B executes in 27.6ms on Apple Silicon MPS (M1 Max)…