🔄 In a Training Loop

SSH PRO

ProCreations

·

https://ssh.codes

AI & ML interests

Making AI’s small, accessible, and easy to run on consumer hardware

Recent Activity

posted an update about 10 hours ago

check out this cool webgpu thing i made! https://huggingface.co/spaces/ProCreations/maple-webgpu runs this model https://huggingface.co/deepgrove/maple-preview with cool performance i found today on WebGPU locally on your computer with very fast speeds. made with a mix of gpt 5.6 sol + opus 5 ultracode, its amazing what agents can do now, just a year ago this wouldve been impossible with them.

liked a model about 10 hours ago

deepgrove/maple-preview

liked a model about 11 hours ago

ProCreations/maple-preview-webgpu-v2

View all activity

Organizations

ProCreations 's Spaces 353

Reproduction - Adaptive Multi-Round Allocation

Exact greedy allocation and a polynomial surrogate DP

Reproduction - Rotation-Invariant Spherical Watermarking via Third-Order SO(3) Coupling

Primary-source rotation and panoramic-fidelity audit

Reproduction: FluxNet

FluxNet conservation audit and table contradiction

Reproduction logbook — Recursive Binding on a Budget: Subspace Carving in Order-p Tensor Memories

Independent paper-setting computation of OSC retrieval

Reproduction - RL for Reachability

Asymptotically optimal reachability-RL audit

Reproduction logbook — A Fully First-Order Layer for Differentiable Optimization

Native evidence audit for FFOLayer's six registered claims.

Reproduction logbook — On the Power of Source Screening for Learning Shared Feature Extractors

Independent five-claim source-screening audit.

Repro - Stability Analysis of Sharpness-Aware Minimization

Full-CIFAR SAM stability reproduction

Independent claim audit

Audit LLM claim consistency with Bayesian stopping analysis

Reproduction logbook — Fixed Budget is No Harder Than Fixed Confidence in Best-Arm Identification up to Logarithmic Factors

Executed reduction and application-scale audit for

Reproduction Token Sample Complexity of Attention

Native BERT/WikiText scale audit

Reproduction: A Capacity-Based Rationale for Multi-Head Attention

Capacity scaling audit for multi-head attention

Reproduction logbook — Provably Data-driven Multiple Hyper-parameter Tuning with Structured Loss Function

Executed pseudo-dimension evidence, six claims

Conformal C2ST: Turning weak classifiers into strong two-sample tests

Explore a reproducibility audit of a two‑sample test paper

Evidence index

Browse evidence for Bayesian inference claims

Reproduction logbook — Learning to Bet for Horizon-Aware Anytime-Valid Testing

Finite-horizon betting theorem and DQN evaluation audit

Reproduction logbook — Loss-aware Distributionally Robust Optimization via Trainable Optimal Transport Ambiguity Sets

Loss-aware OT-DRO formulation and convergence

Calibrated Preference Learning: The Case of Label Ranking

Verify research paper claims with reproducibility logbook

Reproduction logbook — Differentially Private Range Subgraph Counting

Five decisive paper-setting claim checks

ICML 2026 reproduction - Linear Regression with Unknown Truncation

Run reproducible linear regression truncation experiments

Reproduction logbook — Why Linear Recurrent Memory Works in Partially Observable Reinforcement Learning

RingWorld PPO evidence for five paper claims.

Reproduction logbook — LEC: Linear Expectation Constraints for Selection-Conditioned Risk Control in Selective Prediction and Routing Systems

Independent CommonsenseQA LEC power audit.

Reproduction - Rotary Position Encodings for Graphs

WIRE graph rotation and native benchmark repair audit

Reproduction logbook — Distributed Direct Preference Optimization

Official real-model reproduction.