Baseline implementation of recurrent PPO using truncated BPTT
-
Updated
Apr 28, 2024 - Jupyter Notebook
Baseline implementation of recurrent PPO using truncated BPTT
Official PyTorch Implementation for the "Distilling Datasets Into Less Than One Image" paper.
[TMLR] S-TLLR: STDP-inspired Temporal Local Learning Rule for Spiking Neural Networks
SuperMimic: humanoid whole-body control with one differentiable world model — implementation on mjlab / MuJoCo Warp
Character-level vanilla RNN in PyTorch that predicts a name's language of origin, letter by letter. Includes training, confusion matrix evaluation, and inference — plus a real example of dataset distribution shift when tested on modern names.
SuperMimic-mini: teacher-assisted G1 tracking + WM/BPTT negative study (RTX 2060)
Completing the 2x2 factorial: HRM's hierarchy with full BPTT reveals gradient-architecture interaction
Gradient checkpointing and VRAM auto-tuning for spiking neural network training. O(sqrt(T)) activation memory with bit-identical gradients.
GRU-based neural network equalizer in Verilog with FPGA deployment, RTL simulation, and on-chip BPTT training for adaptive communication channel equalization.
SNN training with Surrogate Gradients in pure PyTorch. 95.0% accuracy, 78% sparsity, 43.9x energy advantage on neuromorphic hardware. Timestep scaling + Pareto front analysis on MNIST.
Forward/backward pass of LSTM with peepholes and projections
IMDB sentiment analysis with a from-scratch RNN in low-level TensorFlow 2 (no Keras RNN layers). Padding/truncation, vocab limits, and BPTT training.
Community derivative of CuboiLeo/SuperMimic-Reproduced focused on low-VRAM training safeguards, reproducible engineering, and monitoring.
基于社区实现复现SuperMimic:世界模型与BPTT,RTX4090从零训练,LAFAN1训练库40条完整动作FP32评估与可复验记录。
To associate your repository with the bptt topic, visit your repo's landing page and select "manage topics."