Training Experiment · Python

ASTRO

Pop Alexandru@pop123-ux · Atlas 70.7 · #4= builder

View on GitHub Open Pop Alexandru's Atlas

What it is

ASTRO is a research project developing and evaluating a new optimizer (ASTRO) for language model training, featuring ablation studies, scaling experiments, and a reproducible benchmark harness.

Scored from the code at its last scan, 8 Oct 2026. How merit is scored

Where it counts

Machine Learning90%
NLP & LLMs85%
Data Science10%
MLOps10%

How much of the project is about each field, as scoring counts it. It ranks in the field it is most about.

Around it on the Machine Learning board

Whole board

RankProjectBuilderFieldMerit

26bf16-update-realization-in-transformersTraining Experiment · PythonThe repository investigates the behavior of AdamW optimizer in bfloat16 precision when fine-tuning a small language model (SmolLM2-135M) on GSM8K, by measuring the discrepancy between intended and realized weight updates through custom observer and analysis tools.55.8/100ImplementationRickMachine Learning
27bf16-4-3-overshoot-ratio-investigationTraining Experiment · PythonThe repository investigates the effect of different parameter writing rules (e.g., bfloat16 stochastic vs. round-to-nearest-even) on the overshoot ratio in optimizer steps by replaying recorded Camera-3 optimizer captures from Mamba-3 training. It reconstructs the intended…55.5/100ImplementationRickMachine Learning
28, shared rankchtbx-livive-netML Application · PythonThis repository provides a complete workflow for zero-shot voice cloning using the Chatterbox TTS model. It includes tools for recording a reference voice, synthesizing speech in the cloned voice, and generating web-optimized audio for a set of phrases.55.3/100ImplementationJarad PeckenpaughMachine Learning
28, shared rankASTROTraining Experiment · Python55.3/100ImplementationPop AlexandruMachine Learning
30ai_chat_distil_01ML Application · HTMLThis repository is a proof-of-concept system for reconstructing speech waveforms for a closed set of phrases using lightweight TensorFlow models. It includes data generation, model training, evaluation, conversion to TensorFlow Lite, and a web dashboard for interaction.54.3/100ImplementationJarad PeckenpaughMachine Learning

What it is

Where it counts

Around it on the Machine Learning board

More from Pop Alexandru