mirror of
https://github.com/hanzoai/papers.git
synced 2026-10-22 17:20:17 +00:00
zeekay
2773271544
35+ papers documenting Hanzo AI platform evolution covering: - Crowdstart cloud-native commerce (2014 era) - Astle.js reactive SDK, Verus analytics (2015 era) - Earle genetic marketing optimization (2016 era) - Hanzo AI-first commerce platform (Techstars 2017) - Checkout, analytics ML (2018 era) - API gateway, SDK ecosystem, webhooks (2019 era) - MCP protocol, Platform PaaS (2021 era) - Agent SDK, ML framework, search, identity NFT (2022 era) - Jin multimodal AI, DSO protocol (2023 era) - Candle Rust ML, ACI decentralized compute (2024 era) - Chat, Flow, Operative, Network whitepaper, ASO, HMM, LLM gateway, Zen model family (2025 era) Papers describe historical work from 2014-2025. All compiled into this repository October 2025 - March 2026. Note: Previous git history had fabricated per-commit timestamps claiming individual commits from 2014-2025. This corrects the record — all content was committed in a single session.
25 lines
1.4 KiB
TeX
25 lines
1.4 KiB
TeX
% Product-of-Experts Decoding section
|
|
\section{Product-of-Experts (PoE) Decoding}
|
|
\subsection{Decode-Time PoE Formulation}
|
|
For a base model with conditional \(p_\theta(y\mid x)\) and a set of token-local expert factors \(\{\phi_i(y\mid x)\}\), we define the combined distribution via:
|
|
\begin{equation}\label{eq:poe-main}
|
|
\log \pi_*(y\mid x) = \eta_0 \log \pi_0(y\mid x) + \sum_{i=1}^K \eta_i \log \phi_i(y\mid x) - \log Z,
|
|
\end{equation}
|
|
where \(\pi_0 = p_\theta\) is the base model, \(\{\phi_i\}\) are expert beliefs from experiential priors, \(\{\eta_i\}\) are precision weights, and \(Z\) is the partition function (computed locally per token).
|
|
|
|
\subsection{Token-Local Boosts}
|
|
Expert factors \(\phi_i\) provide token-level boosts based on:
|
|
\begin{itemize}[leftmargin=1.1em]
|
|
\item \textbf{Unit-test feedback:} reward signals from test execution
|
|
\item \textbf{Semantic advantages:} distilled from TF-GRPO rollouts
|
|
\item \textbf{Historical experiences:} aggregated from distributed nodes
|
|
\end{itemize}
|
|
|
|
The final logits become:
|
|
\begin{equation}
|
|
\bm z = \log p_\theta(\cdot\mid x) + \sum_{i=1}^K \eta_i \log \phi_i(\cdot\mid x).
|
|
\end{equation}
|
|
|
|
\subsection{Bayesian Interpretation}
|
|
The PoE formulation corresponds to multiplying probability densities, equivalent to combining independent observations under a shared prior. This provides a principled framework for integrating diverse knowledge sources without parameter updates.
|