---
title: "When decomposing the standard MTP gradient (trained with standard Cross-Entropy), two opposing… — post author — DecisionGateway"
description: "statement by post author, recorded 2026-11-05 — verbatim quote and source citation."
canonical_url: "https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423"
meta:
  "og:description": "statement by post author, recorded 2026-11-05 — verbatim quote and source citation."
  "og:image": "https://dgw.tmarket.io/og.jpg"
  "og:image:alt": "White network of connected nodes crossed by a timeline on black"
  "og:image:height": "1024"
  "og:image:width": "1024"
  "og:title": "When decomposing the standard MTP gradient (trained with standard Cross-Entropy), two opposing… — post author — DecisionGateway"
  "twitter:image": "https://dgw.tmarket.io/og.jpg"
---

> Index of every page on this site: [llms.txt](https://dgw.tmarket.io/llms.txt)

statementSaid on 2026-11-05

# **When decomposing the standard MTP gradient (trained with standard Cross-Entropy), two opposing forces are hidden inside: a useful self-distillation signal and a residual noise component.**

A statement by post author, said on 2026-11-05. Statements are as reported by the cited publishers — not verified by DecisionGateway.

> *כשפירקו את נוסחת הגראדיאנט הרגיל של ה-MTP (שמאומן עם פונקציית Cross-Entropy או CE סטנדרטית), הם גילן שמתחבאים בתוכו שני כוחות מנוגדים:*

mathyaiwithmikePublished 2026-11-05 [ Read source](https://t.me/mathyaiwithmike/1302)

## **Discussion**

## **Related claims**

[**If you train models with MTP architecture, stop injecting noisy CE gradients and instead add the Soft-MTP line of code.**מייק · commitment](https://dgw.tmarket.io/claims/01a1e1ee-4898-7ced-9f43-9f192a44b50c) [**Soft-MTP obtains the same gradient for free by a simple subtraction between gradients already computed, instead of computing KL from scratch (which requires a lot of memory).**post author · statement](https://dgw.tmarket.io/claims/01a1e1ee-ca55-7120-bcef-603814df663d) [**During the backward pass, the MTP residual noise and the NTP residual noise meet with opposite signs and cancel each other out, leaving only the clean self-distillation signal; the result is 0 added memory, 0 impact on token throughput, and full compatibility with Fused kernels.**post author · statement](https://dgw.tmarket.io/claims/01a1e1ee-ca92-7c91-9863-1727b4a888ac) [**Large AI labs train models with MTP, but 80% of the gradient they backpropagate to the model is static noise that interferes with learning.**post author · allegation](https://dgw.tmarket.io/claims/01a1e1ee-c8cc-76f8-a0cf-fdc0d5c6fade) [**Adding the Soft-MTP line of code isolates the Self-Distillation signal and yields a smarter, faster model in speculative decoding without paying an additional penny in compute or memory.**מייק · statement](https://dgw.tmarket.io/claims/01a1e1ee-48b9-76c9-ba93-6ab9598c88a7) [**When the same gradient keeps appearing across several checkpoints, the writers interpret it as "demand" that SGD fails to supply.**the writers · statement](https://dgw.tmarket.io/claims/01a1e1f6-8f5d-7c1b-85c7-28a8997d5012)

[All claims](https://dgw.tmarket.io/claims)

```json
{"@context":"https://schema.org","@graph":[{"@id":"https://dgw.tmarket.io/#website","@type":"WebSite","description":"Decisions, events and falsifiable forecasts from 4,500+ Telegram channels — every claim cited. Read free.","name":"DecisionGateway","url":"https://dgw.tmarket.io/","publisher":{"@id":"https://dgw.tmarket.io/#identity"}},{"@id":"https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423#webpage","@type":"WebPage","description":"statement by post author, recorded 2026-11-05 — verbatim quote and source citation.","name":"When decomposing the standard MTP gradient (trained with standard Cross-Entropy), two opposing… — post author — DecisionGateway","url":"https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423","breadcrumb":{"@id":"https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423#breadcrumb"},"isPartOf":{"@id":"https://dgw.tmarket.io/#website"},"potentialAction":[{"@type":"ReadAction","target":["https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423"]}]},{"@id":"https://dgw.tmarket.io/#identity","@type":"Organization","name":"DecisionGateway","url":"https://dgw.tmarket.io"},{"@id":"https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423#breadcrumb","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","name":"Home","item":"https://dgw.tmarket.io","position":1},{"@type":"ListItem","name":"Claims","item":"https://dgw.tmarket.io/claims","position":2},{"@type":"ListItem","name":"When decomposing the standard MTP gradient (trained with…","position":3}]},{"@id":"https://dgw.tmarket.io/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423#/schema/quotation/1","@type":"Quotation","dateCreated":"2026-11-05","isBasedOn":"https://t.me/mathyaiwithmike/1302","text":"כשפירקו את נוסחת הגראדיאנט הרגיל של ה-MTP (שמאומן עם פונקציית Cross-Entropy או CE סטנדרטית), הם גילן שמתחבאים בתוכו שני כוחות מנוגדים:","url":"/claims/01a1e1ee-c9b5-7ec8-958e-20cbda9fe423"}]}
```

## Sitemap

See the full [sitemap](https://dgw.tmarket.io/sitemap.md) for all pages.
