---
type: "post"
id: "baseten-published-a-technical-explainer-on-selecting-layer-b2914a"
slug: "baseten-published-a-technical-explainer-on-selecting-layer-b2914a"
title: "Baseten published a technical explainer on selecting layer precision for NVFP4 quantization"
description: "Thesis: High-quality NVFP4 quantization depends on choosing precision per layer, and estimating those choices requires accounting for interactions among quantized layers."
retrieval_nugget: "Thesis: High-quality NVFP4 quantization depends on choosing precision per layer, and estimating those choices requires accounting for interactions among quantized layers."
published_at: "2026-10-09"
updated_at: "2026-10-09"
record_date: "2026-10-09"
date_kind: "published_at"
topics: []
source_urls: ["https://www.baseten.co/blog/what-makes-a-high-quality-nvfp4-quantization"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:baseten-published-a-technical-explainer-on-selecting-layer-b2914a"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:d25f6410eff5d9ec7a715a9e59b60d2f6f4549a3d2dde6f94c65c9a1fb747305","reviewed_at":"2026-10-09T20:05:06.596Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":2,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"High-quality NVFP4 quantization depends on choosing precision per layer, and estimating those choices requires accounting for interactions among quantized layers.","observed_facts":[{"text":"The post says NVFP4 divides tensors into blocks of 16 values, each with its own scale.","source_urls":["https://www.baseten.co/blog/what-makes-a-high-quality-nvfp4-quantization"]},{"text":"The post says activation scales require calibration data because activations depend on inputs, while weights are fixed.","source_urls":["https://www.baseten.co/blog/what-makes-a-high-quality-nvfp4-quantization"]},{"text":"The post describes SaturationQuant using Aumann–Shapley scoring to account for interactions among quantized layers.","source_urls":["https://www.baseten.co/blog/what-makes-a-high-quality-nvfp4-quantization"]}],"mechanism":"The post contrasts architecture-based heuristics with sensitivity scoring. Isolated-layer scores measure the effect of quantizing one layer while other layers remain at baseline; Baseten argues those scores cannot simply be added for a mixed-precision recipe because loss can saturate as more layers are quantized. SaturationQuant uses Aumann–Shapley scoring to build this effect into scores that can be summed.","why_now":"NVFP4 offers smaller weights and potentially faster computation on hardware with native FP4 support, so preserving quality while assigning precision has direct deployment relevance.","implications":["Quantization choices should be judged as a whole recipe, not by ranking layers in isolation.","Calibration data and the target workload matter when assigning precision."],"evidence_boundary":"The source explains the method and its rationale, but the supplied material does not provide independent comparative validation.","watch_conditions":["Independent evaluations across models and representative calibration datasets would show whether the described scoring approach predicts quality reliably in practice."],"related_records":[],"new_branch_reason":"No prior New Runtime record is supplied; this technical explainer therefore starts a distinct record on NVFP4 layer selection and saturation-aware quantization scoring."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/baseten-published-a-technical-explainer-on-selecting-layer-b2914a/","markdown":"https://newruntime.com/posts/baseten-published-a-technical-explainer-on-selecting-layer-b2914a.md","json":"https://newruntime.com/posts/baseten-published-a-technical-explainer-on-selecting-layer-b2914a.json"}
---

# Baseten published a technical explainer on selecting layer precision for NVFP4 quantization

## Retrieval answer

Thesis: High-quality NVFP4 quantization depends on choosing precision per layer, and estimating those choices requires accounting for interactions among quantized layers.

Thesis: High-quality NVFP4 quantization depends on choosing precision per layer, and estimating those choices requires accounting for interactions among quantized layers.

Baseten's post states that NVFP4 represents values in 4 bits and divides tensors into blocks of 16 values, each with its own scale. It also explains that activation scales require calibration data because activations depend on model inputs, while weights are fixed. These details make the quantization recipe a measurement and allocation problem, not merely a uniform format conversion.

Mechanism: The post contrasts architecture-based heuristics with sensitivity scoring. Isolated-layer scores measure the effect of quantizing one layer while other layers remain at baseline; Baseten argues those scores cannot simply be added for a mixed-precision recipe because loss can saturate as more layers are quantized. SaturationQuant uses Aumann–Shapley scoring to build this effect into scores that can be summed. Why now: NVFP4 offers smaller weights and potentially faster computation on hardware with native FP4 support, so preserving quality while assigning precision has direct deployment relevance.

Quantization choices should be judged as a whole recipe, not by ranking layers in isolation. This is a vendor-authored technical explanation; it does not establish independent benchmark results or prove that SaturationQuant is best across models and workloads. Evidence boundary: The source explains the method and its rationale, but the supplied material does not provide independent comparative validation. Watch condition: Independent evaluations across models and representative calibration datasets would show whether the described scoring approach predicts quality reliably in practice.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)