Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

published · difficulty 4/5 · 22 min read

Intuition

Build the mental picture first so the rest of the page has something to attach to.

PredictName the object in plain language, then predict what should change.Leave with one reusable mental picture before notation appears.

Long context feels like "the model remembers more." In practice it's two different problems:

Position extrapolation: the model must make sense of token positions it never saw during training (position is out-of-distribution).
Memory and bandwidth: even if the model could use a million tokens, you still have to store and read the KV cache efficiently.

That's why long-context work is a two-front war. RoPE scaling methods (YaRN, LongRoPE, position interpolation) attack the "angle OOD" issue, while KV techniques (GQA/MQA, paging, quantization, eviction) attack the memory wall.

If you want a single production mental model: long-context inference is usually memory-bound, not compute-bound.

Section prompt

Copy-only prompts — each action copies a page-grounded prompt to your clipboard. Nothing is sent by this site.

Math

Translate the story into symbols, assumptions, and a derivation you can inspect.

InspectTrack the same object through the notation and check each symbol.Leave with the invariant the equations preserve.

Equation 1Inspect equation objectM_{ij} = 0 if j <= i and i-j <= W ; -infinity otherwise Equation 2Inspect equation objectAttn(Q,K,V)=softmax ((QK^T)/(sqrt(d_k))+M)V.Equation 3Inspect equation objectMem_{KV} \approx B\cdot L\cdot T\cdot H_{kv}\cdot d_{head}\cdot 2 \cdot bytes.

Sliding window attention (reduce $T^2$ compute)

If each token attends only to the previous $W$ tokens, then attention compute is closer to $O(TW)$ instead of $O(T^2)$ . One way to express this is with a mask:

M_{ij} = \begin{cases} 0 & \text{if } j \le i \text{ and } i-j \le W \\\\ -\infty & \text{otherwise} \end{cases}

and:

\mathrm{Attn}(Q,K,V)=\mathrm{softmax}\!\left(\frac{QK^\top}{\sqrt{d_k}}+M\right)V.

KV cache memory (the bottleneck at long $T$ )

Across a batch of $B$ sequences and $L$ layers, storing keys and values for $T$ tokens costs roughly:

\mathrm{Mem}_{KV} \approx B\cdot L\cdot T\cdot H_{kv}\cdot d_{head}\cdot 2 \cdot \mathrm{bytes}.

This grows linearly in $T$ , and at long context it dominates the serving budget.

RoPE scaling (fix position OOD)

RoPE encodes position by rotating queries/keys. At positions far beyond training, those rotations can become out-of-distribution (especially for low-frequency dimensions). RoPE scaling methods effectively change the mapping $p \mapsto \theta(p)$ so the model sees "less extreme" phases at long context, while preserving short-range detail.

Section prompt

Copy-only prompts — each action copies a page-grounded prompt to your clipboard. Nothing is sent by this site.

Code

Keep the implementation aligned with the notation so the algorithm is legible.

TraceMatch variables to symbols before reading the implementation.Leave with a runnable witness for the math.

Code witness 1import numpy as np def kv_gb(T, L, Hkv, d_head, B=1, bytes_per_elem=2): elems = B * L * T * H...python

import numpy as np

def kv_gb(T, L, Hkv, d_head, B=1, bytes_per_elem=2):
    elems = B * L * T * Hkv * d_head * 2  # K and V
    return elems * bytes_per_elem / 1e9

L, Hkv, d_head, B = 80, 8, 128, 8  # example: 80 layers, GQA with 8 KV heads, fp16
for T in [2048, 8192, 32768, 131072]:
    print("T=", T, "KV~", round(kv_gb(T, L, Hkv, d_head, B), 2), "GB")

def attn_work(T, W, d):
    # crude: per token ~ 2 * W * d multiply-adds (QK + weights*V)
    return T * W * d * 2

T, d = 131072, 128
full = attn_work(T, T, d)
for W in [128, 512, 2048, T]:
    print("window=", W, "relative attention work:", round(attn_work(T, W, d) / full, 6))

Section prompt

Copy-only prompts — each action copies a page-grounded prompt to your clipboard. Nothing is sent by this site.

Interactive Demo

Use direct manipulation to connect the explanation to a moving system.

ManipulateChange one control and predict the visible response before reveal.Leave with the observed invariant or a repaired model.

Live Concept Demo

Explore Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

The stage is code-native and interactive. Use it to test the explanation against the mechanism.

difficulty 4/5graduatecode-aligned

Demo inquiry checkpoint

Manipulate one control and predict the visible change.

01Choose lensTrace a quantity

02ObserveDemo state pending

03GroundName the equation, invariant, or control that explains it.

04CarryNext: SSM Hybrids: Fixed-State Sequence Models for Long Context

Choose what to inspect in Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization. This shared fallback is an observation guide, not evidence of learning.

Loading interactive demo...

Use the demos to connect the "math knobs" to what serving systems actually feel:

Sliding window reduces attention compute (but can hurt long-range retrieval).
RoPE makes relative position usable at longer sequences.
KV cache dashboards make the memory wall concrete.

Section prompt

Copy-only prompts — each action copies a page-grounded prompt to your clipboard. Nothing is sent by this site.

Object

4/4 sections ready

Concept: Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?

BeforeScaled Dot-Product Attention & Transformer LayersNow4/4 sections readyTryManipulate one control and predict the visible change.NextSSM Hybrids: Fixed-State Sequence Models for Long Context

Object contextAttention & Transformers

StateAttention & Transformers

3 sources attachedChecking local snapshot

concept:attention-transformers/long-context

Math3 equation objectsM_{ij} = 0 if j <= i and i-j <= W ; -infinity otherwise CodeLong Context Engineering: RoPE Scaling, KV Compression & Memory Optim...elems = B * L * T * Hkv * d_head * 2 # K and V DemoLong Context Engineering: RoPE Scaling, KV Compression & Memory Optim...Manipulate one control and predict the visible change.

ConceptLearner lens

Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?

Mode questionCan I say the mechanism back in one sentence before I reveal anything?

Start with the prediction checkpoint, then compare the reveal to the mental model.

Take this move

Study modes

Keep the object fixed; change the lens.

01PredictCommit before reveal.02ExplainName the invariant.03Ground3 source links nearby.04ExperimentChange one knob.05CarrySave the next question.

Witness triad

KV memory is the same object in the equation, code, and demo.

The cache stores keys and values for every layer, token, KV head, and head dimension. With batch, layers, KV heads, head size, and bytes fixed, memory grows linearly with context length T.

MathEquation 3\mathrm{Mem}_{KV} \approx B\cdot L\cdot T\cdot H_{kv}\cdot d_{head}\cdot 2 \cdot \mathrm{bytes}.CodeCode witness 1elems = B * L * T * Hkv * d_head * 2 # K and VDemoDemo plannedKV memory estimate updates when context length, layers, heads, width, batch, or precision changes.

Bbatch sizeB

Lnumber of transformer layersL

Tcontext length in tokensT

H_kvkey/value heads after GQA or MQA sharingHkv

d_headwidth of each headd_head

2store both keys and values* 2

bytesbytes per cached elementbytes_per_elem

Predict before revealIf T doubles while B, L, H_kv, d_head, and bytes stay fixed, what should happen to KV memory?

Route back through the notebook

Carry the same object through intuition, math, code, and demo.

4/4 sections ready

Carry inScaled Dot-Product Attention & Transformer Layers

Bring the mental model from Scaled Dot-Product Attention & Transformer Layers; this page will reuse it instead of restarting from zero.

Work hereLong Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

Carry outSSM Hybrids: Fixed-State Sequence Models for Long Context

The next edge should feel earned: use the demo prediction here before following SSM Hybrids: Fixed-State Sequence Models for Long Context.

After The First Pass

Turn the concept into an inspected object.

The lower panels are one second act: keep the object fixed, inspect it visually, check source boundaries, practice transfer, then attach the research question.

ConceptLong Context Engineering: RoPE Scaling, KV Compression & Memory OptimizationAttention & Transformers

01MechanismInspect motionSee how the idea changes form before adding more explanation.02VisualChoose a cueBind one visible mark to one mathematical role.03SourcesCheck boundary3 source links attached to the object.04ClaimAudit evidenceTreat the page claim as something to verify.05PracticeAttempt transferWrite before feedback so the model has something to repair.06RoomAsk groundedAttach the question to the same exact object.

Mechanism Storyboard

See the idea move before the page explains it

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

Demo notes open01 / Intuition

Editorial long-context illustration of a stretched token scroll, rotary position arcs, and paged KV-cache blocks with selective long-range links.

Prediction lens

Start with the picture, metaphor, or geometric mechanism.

Commit first

Before reading further, choose the kind of change Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization should make visible.

Visual Inquiry

Make the image answer a mathematical question

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

4/4 stages readyDemo notes connected

Visual cueWhich visible object should carry the first intuition?

Inspection depth2/4

Prediction

Which visible object should carry the first intuition?

Commit first

Pick the cue that should make Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization easier to reason about before the page gives the answer.

Source Grounding

Canonical references for the mechanism on this page.

Object - ConceptLong Context Engineering: RoPE Scaling, KV Compression & Memory OptimizationQuestion

What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?

concept:attention-transformers/long-context

Boundary

sources: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention

Check

Open the closest source note before trusting the local explanation.

Evidence

3 selected-object sources shown first; 3 references total.

Next move

Audit the claim boundary, then ask from the same selected object.

selected object source · paper · 2021RoFormer: Enhanced Transformer with Rotary Position EmbeddingSu et al.

Located CF editorial boundary

Direct PDF. Grounds RoPE as a rotary position mechanism whose query-key product can express relative offsets; not a RoPE-scaling recipe.

Used here as

RoFormer grounds the rotary relative-position mechanism; ALiBi directly frames train-short/test-long length extrapolation as a position-representation problem; PagedAttention directly fra...

Caveat

Checks two long-context constraints, not every bottleneck; RoFormer is not a RoPE-scaling recipe, ALiBi is not RoPE, and local code/demo are toy witnesses, not production guarantees.

Open source

selected object source · paper · 2021Train Short, Test Long: Attention with Linear Biases Enables Input Length ExtrapolationPress, Smith, and Lewis

Located CF editorial boundary

Direct PDF. Grounds input-length extrapolation as a position-representation problem via ALiBi; this is a contrasting route, not a RoPE variant.

Used here as

RoFormer grounds the rotary relative-position mechanism; ALiBi directly frames train-short/test-long length extrapolation as a position-representation problem; PagedAttention directly fra...

Caveat

Checks two long-context constraints, not every bottleneck; RoFormer is not a RoPE-scaling recipe, ALiBi is not RoPE, and local code/demo are toy witnesses, not production guarantees.

Open source

selected object source · paper · 2023Efficient Memory Management for Large Language Model Serving with PagedAttentionKwon et al.

Located CF editorial boundary

Direct PDF. Grounds long-context serving feasibility as a KV-cache memory-management problem via paged KV-cache allocation.

Used here as

RoFormer grounds the rotary relative-position mechanism; ALiBi directly frames train-short/test-long length extrapolation as a position-representation problem; PagedAttention directly fra...

Caveat

Checks two long-context constraints, not every bottleneck; RoFormer is not a RoPE-scaling recipe, ALiBi is not RoPE, and local code/demo are toy witnesses, not production guarantees.

Open source

Claim Review

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

Object - ConceptLong Context Engineering: RoPE Scaling, KV Compression & Memory OptimizationQuestion

What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?

concept:attention-transformers/long-context

Boundary

sources: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention

Check

Treat every claim as provisional until source support and a local witness agree.

Evidence

1 structured claim check on this concept.

Next move

Run the prediction or practice transfer before asking for a grounded review.

1 CF editorial source-scope review recorded

Publisher-side editorial review is not independent replication. Claims without it still need exact source-support review. 3 references and 3 local witnesses are available for inspection.

Long-context engineering has two separate constraints: positional extrapolation must keep attention usable beyond training lengths, and KV-cache memory management must keep long-context serving feasible.

Registered CF editorial boundary

RoFormer: Enhanced Transformer with Rotary Position EmbeddingDirect PDF. Grounds RoPE as a rotary position mechanism whose query-key product can express relative offsets; not a RoPE-scaling recipe.Train Short, Test Long: Attention with Linear Biases Enables Input Length ExtrapolationDirect PDF. Grounds input-length extrapolation as a position-representation problem via ALiBi; this is a contrasting route, not a RoPE variant.Efficient Memory Management for Large Language Model Serving with PagedAttentionDirect PDF. Grounds long-context serving feasibility as a KV-cache memory-management problem via paged KV-cache allocation.

Used here as

RoFormer grounds the rotary relative-position mechanism; ALiBi directly frames train-short/test-long length extrapolation as a position-representation problem; PagedAttention directly frames high-throughput...

Local witness

source-span-su-2021-roformer source-span-press-2021-alibi source-span-kwon-2023-pagedattention code-witness-1 interactive-demo

Caveat

Checks two long-context constraints, not every bottleneck; RoFormer is not a RoPE-scaling recipe, ALiBi is not RoPE, and local code/demo are toy witnesses, not production guarantees.

Review stateCF editorial source-scope reviewClaim metadata: source checkedPublisher-side editorial review only; not independent replication. Check caveats and exact source scope.

RoFormer supports rotary relative-position mechanics; ALiBi supports train-short/test-long extrapolation as a position-representation issue; PagedAttention supports KV-cache memory as a serving bottleneck. Local code/demo are toy witnesses for KV sizing and phase constraints, not serving-system proof.

Reviewer: codex+oracle; reviewed 2026-05-07

Practice notebook

Use the idea, then test it somewhere new

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

AttemptNo learning claim inferred

Object - ConceptLong Context Engineering: RoPE Scaling, KV Compression & Memory OptimizationQuestion

What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?

concept:attention-transformers/long-context

Boundary

sources: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention

Check

Use one state from Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization to explain what changes, why it changes, and which assumption the explanation needs.

Evidence

No learner move yet; no learning state is inferred.

Next move

Write first, use only the help you need, then try a new case without it.

Explain

Use one state from Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization to explain what changes, why it changes, and which assumption the explanation needs.

Hint 1

Reveal when your model needs a nudge.

Hint 2

Reveal when your model needs a nudge.

Hint 3

Reveal when your model needs a nudge.

Your answer canvas

Attempt context

Learner move

Write an attempt before asking the companion.

Assistance used

0 of 3 progressive hints opened.

Evidence boundary

This draft and any AI response do not establish mastery; a later unassisted case can.

Misconception check

Grounded object roomClose

Selected object routeAsk from this object; carry one invariant back.sources: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention

ObjectConceptLong Context Engineering: RoPE Scaling, KV Compression & Memory Optim...
PredictBefore revealLong Context Engineering: RoPE Scaling, KV Compression & Memory Optim...
WitnessCompare codeLong Context Engineering: RoPE Scaling, KV Compression & Memory Optim...
RoomAsk groundedChecking local snapshot

ConceptLong Context Engineering: RoPE Scaling, KV Compression & Memory OptimizationAttention & Transformers

Code witness comparisonLong Context Engineering: RoPE Scaling, KV Compression & Memory Optimization code witne...elems = B * L * T * Hkv * d_head * 2 # K and V Prediction before revealLong Context Engineering: RoPE Scaling, KV Compression & Memory Optimization predictionManipulate one control and predict the visible change.

Grounded room questionWhat is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?Checking local snapshot

Research Room

Attach the question to an exact object

Pick the concept, equation, source, code witness, claim, misconception, or demo state before asking for help. The handoff stays grounded to that object.

Next local actionNo local draft saved yet

Open the draft below to save one note and next action in this browser.

conceptAttention & Transformers

Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

Anchored question

What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math?

Source boundaryInspect source ids: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattentionStable content-object key attached

Role lenses for this object

These are fixed, deterministic perspectives derived from the selected object. They do not represent people, community contributions, or independent review.

Learner evidence requestAsk what would make "Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization" feel predictable rather than familiar.

Assumption

Source ids su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention must support the exact object, not just the surrounding topic.

Source-checking summary

Treat this as a mechanism object: connect the definition to one equation, code witness, or demo before broadening the discussion.

Proposed experiment

Ask the learner to perturb one representation, then check whether the same invariant survives in math, code, and demo.

Next action

The learner can state the mechanism in their own words

Evidence4 checks

PredictionChecking carried observation

ActionReady for one action

AILearner handoff ready

Open source object

01PredictionChecking browser-local route memory

02EvidenceChecking for a carried observation

03BoundaryInspect source ids: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention

04Next moveSave one next action

Local action draftNo local draft saved yetExpand only when ready to capture one local next action

Local action draft

This draft stays locally in this browser for concept:attention-transformers/long-context.

Draft noteNext action

No local draft saved.

Evidence to inspect

Source ids to inspect: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention
Definition, prerequisite, and contrast concept links
The equation or code witness that makes the concept operational
One demo state that shows the invariant instead of a slogan

What would resolve this

The learner can state the mechanism in their own words
The learner can name the prerequisite that would repair confusion
The learner can predict how the mechanism changes under one perturbation

Object-attached AI handoff

I am working in Continuous Function's research reading room. Object: concept - Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization Object key: concept:attention-transformers/long-context Context: Attention & Transformers Anchor id: concept/concept-notebook/attention-transformers/long-context Open question: What is the smallest example that makes Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization click without losing the math? Evidence to inspect: - Source ids to inspect: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention - Definition, prerequisite, and contrast concept links - The equation or code witness that makes the concept operational - One demo state that shows the invariant instead of a slogan Deterministic role lenses for this object: - Boundary: fixed perspectives, not people, community contributions, or independent review - Source-checking summary: Treat this as a mechanism object: connect the definition to one equation, code witness, or demo before broadening the discussion. - Proposed experiment: Ask the learner to perturb one representation, then check whether the same invariant survives in math, code, and demo. - Teach/transfer move: Turn the mechanism into one sentence that predicts a neighboring concept. - Assumptions: - Source ids su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention must support the exact object, not just the surrounding topic. - The stable content-object key lets local drafts, prompts, and route memory attach without changing the source page. - The concept explanation is local atlas prose until checked against its math, code, and source support. - Prerequisite gaps should become a repair route, not a reason to leave the object vague. - Role-lens requests: - Learner: ask for "Ask what would make "Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization" feel predictable rather than familiar." | assumption: Source ids su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention must support the exact object, not just the surrounding topic. | next action: The learner can state the mechanism in their own words - Researcher: ask for "Source ids to inspect: su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention" | assumption: The stable content-object key lets local drafts, prompts, and route memory attach without changing the source page. | next action: The learner can name the prerequisite that would repair confusion - Experimenter: ask for "Choose one variable or condition to perturb before asking for an explanation." | assumption: The concept explanation is local atlas prose until checked against its math, code, and source support. | next action: The learner can predict how the mechanism changes under one perturbation - Professor: ask for "Find the smallest transferable rule a learner could reuse without the AI." | assumption: Prerequisite gaps should become a repair route, not a reason to leave the object vague. | next action: Teach or transfer: Turn the mechanism into one sentence that predicts a neighboring concept. What would resolve this: - The learner can state the mechanism in their own words - The learner can name the prerequisite that would repair confusion - The learner can predict how the mechanism changes under one perturbation Answer as a careful research tutor: stay source-grounded, separate verified evidence from assumptions, name the relevant math objects, and end with one next action. Current deterministic role lens for this object: - Role lens: Learner - Evidence request: Ask what would make "Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization" feel predictable rather than familiar. - Assumption to keep visible: Source ids su-2021-roformer, press-2021-alibi, kwon-2023-pagedattention must support the exact object, not just the surrounding topic. - Proposed experiment: Ask the learner to perturb one representation, then check whether the same invariant survives in math, code, and demo. - Next action: The learner can state the mechanism in their own words

concept/concept-notebook/attention-transformers/long-context
concept:attention-transformers/long-context

Learning Map

Before / Now / Try / Next

BeforeScaled Dot-Product Attention & Transformer Layers

NowIntuition → Math → Code → Demo

TryManipulate one control and predict the visible change.

NextSSM Hybrids: Fixed-State Sequence Models for Long Context

Learnerpicture first Researchersource check Experimentertest the demo Professorname invariant

01Predict 02Ground 03Code 04Carry

01Intuitionready
02Mathready
03Codeready
04Interactive Demoready

Object Companion

Ask beside the selected object

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

Your question

GoalComfortStyleStuck on

Context prompt

You are my AI learning companion for Continuous Function. Current context: Attention & Transformers concept. Learning surface: Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization. What this page says: How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression). Current section: Intuition, math, code, and interactive demo. Suggested next step: Manipulate one control and predict the visible change.. Learner goal: Understand the idea. Learner comfort level: New to this. Preferred explanation style: Visual first. Task: Explain the central idea in plain language, then restate it with the exact math objects from the page. Answer in a way that helps me learn: ask one clarifying question only if needed, use intuition before notation, and end with one thing I should try on the page.

Copy-only tool: this panel composes a prompt for an AI tool you already use. Nothing you enter here is sent from this page.

Domain

Attention & Transformers

attention-transformerslong-contextkv-cacheropememory

Prerequisites

Scaled Dot-Product Attention & Transformer Layers Rotary Position Embeddings (RoPE)Efficient Attention at Scale: KV Cache, GQA & FlashAttention LLM Serving at Scale: Prefill, Decode & Continuous Batching

Leads To

SSM Hybrids: Fixed-State Sequence Models for Long Context Speculative Decoding: Lossless Multi-Token Generation

Efficiency: Quantization, Distillation, LoRA & Sparse MoE Tokenization & Vocabulary Design

Within this domain

Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

Sliding window attention (reduce T2T^2T2 compute)

KV cache memory (the bottleneck at long TTT)

RoPE scaling (fix position OOD)

Concept: Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

KV memory is the same object in the equation, code, and demo.

Carry the same object through intuition, math, code, and demo.

Turn the concept into an inspected object.

Start with the picture, metaphor, or geometric mechanism.

Which visible object should carry the first intuition?

Canonical references for the mechanism on this page.

How frontier LLMs stretch context windows: positional extrapolation (RoPE scaling) plus KV cache memory tricks (GQA, paging, quantization, compression).

Attach the question to an exact object

Long Context Engineering: RoPE Scaling, KV Compression & Memory Optimization

Sliding window attention (reduce $T^2$ compute)

KV cache memory (the bottleneck at long $T$ )