Preserve

Synthetic Data Generation,
20-30x Faster

Stop generating tokens. Start permuting sentences.
160ms vs 3-5 seconds

Learn More
25x
Cheaper
30x
Faster
O(n)
Complexity

Old vs New: Synthetic Data Generation

Why attention-based permutation beats token generation

Old Method

LLM-Driven Generation

Document → LLM
→ [Generate token 1] ⏱️ 100ms
→ [Generate token 2] ⏱️ 100ms
...
→ [Generate token N] ⏱️ 100ms
= 3-5 seconds
Slow: Sequential token generation
Expensive: $50K+ for 10B tokens
Not scalable: Full generation per example

New Method

Attention-Based Permutation

Document → Extract attention
⏱️ 50ms
→ Build DAG
⏱️ 10ms
→ Permute sentences
⏱️ 30ms
→ Validate permutations
⏱️ 40ms
= 160ms total
Fast: No token generation needed
Cheap: Only prefill, 25x cost reduction
Scalable: Multiple permutations instantly

Key Insight

Instead of generating new text token-by-token, we reuse existing sentences in valid orders based on their semantic dependencies extracted from attention patterns.

Technical Deep Dive

How we achieve O(n) speedup with adaptive constraint relaxation

Pattern Extraction

We run a single forward pass through the model with output_attentions=True, then filter to only "structural" attention heads that have >30% cross-sentence attention.

These structural heads reveal which sentences semantically depend on which others. For example: "It does X" attends strongly to the sentence defining "it".

Input
Document text
Output
Dependency graph

The Unique Solution

We convert dependencies into a DAG (Directed Acyclic Graph) where sentences are nodes and edges represent semantic dependencies. Then we generate permutations via topological sorts of this DAG.

The Challenge: "Almost Linear" DAGs

When DAGs are too restrictive (avg <2.5 dependencies/node), we get limited permutation diversity.

Our Innovation: Adaptive Constraint Relaxation

We dynamically detect restrictive DAGs and apply strategies:

  • Relax position constraints
  • Allow "sibling" swaps (sentences sharing dependencies)
  • Use fallback strategies for diversity

Result: O(seqlen) speedup while preventing semantic breakage and ensuring permutation diversity

🎯

Semantic Preservation

Dependencies maintained

Lightning Fast

No token generation

🎨

High Diversity

Multiple valid permutations

Join the Waitlist

Be among the first to access Preserve and revolutionize your synthetic data generation