Skip to content
sinc-LLMPreview this diagnostic

Engineering instrument

Embedding Distance Visualizer

Compare two prompts as vectors in semantic embedding space. Returns cosine similarity, semantic-shift analysis, shared vs unique concepts, intent match, and a likelihood prediction of whether both prompts produce the same response.

Preview this diagnosticCompare services
// Tool 7/8 · Linear Algebra

Compare two prompts as vectors in semantic embedding space. Returns cosine similarity, semantic-shift analysis, shared vs unique concepts, intent match, and a likelihood prediction of whether both prompts produce the same response.

// Grounded in transformer-internals.md (Engineering Analysis 34 + linear-algebra concept 31 sources)
// awaiting input
tools-api v1.0
Nemotron 120B → Gemma 31B → MiniMax → Liquid → router

// Input

// Output

// Raw JSON Response

// Want this scaled?

Transformer Internals Audit (Service #40)

Read what your model is actually doing. Attention-head analysis, embedding-space probing, SVD-based LoRA design. When prompt-tuning has plateaued, this is the layer beneath.

→ See service · $20K – $60K + $1K – $3K/mo
Original source record

sinc-LLM About Blog Services

// MARIO ALEXANDRE — DLux Digital · Free tool, visible engineering, OpenRouter free-tier models. Audit feed: /api/tools/audit · Metrics: /api/tools/metrics