Embeddings

Create vectors for several values and compare them with cosine similarity.

embedMany preserves the input order, so each vector can be matched to its original text. The example compares related beach and ocean phrases with an unrelated compiler phrase and prints the provider’s token usage.

import std/[os, strformat]
import nimgent
import nimgent/providers/openai

let model: EmbeddingModel = openAI(getEnv("OPENAI_API_KEY")).embeddingModel(
  "text-embedding-3-small")

let values = @[
  "sunny day at the beach",
  "warm afternoon by the ocean",
  "debugging a compiler error"
]

echo "creating embeddings..."
let result: EmbedManyResult = embedMany(model, values)

for i in 1 ..< values.len:
  let similarity = cosineSimilarity(result.embeddings[0], result.embeddings[i])
  echo &"similarity between item 0 and item {i}: {similarity:.3f}"

echo "input tokens: ", result.usage.tokens

Run it from the repository root:

OPENAI_API_KEY=... nim c -r examples/embeddings.nim

View the source example