55 lines
1.2 KiB
Go
55 lines
1.2 KiB
Go
package embedding
|
|
|
|
import (
|
|
"context"
|
|
"crypto/sha256"
|
|
"encoding/binary"
|
|
"math"
|
|
)
|
|
|
|
type FakeProvider struct {
|
|
dims int
|
|
model string
|
|
}
|
|
|
|
func NewFakeProvider(dims int, model string) *FakeProvider {
|
|
if dims <= 0 {
|
|
dims = 1536
|
|
}
|
|
if model == "" {
|
|
model = "fake"
|
|
}
|
|
return &FakeProvider{dims: dims, model: model}
|
|
}
|
|
func (p *FakeProvider) Dimensions() int { return p.dims }
|
|
func (p *FakeProvider) Model() string { return p.model }
|
|
func (p *FakeProvider) ProviderName() string { return "fake" }
|
|
func (p *FakeProvider) Embed(ctx context.Context, input []string) ([]Vector, error) {
|
|
out := make([]Vector, 0, len(input))
|
|
for _, text := range input {
|
|
select {
|
|
case <-ctx.Done():
|
|
return nil, ctx.Err()
|
|
default:
|
|
}
|
|
v := make(Vector, p.dims)
|
|
seed := sha256.Sum256([]byte(text))
|
|
var norm float64
|
|
for i := 0; i < p.dims; i++ {
|
|
b := sha256.Sum256(append(seed[:], byte(i), byte(i>>8)))
|
|
n := binary.BigEndian.Uint32(b[:4])
|
|
val := (float64(n%2000000)/1000000.0 - 1.0)
|
|
v[i] = float32(val)
|
|
norm += val * val
|
|
}
|
|
if norm > 0 {
|
|
scale := float32(1 / math.Sqrt(norm))
|
|
for i := range v {
|
|
v[i] *= scale
|
|
}
|
|
}
|
|
out = append(out, v)
|
|
}
|
|
return out, nil
|
|
}
|