blob: 289e55fc5c88bb93d94b3537a60446e6a504c9ff (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
|
package main
import (
"fmt"
"log"
"go.jknobloc.com/x/llmc"
"go.jknobloc.com/x/tokenizer/bpe"
)
func main() {
var data llmc.DataFile[uint16]
var gold llmc.DataFile[uint16]
_ = must(llmc.Deserialize("artifacts/data/llmc/tinyshakespeare/tiny_shakespeare_train.bin", &data))
_ = must(llmc.Deserialize("artifacts/test/llmc/tinyshakespeare/tiny_shakespeare_train.bin", &gold))
t := must(bpe.NewTokenizerFromFiles("gpt2/models/base/vocab.json", "gpt2/models/base/merges.txt"))
itoa := bpe.Itoa(t)
fmt.Println(len(data.Tokens))
fmt.Println(len(gold.Tokens))
if len(data.Tokens) != len(gold.Tokens) {
log.Fatal("length mismatch")
}
for i, a := range data.Tokens {
b := gold.Tokens[i]
if a != b {
fmt.Printf("token mismatch: %d %d at index %d\n", a, b, i)
fmt.Println("data", itoa[int64(a)])
fmt.Println("gold", itoa[int64(b)])
break
}
}
}
func must[T any](v T, err error) T {
if err != nil {
log.Fatal(err)
}
return v
}
|
