gguf_test.go 4.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187
  1. package llm
  2. import (
  3. "crypto/sha256"
  4. "fmt"
  5. "io"
  6. "math"
  7. "os"
  8. "path/filepath"
  9. "testing"
  10. "github.com/google/go-cmp/cmp"
  11. )
  12. // TestGGUFDecode tests the decoding and rewriting of (unsorted) GGUF files
  13. // To run, add GGUF files to /llm/testdata and add the name of the file to the tests slice
  14. // Should comment //sort.Sort(tensors) in gguf.go
  15. // This creates a temporary file in /llm/testdata that will deleted only if the test passes
  16. func TestGGUFRewrite(t *testing.T) {
  17. tests := []string{
  18. "nutiny.gguf",
  19. }
  20. for i := range tests {
  21. tt := tests[i]
  22. t.Run(tt, func(t *testing.T) {
  23. t.Parallel()
  24. p := filepath.Join("testdata", tt)
  25. if _, err := os.Stat(p); err != nil {
  26. t.Skip("file not found", p)
  27. }
  28. wantFile, err := os.Open(p)
  29. if err != nil {
  30. t.Fatal(err)
  31. }
  32. defer wantFile.Close()
  33. // decode original gguf
  34. _, wantGGML, err := decodeGGML(t, wantFile)
  35. if err != nil {
  36. t.Fatal(err)
  37. }
  38. gotFile, err := os.CreateTemp("testdata", tt)
  39. if err != nil {
  40. t.Fatal(err)
  41. }
  42. defer func() {
  43. gotFile.Close()
  44. if !t.Failed() {
  45. os.Remove(gotFile.Name())
  46. }
  47. }()
  48. _, gotGGML, err := rewriteGGML(t, wantGGML, gotFile, wantFile)
  49. if err != nil {
  50. t.Fatal(err)
  51. }
  52. diff, diff2 := compareGGML(t, gotGGML, wantGGML, gotFile, wantFile)
  53. if cmp.Diff(diff, diff2) != "" {
  54. t.Fatalf("diff: \n%s", cmp.Diff(diff, diff2))
  55. }
  56. })
  57. }
  58. }
  59. func compareGGML(t *testing.T, gotGGML, wantGGML *GGML, f *os.File, f2 *os.File) (map[string]string, map[string]string) {
  60. got := make(map[string]string)
  61. want := make(map[string]string)
  62. gotKV := gotGGML.KV()
  63. wantKV := wantGGML.KV()
  64. if len(gotKV) != len(wantKV) {
  65. t.Fatalf("got length: %d != want length: %d", len(gotKV), len(wantKV))
  66. }
  67. for k, v := range gotKV {
  68. switch t := v.(type) {
  69. case *array:
  70. if diffy := cmp.Diff(t.values, wantKV[k].(*array).values); diffy != "" {
  71. got[k] = diffy
  72. }
  73. default:
  74. if v != wantKV[k] {
  75. got[k] = fmt.Sprintf("kv1: %v, kv2: %v", v, want[k])
  76. }
  77. }
  78. }
  79. gotTensors := gotGGML.Tensors().Items
  80. gotOffset := gotGGML.Tensors().Offset
  81. wantTensors := wantGGML.Tensors().Items
  82. wantOffset := wantGGML.Tensors().Offset
  83. if len(gotTensors) != len(wantTensors) {
  84. got["lenTensors"] = fmt.Sprintf("t1: %d, t2: %d", len(gotTensors), len(wantTensors))
  85. }
  86. for _, tensor := range gotTensors {
  87. sha256sum := sha256.New()
  88. sr := io.NewSectionReader(f, gotOffset+int64(tensor.Offset), int64(tensor.Size()))
  89. var s int64
  90. s, err := io.Copy(sha256sum, sr)
  91. if err != nil {
  92. t.Fatalf("error: %v", err)
  93. }
  94. got[tensor.Name] = fmt.Sprintf("%x", sha256sum.Sum(nil))
  95. got[tensor.Name+" size"] = fmt.Sprintf("%d", s)
  96. got[tensor.Name+" offset"] = fmt.Sprintf("%v", tensor.Offset)
  97. }
  98. for _, tensor := range wantTensors {
  99. sha256sum := sha256.New()
  100. var s int64
  101. sr := io.NewSectionReader(f2, wantOffset +int64(tensor.Offset), int64(tensor.Size()))
  102. s, err := io.Copy(sha256sum, sr)
  103. if err != nil {
  104. t.Fatalf("error: %v", err)
  105. }
  106. want[tensor.Name] = fmt.Sprintf("%x", sha256sum.Sum(nil))
  107. want[tensor.Name+" size"] = fmt.Sprintf("%d", s)
  108. want[tensor.Name+" offset"] = fmt.Sprintf("%v", tensor.Offset)
  109. }
  110. return got, want
  111. }
  112. func decodeGGML(t *testing.T, f *os.File) (int64, *GGML, error) {
  113. ggml, n, err := DecodeGGML(f, math.MaxInt)
  114. if err != nil {
  115. t.Fatal(err)
  116. }
  117. return n, ggml, nil
  118. }
  119. func rewriteGGML(t *testing.T, ggml *GGML, gotFile *os.File, wantFile *os.File) (int64, *GGML, error) {
  120. var tensors []*Tensor
  121. for _, tensor := range ggml.Tensors().Items {
  122. shape := make([]uint64, len(tensor.Shape))
  123. for i := range len(tensor.Shape) {
  124. shape[i] = tensor.Shape[len(tensor.Shape)-i-1]
  125. }
  126. tensors = append(tensors, &Tensor{
  127. Name: tensor.Name,
  128. Kind: tensor.Kind,
  129. Shape: shape,
  130. WriterTo: TensorWriter{
  131. Reader: io.NewSectionReader(wantFile, ggml.Tensors().Offset+int64(tensor.Offset), int64(tensor.Size())),
  132. },
  133. })
  134. }
  135. reader := &GGUFWriter{
  136. KV: ggml.KV(),
  137. Tensors: Tensors{
  138. Items: tensors,
  139. Offset: ggml.Tensors().Offset,
  140. },
  141. }
  142. n, err := io.Copy(gotFile, reader)
  143. if err != nil {
  144. t.Fatal(err)
  145. }
  146. file, err := os.Open(gotFile.Name())
  147. if err != nil {
  148. t.Fatal(err)
  149. }
  150. ggml2, _, err := DecodeGGML(file, math.MaxInt)
  151. if err != nil {
  152. t.Fatal(err)
  153. }
  154. return n, ggml2, nil
  155. }