create.go 17.1 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
package server

import (
	"bytes"
	"cmp"
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"io"
11
	"io/fs"
12
13
14
15
16
17
18
19
20
21
22
23
24
	"log/slog"
	"net/http"
	"os"
	"path/filepath"
	"slices"
	"strings"

	"github.com/gin-gonic/gin"

	"github.com/ollama/ollama/api"
	"github.com/ollama/ollama/convert"
	"github.com/ollama/ollama/envconfig"
	"github.com/ollama/ollama/format"
Michael Yang's avatar
Michael Yang committed
25
	"github.com/ollama/ollama/fs/ggml"
26
27
28
29
30
31
32
33
34
35
36
	"github.com/ollama/ollama/llama"
	"github.com/ollama/ollama/template"
	"github.com/ollama/ollama/types/errtypes"
	"github.com/ollama/ollama/types/model"
)

var (
	errNoFilesProvided         = errors.New("no files provided to convert")
	errOnlyOneAdapterSupported = errors.New("only one adapter is currently supported")
	errOnlyGGUFSupported       = errors.New("supplied file was not in GGUF format")
	errUnknownType             = errors.New("unknown type")
37
	errNeitherFromOrFiles      = errors.New("neither 'from' or 'files' was specified")
38
	errFilePath                = errors.New("file path must be relative")
39
40
41
42
43
44
45
46
47
48
49
50
)

func (s *Server) CreateHandler(c *gin.Context) {
	var r api.CreateRequest
	if err := c.ShouldBindJSON(&r); errors.Is(err, io.EOF) {
		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
		return
	} else if err != nil {
		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
		return
	}

51
52
53
54
55
56
57
	for v := range r.Files {
		if !fs.ValidPath(v) {
			c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": errFilePath.Error()})
			return
		}
	}

58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
	name := model.ParseName(cmp.Or(r.Model, r.Name))
	if !name.IsValid() {
		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": errtypes.InvalidModelNameErrMsg})
		return
	}

	name, err := getExistingName(name)
	if err != nil {
		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
		return
	}

	ch := make(chan any)
	go func() {
		defer close(ch)
		fn := func(resp api.ProgressResponse) {
			ch <- resp
		}

		oldManifest, _ := ParseNamedManifest(name)

		var baseLayers []*layerGGML
		if r.From != "" {
			slog.Debug("create model from model name")
			fromName := model.ParseName(r.From)
			if !fromName.IsValid() {
				ch <- gin.H{"error": errtypes.InvalidModelNameErrMsg, "status": http.StatusBadRequest}
				return
			}

			ctx, cancel := context.WithCancel(c.Request.Context())
			defer cancel()

			baseLayers, err = parseFromModel(ctx, fromName, fn)
			if err != nil {
				ch <- gin.H{"error": err.Error()}
			}
		} else if r.Files != nil {
			baseLayers, err = convertModelFromFiles(r.Files, baseLayers, false, fn)
			if err != nil {
				for _, badReq := range []error{errNoFilesProvided, errOnlyGGUFSupported, errUnknownType} {
					if errors.Is(err, badReq) {
						ch <- gin.H{"error": err.Error(), "status": http.StatusBadRequest}
						return
					}
				}
				ch <- gin.H{"error": err.Error()}
				return
			}
		} else {
108
			ch <- gin.H{"error": errNeitherFromOrFiles.Error(), "status": http.StatusBadRequest}
109
110
111
112
113
114
115
			return
		}

		var adapterLayers []*layerGGML
		if r.Adapters != nil {
			adapterLayers, err = convertModelFromFiles(r.Adapters, baseLayers, true, fn)
			if err != nil {
116
				for _, badReq := range []error{errNoFilesProvided, errOnlyOneAdapterSupported, errOnlyGGUFSupported, errUnknownType, errFilePath} {
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
					if errors.Is(err, badReq) {
						ch <- gin.H{"error": err.Error(), "status": http.StatusBadRequest}
						return
					}
				}
				ch <- gin.H{"error": err.Error(), "status": http.StatusBadRequest}
				return
			}
		}

		if len(adapterLayers) > 0 {
			baseLayers = append(baseLayers, adapterLayers...)
		}

		if err := createModel(r, name, baseLayers, fn); err != nil {
			if errors.Is(err, errBadTemplate) {
				ch <- gin.H{"error": err.Error(), "status": http.StatusBadRequest}
				return
			}
			ch <- gin.H{"error": err.Error()}
			return
		}

		if !envconfig.NoPrune() && oldManifest != nil {
			if err := oldManifest.RemoveLayers(); err != nil {
				ch <- gin.H{"error": err.Error()}
			}
		}

		ch <- api.ProgressResponse{Status: "success"}
	}()

	if r.Stream != nil && !*r.Stream {
		waitForStream(c, ch)
		return
	}

	streamResponse(c, ch)
}

func convertModelFromFiles(files map[string]string, baseLayers []*layerGGML, isAdapter bool, fn func(resp api.ProgressResponse)) ([]*layerGGML, error) {
	switch detectModelTypeFromFiles(files) {
	case "safetensors":
		layers, err := convertFromSafetensors(files, baseLayers, isAdapter, fn)
		if err != nil {
			slog.Error("error converting from safetensors", "error", err)
			return nil, err
		}
		return layers, nil
	case "gguf":
		if len(files) == 0 {
			return nil, errNoFilesProvided
		} else if len(files) > 1 && isAdapter {
			return nil, errOnlyOneAdapterSupported
		}

		var digest string
		var allLayers []*layerGGML
		for _, v := range files {
			digest = v
			layers, err := ggufLayers(digest, fn)
			if err != nil {
				return nil, err
			}
			allLayers = append(allLayers, layers...)
		}
		return allLayers, nil
	default:
		return nil, errUnknownType
	}
}

func detectModelTypeFromFiles(files map[string]string) string {
	for fn := range files {
		if strings.HasSuffix(fn, ".safetensors") {
			return "safetensors"
193
		} else if strings.HasSuffix(fn, ".gguf") {
194
			return "gguf"
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
		} else {
			// try to see if we can find a gguf file even without the file extension
			blobPath, err := GetBlobsPath(files[fn])
			if err != nil {
				slog.Error("error getting blobs path", "file", fn)
				return ""
			}

			f, err := os.Open(blobPath)
			if err != nil {
				slog.Error("error reading file", "error", err)
				return ""
			}
			defer f.Close()

			buf := make([]byte, 4)
			_, err = f.Read(buf)
			if err != nil {
				slog.Error("error reading file", "error", err)
				return ""
			}

Michael Yang's avatar
Michael Yang committed
217
			ct := ggml.DetectContentType(buf)
218
219
220
			if ct == "gguf" {
				return "gguf"
			}
221
222
223
224
225
226
227
228
229
230
231
232
		}
	}

	return ""
}

func convertFromSafetensors(files map[string]string, baseLayers []*layerGGML, isAdapter bool, fn func(resp api.ProgressResponse)) ([]*layerGGML, error) {
	tmpDir, err := os.MkdirTemp("", "ollama-safetensors")
	if err != nil {
		return nil, err
	}
	defer os.RemoveAll(tmpDir)
233
234
235
236
237
238
	// Set up a root to validate paths
	root, err := os.OpenRoot(tmpDir)
	if err != nil {
		return nil, err
	}
	defer root.Close()
239
240

	for fp, digest := range files {
241
242
243
244
245
246
247
248
		if !fs.ValidPath(fp) {
			return nil, fmt.Errorf("%w: %s", errFilePath, fp)
		}
		if _, err := root.Stat(fp); err != nil && !errors.Is(err, fs.ErrNotExist) {
			// Path is likely outside the root
			return nil, fmt.Errorf("%w: %s: %s", errFilePath, err, fp)
		}

249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
		blobPath, err := GetBlobsPath(digest)
		if err != nil {
			return nil, err
		}
		if err := createLink(blobPath, filepath.Join(tmpDir, fp)); err != nil {
			return nil, err
		}
	}

	t, err := os.CreateTemp(tmpDir, "fp16")
	if err != nil {
		return nil, err
	}
	defer t.Close()

	var mediaType string
	if !isAdapter {
		fn(api.ProgressResponse{Status: "converting model"})
		mediaType = "application/vnd.ollama.image.model"
		if err := convert.ConvertModel(os.DirFS(tmpDir), t); err != nil {
			return nil, err
		}
	} else {
		kv, err := kvFromLayers(baseLayers)
		if err != nil {
			return nil, err
		}
		fn(api.ProgressResponse{Status: "converting adapter"})
		mediaType = "application/vnd.ollama.image.adapter"
		if err := convert.ConvertAdapter(os.DirFS(tmpDir), t, kv); err != nil {
			return nil, err
		}
	}

	if _, err := t.Seek(0, io.SeekStart); err != nil {
		return nil, err
	}

	layer, err := NewLayer(t, mediaType)
	if err != nil {
		return nil, err
	}

	bin, err := layer.Open()
	if err != nil {
		return nil, err
	}
296
	defer bin.Close()
297

Michael Yang's avatar
Michael Yang committed
298
	f, _, err := ggml.Decode(bin, 0)
299
300
301
	if err != nil {
		return nil, err
	}
Michael Yang's avatar
Michael Yang committed
302
	layers := []*layerGGML{{layer, f}}
303
304
305
306
307
308
309

	if !isAdapter {
		return detectChatTemplate(layers)
	}
	return layers, nil
}

Michael Yang's avatar
Michael Yang committed
310
func kvFromLayers(baseLayers []*layerGGML) (ggml.KV, error) {
311
312
313
314
315
	for _, l := range baseLayers {
		if l.GGML != nil {
			return l.KV(), nil
		}
	}
Michael Yang's avatar
Michael Yang committed
316
	return ggml.KV{}, fmt.Errorf("no base model was found")
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
}

func createModel(r api.CreateRequest, name model.Name, baseLayers []*layerGGML, fn func(resp api.ProgressResponse)) (err error) {
	config := ConfigV2{
		OS:           "linux",
		Architecture: "amd64",
		RootFS: RootFS{
			Type: "layers",
		},
	}

	var layers []Layer
	for _, layer := range baseLayers {
		if layer.GGML != nil {
			quantType := strings.ToUpper(cmp.Or(r.Quantize, r.Quantization))
			if quantType != "" && layer.GGML.Name() == "gguf" && layer.MediaType == "application/vnd.ollama.image.model" {
Michael Yang's avatar
Michael Yang committed
333
				want, err := ggml.ParseFileType(quantType)
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
				if err != nil {
					return err
				}

				ft := layer.GGML.KV().FileType()
				if !slices.Contains([]string{"F16", "F32"}, ft.String()) {
					return errors.New("quantization is only supported for F16 and F32 models")
				} else if ft != want {
					layer, err = quantizeLayer(layer, quantType, fn)
					if err != nil {
						return err
					}
				}
			}
			config.ModelFormat = cmp.Or(config.ModelFormat, layer.GGML.Name())
			config.ModelFamily = cmp.Or(config.ModelFamily, layer.GGML.KV().Architecture())
			config.ModelType = cmp.Or(config.ModelType, format.HumanNumber(layer.GGML.KV().ParameterCount()))
			config.FileType = cmp.Or(config.FileType, layer.GGML.KV().FileType().String())
			config.ModelFamilies = append(config.ModelFamilies, layer.GGML.KV().Architecture())
		}
		layers = append(layers, layer.Layer)
	}

	if r.Template != "" {
		layers, err = setTemplate(layers, r.Template)
		if err != nil {
			return err
		}
	}

	if r.System != "" {
		layers, err = setSystem(layers, r.System)
		if err != nil {
			return err
		}
	}

	if r.License != nil {
		switch l := r.License.(type) {
		case string:
			if l != "" {
				layers, err = setLicense(layers, l)
				if err != nil {
					return err
				}
			}
		case any:
			var licenses []string
			b, _ := json.Marshal(l) // re-marshal to JSON
			if err := json.Unmarshal(b, &licenses); err != nil {
				return err
			}
			for _, v := range licenses {
				layers, err = setLicense(layers, v)
				if err != nil {
					return err
				}
			}
		default:
			return fmt.Errorf("unknown license type: %T", l)
		}
	}

	layers, err = setParameters(layers, r.Parameters)
	if err != nil {
		return err
	}

	layers, err = setMessages(layers, r.Messages)
	if err != nil {
		return err
	}

	configLayer, err := createConfigLayer(layers, config)
	if err != nil {
		return err
	}

	for _, layer := range layers {
		if layer.status != "" {
			fn(api.ProgressResponse{Status: layer.status})
		}
	}

	fn(api.ProgressResponse{Status: "writing manifest"})
	if err := WriteManifest(name, *configLayer, layers); err != nil {
		return err
	}

	return nil
}

func quantizeLayer(layer *layerGGML, quantizeType string, fn func(resp api.ProgressResponse)) (*layerGGML, error) {
	ft := layer.GGML.KV().FileType()
	fn(api.ProgressResponse{Status: fmt.Sprintf("quantizing %s model to %s", ft, quantizeType)})

Michael Yang's avatar
Michael Yang committed
430
	want, err := ggml.ParseFileType(quantizeType)
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
	if err != nil {
		return nil, err
	}

	blob, err := GetBlobsPath(layer.Digest)
	if err != nil {
		return nil, err
	}

	temp, err := os.CreateTemp(filepath.Dir(blob), quantizeType)
	if err != nil {
		return nil, err
	}
	defer temp.Close()
	defer os.Remove(temp.Name())

	if err := llama.Quantize(blob, temp.Name(), uint32(want)); err != nil {
		return nil, err
	}

	newLayer, err := NewLayer(temp, layer.MediaType)
	if err != nil {
		return nil, err
	}

	if _, err := temp.Seek(0, io.SeekStart); err != nil {
		return nil, err
	}

Michael Yang's avatar
Michael Yang committed
460
	f, _, err := ggml.Decode(temp, 0)
461
462
463
464
465
	if err != nil {
		slog.Error(fmt.Sprintf("error decoding ggml: %s\n", err))
		return nil, err
	}

Michael Yang's avatar
Michael Yang committed
466
	return &layerGGML{newLayer, f}, nil
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
}

func ggufLayers(digest string, fn func(resp api.ProgressResponse)) ([]*layerGGML, error) {
	var layers []*layerGGML

	fn(api.ProgressResponse{Status: "parsing GGUF"})
	blobPath, err := GetBlobsPath(digest)
	if err != nil {
		return nil, err
	}

	blob, err := os.Open(blobPath)
	if err != nil {
		return nil, err
	}
	defer blob.Close()

	sr := io.NewSectionReader(blob, 0, 512)
	contentType, err := detectContentType(sr)
	if err != nil {
		return nil, err
	}

	if contentType != "gguf" {
		slog.Error(fmt.Sprintf("unsupported content type: %s", contentType))
		return nil, errOnlyGGUFSupported
	}

	stat, err := blob.Stat()
	if err != nil {
		return nil, err
	}

	var offset int64
	for offset < stat.Size() {
Michael Yang's avatar
Michael Yang committed
502
		f, n, err := ggml.Decode(blob, 0)
503
504
505
506
507
508
509
		if errors.Is(err, io.EOF) {
			break
		} else if err != nil {
			return nil, err
		}

		mediatype := "application/vnd.ollama.image.model"
Michael Yang's avatar
Michael Yang committed
510
		if f.KV().Kind() == "adapter" {
511
			mediatype = "application/vnd.ollama.image.adapter"
Michael Yang's avatar
Michael Yang committed
512
		} else if _, ok := f.KV()[fmt.Sprintf("%s.vision.block_count", f.KV().Architecture())]; ok || f.KV().Kind() == "projector" {
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
			mediatype = "application/vnd.ollama.image.projector"
		}

		var layer Layer
		if digest != "" && n == stat.Size() && offset == 0 {
			layer, err = NewLayerFromLayer(digest, mediatype, blob.Name())
			if err != nil {
				slog.Debug("could not create new layer from layer", "error", err)
				return nil, err
			}
		}

		// Fallback to creating layer from file copy (either NewLayerFromLayer failed, or digest empty/n != stat.Size())
		if layer.Digest == "" {
			layer, err = NewLayer(io.NewSectionReader(blob, offset, n), mediatype)
			if err != nil {
				return nil, err
			}
		}

Michael Yang's avatar
Michael Yang committed
533
		layers = append(layers, &layerGGML{layer, f})
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
		offset = n
	}

	return detectChatTemplate(layers)
}

func removeLayer(layers []Layer, mediatype string) []Layer {
	return slices.DeleteFunc(layers, func(layer Layer) bool {
		if layer.MediaType != mediatype {
			return false
		}

		if err := layer.Remove(); err != nil {
			slog.Warn("couldn't remove blob", "digest", layer.Digest, "error", err)
			return true
		}

		return true
	})
}

func setTemplate(layers []Layer, t string) ([]Layer, error) {
	layers = removeLayer(layers, "application/vnd.ollama.image.template")
	if _, err := template.Parse(t); err != nil {
		return nil, fmt.Errorf("%w: %s", errBadTemplate, err)
	}
	if _, err := template.Parse(t); err != nil {
		return nil, fmt.Errorf("%w: %s", errBadTemplate, err)
	}

	blob := strings.NewReader(t)
	layer, err := NewLayer(blob, "application/vnd.ollama.image.template")
	if err != nil {
		return nil, err
	}

	layers = append(layers, layer)
	return layers, nil
}

func setSystem(layers []Layer, s string) ([]Layer, error) {
	layers = removeLayer(layers, "application/vnd.ollama.image.system")
	if s != "" {
		blob := strings.NewReader(s)
		layer, err := NewLayer(blob, "application/vnd.ollama.image.system")
		if err != nil {
			return nil, err
		}
		layers = append(layers, layer)
	}
	return layers, nil
}

func setLicense(layers []Layer, l string) ([]Layer, error) {
	blob := strings.NewReader(l)
	layer, err := NewLayer(blob, "application/vnd.ollama.image.license")
	if err != nil {
		return nil, err
	}
	layers = append(layers, layer)
	return layers, nil
}

func setParameters(layers []Layer, p map[string]any) ([]Layer, error) {
	if p == nil {
		p = make(map[string]any)
	}
	for _, layer := range layers {
		if layer.MediaType != "application/vnd.ollama.image.params" {
			continue
		}

		digestPath, err := GetBlobsPath(layer.Digest)
		if err != nil {
			return nil, err
		}

		fn, err := os.Open(digestPath)
		if err != nil {
			return nil, err
		}
		defer fn.Close()

		var existing map[string]any
		if err := json.NewDecoder(fn).Decode(&existing); err != nil {
			return nil, err
		}

		for k, v := range existing {
			if _, exists := p[k]; exists {
				continue
			}
			p[k] = v
		}
	}

	if len(p) == 0 {
		return layers, nil
	}

	layers = removeLayer(layers, "application/vnd.ollama.image.params")

	var b bytes.Buffer
	if err := json.NewEncoder(&b).Encode(p); err != nil {
		return nil, err
	}
	layer, err := NewLayer(&b, "application/vnd.ollama.image.params")
	if err != nil {
		return nil, err
	}
	layers = append(layers, layer)
	return layers, nil
}

func setMessages(layers []Layer, m []api.Message) ([]Layer, error) {
	// this leaves the old messages intact if no new messages were specified
	// which may not be the correct behaviour
	if len(m) == 0 {
		return layers, nil
	}

	fmt.Printf("removing old messages\n")
	layers = removeLayer(layers, "application/vnd.ollama.image.messages")
	var b bytes.Buffer
	if err := json.NewEncoder(&b).Encode(m); err != nil {
		return nil, err
	}
	layer, err := NewLayer(&b, "application/vnd.ollama.image.messages")
	if err != nil {
		return nil, err
	}
	layers = append(layers, layer)
	return layers, nil
}

func createConfigLayer(layers []Layer, config ConfigV2) (*Layer, error) {
	digests := make([]string, len(layers))
	for i, layer := range layers {
		digests[i] = layer.Digest
	}
	config.RootFS.DiffIDs = digests

	var b bytes.Buffer
	if err := json.NewEncoder(&b).Encode(config); err != nil {
		return nil, err
	}
	layer, err := NewLayer(&b, "application/vnd.docker.container.image.v1+json")
	if err != nil {
		return nil, err
	}
	return &layer, nil
}

func createLink(src, dst string) error {
	// make any subdirs for dst
	if err := os.MkdirAll(filepath.Dir(dst), 0o755); err != nil {
		return err
	}

	_ = os.Remove(dst)
	if err := os.Symlink(src, dst); err != nil {
		if err := copyFile(src, dst); err != nil {
			return err
		}
	}
	return nil
}

func copyFile(src, dst string) error {
	srcFile, err := os.Open(src)
	if err != nil {
		return err
	}
	defer srcFile.Close()

	dstFile, err := os.Create(dst)
	if err != nil {
		return err
	}
	defer dstFile.Close()

	_, err = io.Copy(dstFile, srcFile)
	return err
}