model.go 9.14 KB
Newer Older
Michael Yang's avatar
Michael Yang committed
1
2
3
4
5
6
package server

import (
	"archive/zip"
	"bytes"
	"context"
Michael Yang's avatar
tools  
Michael Yang committed
7
	"encoding/json"
Michael Yang's avatar
Michael Yang committed
8
9
10
	"errors"
	"fmt"
	"io"
11
	"log/slog"
Michael Yang's avatar
Michael Yang committed
12
13
14
	"net/http"
	"os"
	"path/filepath"
Michael Yang's avatar
tools  
Michael Yang committed
15
16
17
	"slices"
	"strings"
	"text/template/parse"
Michael Yang's avatar
Michael Yang committed
18
19
20
21

	"github.com/ollama/ollama/api"
	"github.com/ollama/ollama/convert"
	"github.com/ollama/ollama/llm"
Michael Yang's avatar
Michael Yang committed
22
	"github.com/ollama/ollama/template"
Michael Yang's avatar
Michael Yang committed
23
24
25
	"github.com/ollama/ollama/types/model"
)

Michael Yang's avatar
Michael Yang committed
26
var intermediateBlobs map[string]string = make(map[string]string)
27

28
type layerGGML struct {
29
	Layer
Michael Yang's avatar
Michael Yang committed
30
31
32
	*llm.GGML
}

33
func parseFromModel(ctx context.Context, name model.Name, fn func(api.ProgressResponse)) (layers []*layerGGML, err error) {
34
	m, err := ParseNamedManifest(name)
Michael Yang's avatar
Michael Yang committed
35
36
	switch {
	case errors.Is(err, os.ErrNotExist):
Michael Yang's avatar
Michael Yang committed
37
		if err := PullModel(ctx, name.String(), &registryOptions{}, fn); err != nil {
Michael Yang's avatar
Michael Yang committed
38
39
40
			return nil, err
		}

41
		m, err = ParseNamedManifest(name)
Michael Yang's avatar
Michael Yang committed
42
43
44
		if err != nil {
			return nil, err
		}
Michael Yang's avatar
Michael Yang committed
45
46
47
48
	case err != nil:
		return nil, err
	}

49
50
	for _, layer := range m.Layers {
		layer, err := NewLayerFromLayer(layer.Digest, layer.MediaType, name.DisplayShortest())
Michael Yang's avatar
Michael Yang committed
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
		if err != nil {
			return nil, err
		}

		switch layer.MediaType {
		case "application/vnd.ollama.image.model",
			"application/vnd.ollama.image.projector",
			"application/vnd.ollama.image.adapter":
			blobpath, err := GetBlobsPath(layer.Digest)
			if err != nil {
				return nil, err
			}

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

70
			ggml, _, err := llm.DecodeGGML(blob, 0)
Michael Yang's avatar
Michael Yang committed
71
72
73
			if err != nil {
				return nil, err
			}
Michael Yang's avatar
Michael Yang committed
74

75
			layers = append(layers, &layerGGML{layer, ggml})
Michael Yang's avatar
Michael Yang committed
76
		default:
77
			layers = append(layers, &layerGGML{layer, nil})
Michael Yang's avatar
Michael Yang committed
78
79
80
81
82
83
		}
	}

	return layers, nil
}

84
func parseFromZipFile(_ context.Context, command string, baseLayers []*layerGGML, f *os.File, digest string, fn func(api.ProgressResponse)) (layers []*layerGGML, err error) {
85
	fi, err := f.Stat()
Michael Yang's avatar
Michael Yang committed
86
	if err != nil {
87
		return nil, err
Michael Yang's avatar
Michael Yang committed
88
89
	}

90
	r, err := zip.NewReader(f, fi.Size())
91
92
93
94
	if err != nil {
		return nil, err
	}

95
96
	p, err := os.MkdirTemp(filepath.Dir(f.Name()), "")
	if err != nil {
97
98
		return nil, err
	}
99
	defer os.RemoveAll(p)
100

Michael Yang's avatar
Michael Yang committed
101
102
103
	fn(api.ProgressResponse{Status: "converting model"})
	// TODO(mxyng): this should write directly into a layer
	// e.g. NewLayer(arch.Reader(), "application/vnd.ollama.image.model")
104
	t, err := os.CreateTemp(p, "fp16")
Michael Yang's avatar
Michael Yang committed
105
106
107
	if err != nil {
		return nil, err
	}
108
109
	defer t.Close()
	defer os.Remove(t.Name())
Michael Yang's avatar
Michael Yang committed
110

111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
	var layerType string

	switch command {
	case "adapter":
		var baseModel *llm.GGML
		for _, l := range baseLayers {
			if l.GGML != nil {
				baseModel = l.GGML
				break
			}
		}

		if baseModel == nil {
			return nil, fmt.Errorf("no base model specified for the adapter")
		}

		if err := convert.ConvertAdapter(convert.NewZipReader(r, p, 32<<20), t, baseModel.KV()); err != nil {
			return nil, err
		}
		layerType = "application/vnd.ollama.image.adapter"
	case "model":
		if err := convert.ConvertModel(convert.NewZipReader(r, p, 32<<20), t); err != nil {
			return nil, err
		}
		layerType = "application/vnd.ollama.image.model"
Michael Yang's avatar
Michael Yang committed
136
137
	}

138
	if _, err := t.Seek(0, io.SeekStart); err != nil {
Michael Yang's avatar
Michael Yang committed
139
140
141
		return nil, err
	}

142
	layer, err := NewLayer(t, layerType)
Michael Yang's avatar
Michael Yang committed
143
	if err != nil {
Michael Yang's avatar
Michael Yang committed
144
		return nil, err
Michael Yang's avatar
Michael Yang committed
145
146
	}

147
	bin, err := layer.Open()
Michael Yang's avatar
Michael Yang committed
148
149
150
151
152
	if err != nil {
		return nil, err
	}
	defer bin.Close()

153
	ggml, _, err := llm.DecodeGGML(bin, 0)
Michael Yang's avatar
Michael Yang committed
154
155
156
157
	if err != nil {
		return nil, err
	}

158
	layers = append(layers, &layerGGML{layer, ggml})
159

Michael Yang's avatar
Michael Yang committed
160
	intermediateBlobs[digest] = layer.Digest
161
	return detectChatTemplate(layers)
Michael Yang's avatar
Michael Yang committed
162
163
}

164
func parseFromFile(ctx context.Context, command string, baseLayers []*layerGGML, file *os.File, digest string, fn func(api.ProgressResponse)) (layers []*layerGGML, err error) {
Michael Yang's avatar
Michael Yang committed
165
166
167
168
169
170
171
172
173
174
	sr := io.NewSectionReader(file, 0, 512)
	contentType, err := detectContentType(sr)
	if err != nil {
		return nil, err
	}

	switch contentType {
	case "gguf", "ggla":
		// noop
	case "application/zip":
175
		return parseFromZipFile(ctx, command, baseLayers, file, digest, fn)
Michael Yang's avatar
Michael Yang committed
176
177
178
179
180
181
182
183
184
185
186
	default:
		return nil, fmt.Errorf("unsupported content type: %s", contentType)
	}

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

	var offset int64
	for offset < stat.Size() {
187
		ggml, n, err := llm.DecodeGGML(file, 0)
Michael Yang's avatar
Michael Yang committed
188
189
190
191
192
193
194
		if errors.Is(err, io.EOF) {
			break
		} else if err != nil {
			return nil, err
		}

		mediatype := "application/vnd.ollama.image.model"
195
		if ggml.Name() == "ggla" || ggml.KV().Kind() == "adapter" {
Michael Yang's avatar
Michael Yang committed
196
			mediatype = "application/vnd.ollama.image.adapter"
197
198
199
		}

		if _, ok := ggml.KV()[fmt.Sprintf("%s.vision.block_count", ggml.KV().Architecture())]; ok || ggml.KV().Kind() == "projector" {
Michael Yang's avatar
Michael Yang committed
200
201
202
			mediatype = "application/vnd.ollama.image.projector"
		}

Josh's avatar
Josh committed
203
204
205
206
207
208
209
210
211
212
213
214
215
216
		var layer Layer
		if digest != "" && n == stat.Size() && offset == 0 {
			layer, err = NewLayerFromLayer(digest, mediatype, file.Name())
			if err != nil {
				slog.Debug("could not create new layer from layer", "error", 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(file, offset, n), mediatype)
			if err != nil {
				return nil, err
			}
Michael Yang's avatar
Michael Yang committed
217
218
		}

219
		layers = append(layers, &layerGGML{layer, ggml})
Michael Yang's avatar
Michael Yang committed
220
221
222
		offset = n
	}

223
224
225
226
227
228
	return detectChatTemplate(layers)
}

func detectChatTemplate(layers []*layerGGML) ([]*layerGGML, error) {
	for _, layer := range layers {
		if s := layer.GGML.KV().ChatTemplate(); s != "" {
Michael Yang's avatar
Michael Yang committed
229
			if t, err := template.Named(s); err != nil {
230
231
				slog.Debug("template detection", "error", err)
			} else {
232
				layer, err := NewLayer(t.Reader(), "application/vnd.ollama.image.template")
233
234
235
236
				if err != nil {
					return nil, err
				}

237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
				layer.status = fmt.Sprintf("using autodetected template %s", t.Name)
				layers = append(layers, &layerGGML{layer, nil})

				if t.Parameters != nil {
					var b bytes.Buffer
					if err := json.NewEncoder(&b).Encode(t.Parameters); err != nil {
						return nil, err
					}

					layer, err := NewLayer(&b, "application/vnd.ollama.image.params")
					if err != nil {
						return nil, err
					}

					layers = append(layers, &layerGGML{layer, nil})
				}
253
254
255
256
			}
		}
	}

Michael Yang's avatar
Michael Yang committed
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
	return layers, nil
}

func detectContentType(r io.Reader) (string, error) {
	var b bytes.Buffer
	if _, err := io.Copy(&b, r); err != nil {
		return "", err
	}

	if contentType := llm.DetectGGMLType(b.Bytes()); contentType != "" {
		return contentType, nil
	}

	if contentType := http.DetectContentType(b.Bytes()); contentType != "application/octet-stream" {
		return contentType, nil
	}

	return "unknown", nil
}
Michael Yang's avatar
tools  
Michael Yang committed
276

277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
func parseObjects(s string) []map[string]any {
	var objs []map[string]any
	for offset := 0; offset < len(s); {
		var obj map[string]any
		decoder := json.NewDecoder(strings.NewReader(s[offset:]))
		if err := decoder.Decode(&obj); errors.Is(err, io.EOF) || errors.Is(err, io.ErrUnexpectedEOF) {
			break
		} else if syntax := &(json.SyntaxError{}); errors.As(err, &syntax) {
			// skip over any syntax errors
			offset += int(syntax.Offset)
		} else if unmarshalType := &(json.UnmarshalTypeError{}); errors.As(err, &unmarshalType) {
			// skip over any unmarshalable types
			offset += int(unmarshalType.Offset)
		} else if err != nil {
			return nil
		} else {
			offset += int(decoder.InputOffset())
			objs = append(objs, obj)
		}
	}

	return objs
}

Michael Yang's avatar
tools  
Michael Yang committed
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
// parseToolCalls attempts to parse a JSON string into a slice of ToolCalls.
// mxyng: this only really works if the input contains tool calls in some JSON format
func (m *Model) parseToolCalls(s string) ([]api.ToolCall, bool) {
	// create a subtree from the node that ranges over .ToolCalls
	tmpl := m.Template.Subtree(func(n parse.Node) bool {
		if t, ok := n.(*parse.RangeNode); ok {
			return slices.Contains(template.Identifiers(t.Pipe), "ToolCalls")
		}

		return false
	})

	if tmpl == nil {
		return nil, false
	}

	var b bytes.Buffer
318
	if err := tmpl.Execute(&b, map[string][]api.ToolCall{
Michael Yang's avatar
tools  
Michael Yang committed
319
320
		"ToolCalls": {
			{
321
322
323
324
325
				Function: api.ToolCallFunction{
					Name: "@@name@@",
					Arguments: api.ToolCallFunctionArguments{
						"@@argument@@": 1,
					},
Michael Yang's avatar
tools  
Michael Yang committed
326
327
328
329
330
331
332
				},
			},
		},
	}); err != nil {
		return nil, false
	}

333
334
	templateObjects := parseObjects(b.String())
	if len(templateObjects) == 0 {
Michael Yang's avatar
tools  
Michael Yang committed
335
336
337
338
339
		return nil, false
	}

	// find the keys that correspond to the name and arguments fields
	var name, arguments string
340
	for k, v := range templateObjects[0] {
341
342
		switch v.(type) {
		case string:
Michael Yang's avatar
tools  
Michael Yang committed
343
			name = k
344
		case map[string]any:
Michael Yang's avatar
tools  
Michael Yang committed
345
346
347
348
			arguments = k
		}
	}

349
350
351
352
	if name == "" || arguments == "" {
		return nil, false
	}

353
354
355
356
	responseObjects := parseObjects(s)
	if len(responseObjects) == 0 {
		return nil, false
	}
357

358
359
360
361
362
363
364
365
366
367
368
369
	// collect all nested objects
	var collect func(any) []map[string]any
	collect = func(obj any) (all []map[string]any) {
		switch o := obj.(type) {
		case map[string]any:
			all = append(all, o)
			for _, v := range o {
				all = append(all, collect(v)...)
			}
		case []any:
			for _, v := range o {
				all = append(all, collect(v)...)
370
			}
Michael Yang's avatar
tools  
Michael Yang committed
371
		}
372
373
374
375
376
377
378

		return all
	}

	var objs []map[string]any
	for _, p := range responseObjects {
		objs = append(objs, collect(p)...)
Michael Yang's avatar
tools  
Michael Yang committed
379
380
381
	}

	var toolCalls []api.ToolCall
Michael Yang's avatar
Michael Yang committed
382
	for _, kv := range objs {
383
384
385
386
387
388
389
390
391
		n, nok := kv[name].(string)
		a, aok := kv[arguments].(map[string]any)
		if nok && aok {
			toolCalls = append(toolCalls, api.ToolCall{
				Function: api.ToolCallFunction{
					Name:      n,
					Arguments: a,
				},
			})
Michael Yang's avatar
tools  
Michael Yang committed
392
393
394
		}
	}

Michael Yang's avatar
Michael Yang committed
395
	return toolCalls, len(toolCalls) > 0
Michael Yang's avatar
tools  
Michael Yang committed
396
}