types.go 10 KB
Newer Older
Jeffrey Morgan's avatar
Jeffrey Morgan committed
1
2
package api

3
import (
Michael Yang's avatar
Michael Yang committed
4
	"encoding/json"
5
	"fmt"
Michael Yang's avatar
Michael Yang committed
6
	"math"
7
	"os"
8
9
	"reflect"
	"strings"
10
11
	"time"
)
Michael Yang's avatar
Michael Yang committed
12

Patrick Devine's avatar
Patrick Devine committed
13
type StatusError struct {
14
15
16
	StatusCode   int
	Status       string
	ErrorMessage string `json:"error"`
Patrick Devine's avatar
Patrick Devine committed
17
18
19
}

func (e StatusError) Error() string {
20
21
22
23
24
25
26
27
28
29
	switch {
	case e.Status != "" && e.ErrorMessage != "":
		return fmt.Sprintf("%s: %s", e.Status, e.ErrorMessage)
	case e.Status != "":
		return e.Status
	case e.ErrorMessage != "":
		return e.ErrorMessage
	default:
		// this should not happen
		return "something went wrong, please see the ollama server logs for details"
Patrick Devine's avatar
Patrick Devine committed
30
31
32
	}
}

33
type GenerateRequest struct {
34
35
36
37
38
	Model    string `json:"model"`
	Prompt   string `json:"prompt"`
	System   string `json:"system"`
	Template string `json:"template"`
	Context  []int  `json:"context,omitempty"`
39
	Stream   *bool  `json:"stream,omitempty"`
40
	Raw      bool   `json:"raw,omitempty"`
41
	Format   string `json:"format"`
42

43
	Options map[string]interface{} `json:"options"`
44
45
}

46
47
48
49
50
51
52
53
54
55
56
57
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
// Options specfied in GenerateRequest, if you add a new option here add it to the API docs also
type Options struct {
	Runner

	// Predict options used at runtime
	NumKeep          int      `json:"num_keep,omitempty"`
	Seed             int      `json:"seed,omitempty"`
	NumPredict       int      `json:"num_predict,omitempty"`
	TopK             int      `json:"top_k,omitempty"`
	TopP             float32  `json:"top_p,omitempty"`
	TFSZ             float32  `json:"tfs_z,omitempty"`
	TypicalP         float32  `json:"typical_p,omitempty"`
	RepeatLastN      int      `json:"repeat_last_n,omitempty"`
	Temperature      float32  `json:"temperature,omitempty"`
	RepeatPenalty    float32  `json:"repeat_penalty,omitempty"`
	PresencePenalty  float32  `json:"presence_penalty,omitempty"`
	FrequencyPenalty float32  `json:"frequency_penalty,omitempty"`
	Mirostat         int      `json:"mirostat,omitempty"`
	MirostatTau      float32  `json:"mirostat_tau,omitempty"`
	MirostatEta      float32  `json:"mirostat_eta,omitempty"`
	PenalizeNewline  bool     `json:"penalize_newline,omitempty"`
	Stop             []string `json:"stop,omitempty"`
}

// Runner options which must be set when the model is loaded into memory
type Runner struct {
	UseNUMA            bool    `json:"numa,omitempty"`
	NumCtx             int     `json:"num_ctx,omitempty"`
	NumBatch           int     `json:"num_batch,omitempty"`
	NumGQA             int     `json:"num_gqa,omitempty"`
	NumGPU             int     `json:"num_gpu,omitempty"`
	MainGPU            int     `json:"main_gpu,omitempty"`
	LowVRAM            bool    `json:"low_vram,omitempty"`
	F16KV              bool    `json:"f16_kv,omitempty"`
	LogitsAll          bool    `json:"logits_all,omitempty"`
	VocabOnly          bool    `json:"vocab_only,omitempty"`
	UseMMap            bool    `json:"use_mmap,omitempty"`
	UseMLock           bool    `json:"use_mlock,omitempty"`
	EmbeddingOnly      bool    `json:"embedding_only,omitempty"`
	RopeFrequencyBase  float32 `json:"rope_frequency_base,omitempty"`
	RopeFrequencyScale float32 `json:"rope_frequency_scale,omitempty"`
	NumThread          int     `json:"num_thread,omitempty"`
}

Bruce MacDonald's avatar
Bruce MacDonald committed
90
91
92
93
94
95
96
97
98
99
100
type EmbeddingRequest struct {
	Model  string `json:"model"`
	Prompt string `json:"prompt"`

	Options map[string]interface{} `json:"options"`
}

type EmbeddingResponse struct {
	Embedding []float64 `json:"embedding"`
}

101
type CreateRequest struct {
102
103
104
	Name   string `json:"name"`
	Path   string `json:"path"`
	Stream *bool  `json:"stream,omitempty"`
105
106
}

107
108
109
110
type DeleteRequest struct {
	Name string `json:"name"`
}

Patrick Devine's avatar
Patrick Devine committed
111
112
113
114
115
116
117
118
119
120
121
122
type ShowRequest struct {
	Name string `json:"name"`
}

type ShowResponse struct {
	License    string `json:"license,omitempty"`
	Modelfile  string `json:"modelfile,omitempty"`
	Parameters string `json:"parameters,omitempty"`
	Template   string `json:"template,omitempty"`
	System     string `json:"system,omitempty"`
}

Patrick Devine's avatar
Patrick Devine committed
123
124
125
126
127
type CopyRequest struct {
	Source      string `json:"source"`
	Destination string `json:"destination"`
}

Bruce MacDonald's avatar
Bruce MacDonald committed
128
type PullRequest struct {
129
	Name     string `json:"name"`
130
	Insecure bool   `json:"insecure,omitempty"`
131
132
133
	Username string `json:"username"`
	Password string `json:"password"`
	Stream   *bool  `json:"stream,omitempty"`
Bruce MacDonald's avatar
Bruce MacDonald committed
134
135
}

136
type ProgressResponse struct {
137
138
	Status    string `json:"status"`
	Digest    string `json:"digest,omitempty"`
Michael Yang's avatar
Michael Yang committed
139
140
	Total     int64  `json:"total,omitempty"`
	Completed int64  `json:"completed,omitempty"`
Bruce MacDonald's avatar
Bruce MacDonald committed
141
142
}

143
144
type PushRequest struct {
	Name     string `json:"name"`
145
	Insecure bool   `json:"insecure,omitempty"`
146
147
148
	Username string `json:"username"`
	Password string `json:"password"`
	Stream   *bool  `json:"stream,omitempty"`
149
}
150

Patrick Devine's avatar
Patrick Devine committed
151
type ListResponse struct {
152
	Models []ModelResponse `json:"models"`
Patrick Devine's avatar
Patrick Devine committed
153
154
}

155
type ModelResponse struct {
Patrick Devine's avatar
Patrick Devine committed
156
157
	Name       string    `json:"name"`
	ModifiedAt time.Time `json:"modified_at"`
Michael Yang's avatar
Michael Yang committed
158
	Size       int64     `json:"size"`
Patrick Devine's avatar
Patrick Devine committed
159
	Digest     string    `json:"digest"`
Patrick Devine's avatar
Patrick Devine committed
160
161
}

Patrick Devine's avatar
Patrick Devine committed
162
163
164
165
type TokenResponse struct {
	Token string `json:"token"`
}

Michael Yang's avatar
Michael Yang committed
166
type GenerateResponse struct {
167
168
	Model     string    `json:"model"`
	CreatedAt time.Time `json:"created_at"`
169
	Response  string    `json:"response"`
170

Michael Yang's avatar
Michael Yang committed
171
172
	Done    bool  `json:"done"`
	Context []int `json:"context,omitempty"`
173
174

	TotalDuration      time.Duration `json:"total_duration,omitempty"`
Michael Yang's avatar
Michael Yang committed
175
	LoadDuration       time.Duration `json:"load_duration,omitempty"`
176
177
178
179
180
181
182
183
184
185
186
	PromptEvalCount    int           `json:"prompt_eval_count,omitempty"`
	PromptEvalDuration time.Duration `json:"prompt_eval_duration,omitempty"`
	EvalCount          int           `json:"eval_count,omitempty"`
	EvalDuration       time.Duration `json:"eval_duration,omitempty"`
}

func (r *GenerateResponse) Summary() {
	if r.TotalDuration > 0 {
		fmt.Fprintf(os.Stderr, "total duration:       %v\n", r.TotalDuration)
	}

Michael Yang's avatar
Michael Yang committed
187
188
189
190
	if r.LoadDuration > 0 {
		fmt.Fprintf(os.Stderr, "load duration:        %v\n", r.LoadDuration)
	}

191
192
193
194
195
196
197
198
199
200
201
202
203
204
	if r.PromptEvalCount > 0 {
		fmt.Fprintf(os.Stderr, "prompt eval count:    %d token(s)\n", r.PromptEvalCount)
	}

	if r.PromptEvalDuration > 0 {
		fmt.Fprintf(os.Stderr, "prompt eval duration: %s\n", r.PromptEvalDuration)
		fmt.Fprintf(os.Stderr, "prompt eval rate:     %.2f tokens/s\n", float64(r.PromptEvalCount)/r.PromptEvalDuration.Seconds())
	}

	if r.EvalCount > 0 {
		fmt.Fprintf(os.Stderr, "eval count:           %d token(s)\n", r.EvalCount)
	}

	if r.EvalDuration > 0 {
Michael Yang's avatar
Michael Yang committed
205
		fmt.Fprintf(os.Stderr, "eval duration:        %s\n", r.EvalDuration)
206
207
		fmt.Fprintf(os.Stderr, "eval rate:            %.2f tokens/s\n", float64(r.EvalCount)/r.EvalDuration.Seconds())
	}
208
209
}

210
211
var ErrInvalidOpts = fmt.Errorf("invalid options")

212
213
214
215
216
217
218
219
220
221
222
223
224
func (opts *Options) FromMap(m map[string]interface{}) error {
	valueOpts := reflect.ValueOf(opts).Elem() // names of the fields in the options struct
	typeOpts := reflect.TypeOf(opts).Elem()   // types of the fields in the options struct

	// build map of json struct tags to their types
	jsonOpts := make(map[string]reflect.StructField)
	for _, field := range reflect.VisibleFields(typeOpts) {
		jsonTag := strings.Split(field.Tag.Get("json"), ",")[0]
		if jsonTag != "" {
			jsonOpts[jsonTag] = field
		}
	}

225
	invalidOpts := []string{}
226
227
228
229
	for key, val := range m {
		if opt, ok := jsonOpts[key]; ok {
			field := valueOpts.FieldByName(opt.Name)
			if field.IsValid() && field.CanSet() {
Michael Yang's avatar
Michael Yang committed
230
231
232
233
				if val == nil {
					continue
				}

234
235
				switch field.Kind() {
				case reflect.Int:
Michael Yang's avatar
Michael Yang committed
236
237
238
239
240
241
242
					switch t := val.(type) {
					case int64:
						field.SetInt(t)
					case float64:
						// when JSON unmarshals numbers, it uses float64, not int
						field.SetInt(int64(t))
					default:
243
						return fmt.Errorf("option %q must be of type integer", key)
244
245
246
247
					}
				case reflect.Bool:
					val, ok := val.(bool)
					if !ok {
248
						return fmt.Errorf("option %q must be of type boolean", key)
249
250
251
252
253
254
					}
					field.SetBool(val)
				case reflect.Float32:
					// JSON unmarshals to float64
					val, ok := val.(float64)
					if !ok {
255
						return fmt.Errorf("option %q must be of type float32", key)
256
257
258
259
260
					}
					field.SetFloat(val)
				case reflect.String:
					val, ok := val.(string)
					if !ok {
261
						return fmt.Errorf("option %q must be of type string", key)
262
263
264
265
266
267
					}
					field.SetString(val)
				case reflect.Slice:
					// JSON unmarshals to []interface{}, not []string
					val, ok := val.([]interface{})
					if !ok {
268
						return fmt.Errorf("option %q must be of type array", key)
269
270
271
272
273
274
					}
					// convert []interface{} to []string
					slice := make([]string, len(val))
					for i, item := range val {
						str, ok := item.(string)
						if !ok {
275
							return fmt.Errorf("option %q must be of an array of strings", key)
276
277
278
279
280
281
282
283
						}
						slice[i] = str
					}
					field.Set(reflect.ValueOf(slice))
				default:
					return fmt.Errorf("unknown type loading config params: %v", field.Kind())
				}
			}
284
285
		} else {
			invalidOpts = append(invalidOpts, key)
286
287
		}
	}
288
289
290
291

	if len(invalidOpts) > 0 {
		return fmt.Errorf("%w: %v", ErrInvalidOpts, strings.Join(invalidOpts, ", "))
	}
292
293
294
	return nil
}

Michael Yang's avatar
Michael Yang committed
295
296
func DefaultOptions() Options {
	return Options{
297
298
		// options set on request to runner
		NumPredict:       -1,
299
		NumKeep:          0,
Michael Yang's avatar
Michael Yang committed
300
301
302
303
304
		Temperature:      0.8,
		TopK:             40,
		TopP:             0.9,
		TFSZ:             1.0,
		TypicalP:         1.0,
305
306
307
308
		RepeatLastN:      64,
		RepeatPenalty:    1.1,
		PresencePenalty:  0.0,
		FrequencyPenalty: 0.0,
Michael Yang's avatar
Michael Yang committed
309
310
311
		Mirostat:         0,
		MirostatTau:      5.0,
		MirostatEta:      0.1,
Michael Yang's avatar
Michael Yang committed
312
		PenalizeNewline:  true,
313
		Seed:             -1,
Michael Yang's avatar
Michael Yang committed
314

315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
		Runner: Runner{
			// options set when the model is loaded
			NumCtx:             2048,
			RopeFrequencyBase:  10000.0,
			RopeFrequencyScale: 1.0,
			NumBatch:           512,
			NumGPU:             -1, // -1 here indicates that NumGPU should be set dynamically
			NumGQA:             1,
			NumThread:          0, // let the runtime decide
			LowVRAM:            false,
			F16KV:              true,
			UseMLock:           false,
			UseMMap:            true,
			UseNUMA:            false,
			EmbeddingOnly:      true,
		},
Michael Yang's avatar
Michael Yang committed
331
	}
Jeffrey Morgan's avatar
Jeffrey Morgan committed
332
}
Michael Yang's avatar
Michael Yang committed
333
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

type Duration struct {
	time.Duration
}

func (d *Duration) UnmarshalJSON(b []byte) (err error) {
	var v any
	if err := json.Unmarshal(b, &v); err != nil {
		return err
	}

	d.Duration = 5 * time.Minute

	switch t := v.(type) {
	case float64:
		if t < 0 {
			t = math.MaxFloat64
		}

		d.Duration = time.Duration(t)
	case string:
		d.Duration, err = time.ParseDuration(t)
		if err != nil {
			return err
		}
	}

	return nil
}