lib.rs 23.5 KB
Newer Older
1
mod health;
2
/// Text Generation Inference Webserver
3
mod infer;
4
mod queue;
Olivier Dehaene's avatar
Olivier Dehaene committed
5
pub mod server;
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
6
mod validation;
Olivier Dehaene's avatar
Olivier Dehaene committed
7

8
use infer::{Infer, InferError, InferStreamResponse};
9
use queue::{Entry, Queue};
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
10
use serde::{Deserialize, Serialize};
11
12
use tokio::sync::OwnedSemaphorePermit;
use tokio_stream::wrappers::UnboundedReceiverStream;
13
use utoipa::ToSchema;
Olivier Dehaene's avatar
Olivier Dehaene committed
14
use validation::Validation;
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
15

16
17
18
19
20
21
22
/// Type alias for generation responses
pub(crate) type GenerateStreamResponse = (
    OwnedSemaphorePermit,
    u32, // input_length
    UnboundedReceiverStream<Result<InferStreamResponse, InferError>>,
);

drbh's avatar
drbh committed
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
#[derive(Clone, Deserialize, ToSchema)]
pub(crate) struct VertexInstance {
    #[schema(example = "What is Deep Learning?")]
    pub inputs: String,
    #[schema(nullable = true, default = "null", example = "null")]
    pub parameters: Option<GenerateParameters>,
}

#[derive(Deserialize, ToSchema)]
pub(crate) struct VertexRequest {
    #[serde(rename = "instances")]
    pub instances: Vec<VertexInstance>,
}

#[derive(Clone, Deserialize, ToSchema, Serialize)]
pub(crate) struct VertexResponse {
    pub predictions: Vec<String>,
}

42
43
/// Hub type
#[derive(Clone, Debug, Deserialize)]
44
pub struct HubModelInfo {
45
46
47
48
49
50
    #[serde(rename(deserialize = "id"))]
    pub model_id: String,
    pub sha: Option<String>,
    pub pipeline_tag: Option<String>,
}

51
52
53
#[derive(Clone, Deserialize, Default)]
pub struct HubTokenizerConfig {
    pub chat_template: Option<String>,
54
    #[serde(deserialize_with = "token_serde::deserialize")]
55
    pub bos_token: Option<String>,
56
    #[serde(deserialize_with = "token_serde::deserialize")]
57
    pub eos_token: Option<String>,
58
59
60
}

impl HubTokenizerConfig {
61
    pub fn from_file(filename: &std::path::Path) -> Self {
62
63
64
65
66
        let content = std::fs::read_to_string(filename).unwrap();
        serde_json::from_str(&content).unwrap_or_default()
    }
}

drbh's avatar
drbh committed
67
#[derive(Clone, Debug, Deserialize, ToSchema)]
drbh's avatar
drbh committed
68
69
#[serde(tag = "type", content = "value")]
pub(crate) enum GrammarType {
70
71
72
73
74
75
76
    /// A string that represents a [JSON Schema](https://json-schema.org/).
    ///
    /// JSON Schema is a declarative language that allows to annotate JSON documents
    /// with types and descriptions.
    #[serde(rename = "json")]
    #[schema(example = json ! ({"properties": {"location":{"type": "string"}}}))]
    Json(serde_json::Value),
drbh's avatar
drbh committed
77
78
79
80
    #[serde(rename = "regex")]
    Regex(String),
}

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
108
mod token_serde {
    use super::*;
    use serde::de;
    use serde::Deserializer;
    use serde_json::Value;

    pub fn deserialize<'de, D>(deserializer: D) -> Result<Option<String>, D::Error>
    where
        D: Deserializer<'de>,
    {
        let value = Value::deserialize(deserializer)?;

        match value {
            Value::String(s) => Ok(Some(s)),
            Value::Object(map) => {
                if let Some(content) = map.get("content").and_then(|v| v.as_str()) {
                    Ok(Some(content.to_string()))
                } else {
                    Err(de::Error::custom(
                        "content key not found in structured token",
                    ))
                }
            }
            _ => Err(de::Error::custom("invalid token format")),
        }
    }
}

109
110
#[derive(Clone, Debug, Serialize, ToSchema)]
pub struct Info {
111
    /// Model info
112
113
114
115
    #[schema(example = "bigscience/blomm-560m")]
    pub model_id: String,
    #[schema(nullable = true, example = "e985a63cdc139290c5f700ff1929f0b5942cced2")]
    pub model_sha: Option<String>,
116
117
118
119
    #[schema(example = "torch.float16")]
    pub model_dtype: String,
    #[schema(example = "cuda")]
    pub model_device_type: String,
120
121
    #[schema(nullable = true, example = "text-generation")]
    pub model_pipeline_tag: Option<String>,
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
    /// Router Parameters
    #[schema(example = "128")]
    pub max_concurrent_requests: usize,
    #[schema(example = "2")]
    pub max_best_of: usize,
    #[schema(example = "4")]
    pub max_stop_sequences: usize,
    #[schema(example = "1024")]
    pub max_input_length: usize,
    #[schema(example = "2048")]
    pub max_total_tokens: usize,
    #[schema(example = "1.2")]
    pub waiting_served_ratio: f32,
    #[schema(example = "32000")]
    pub max_batch_total_tokens: u32,
    #[schema(example = "20")]
    pub max_waiting_tokens: usize,
139
140
    #[schema(nullable = true, example = "null")]
    pub max_batch_size: Option<usize>,
141
142
143
    #[schema(example = "2")]
    pub validation_workers: usize,
    /// Router Info
144
145
146
147
    #[schema(example = "0.5.0")]
    pub version: &'static str,
    #[schema(nullable = true, example = "null")]
    pub sha: Option<&'static str>,
148
149
    #[schema(nullable = true, example = "null")]
    pub docker_label: Option<&'static str>,
150
151
}

drbh's avatar
drbh committed
152
#[derive(Clone, Debug, Deserialize, ToSchema, Default)]
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
153
pub(crate) struct GenerateParameters {
154
155
156
    #[serde(default)]
    #[schema(exclusive_minimum = 0, nullable = true, default = "null", example = 1)]
    pub best_of: Option<usize>,
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
    #[serde(default)]
    #[schema(
        exclusive_minimum = 0.0,
        nullable = true,
        default = "null",
        example = 0.5
    )]
    pub temperature: Option<f32>,
    #[serde(default)]
    #[schema(
        exclusive_minimum = 0.0,
        nullable = true,
        default = "null",
        example = 1.03
    )]
    pub repetition_penalty: Option<f32>,
    #[serde(default)]
174
175
176
177
178
179
180
181
    #[schema(
        exclusive_minimum = -2.0,
        nullable = true,
        default = "null",
        example = 0.1
    )]
    pub frequency_penalty: Option<f32>,
    #[serde(default)]
182
183
184
185
186
187
188
189
190
191
192
    #[schema(exclusive_minimum = 0, nullable = true, default = "null", example = 10)]
    pub top_k: Option<i32>,
    #[serde(default)]
    #[schema(
        exclusive_minimum = 0.0,
        maximum = 1.0,
        nullable = true,
        default = "null",
        example = 0.95
    )]
    pub top_p: Option<f32>,
193
    #[serde(default)]
194
195
196
197
198
199
200
201
202
    #[schema(
        exclusive_minimum = 0.0,
        maximum = 1.0,
        nullable = true,
        default = "null",
        example = 0.95
    )]
    pub typical_p: Option<f32>,
    #[serde(default)]
203
    #[schema(default = "false", example = true)]
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
204
205
    pub do_sample: bool,
    #[serde(default = "default_max_new_tokens")]
206
    #[schema(nullable = true, default = "100", example = "20")]
207
    pub max_new_tokens: Option<u32>,
OlivierDehaene's avatar
OlivierDehaene committed
208
    #[serde(default)]
209
    #[schema(nullable = true, default = "null", example = false)]
210
211
    pub return_full_text: Option<bool>,
    #[serde(default)]
212
    #[schema(inline, max_items = 4, example = json ! (["photographer"]))]
213
    pub stop: Vec<String>,
OlivierDehaene's avatar
OlivierDehaene committed
214
    #[serde(default)]
215
    #[schema(nullable = true, default = "null", example = "null")]
216
217
    pub truncate: Option<usize>,
    #[serde(default)]
218
219
220
    #[schema(default = "false", example = true)]
    pub watermark: bool,
    #[serde(default)]
221
    #[schema(default = "true")]
OlivierDehaene's avatar
OlivierDehaene committed
222
    pub details: bool,
223
    #[serde(default)]
224
225
226
    #[schema(default = "true")]
    pub decoder_input_details: bool,
    #[serde(default)]
227
228
229
230
231
232
    #[schema(
        exclusive_minimum = 0,
        nullable = true,
        default = "null",
        example = "null"
    )]
233
    pub seed: Option<u64>,
Nicolas Patry's avatar
Nicolas Patry committed
234
235
236
    #[serde(default)]
    #[schema(exclusive_minimum = 0, nullable = true, default = "null", example = 5)]
    pub top_n_tokens: Option<u32>,
drbh's avatar
drbh committed
237
238
    #[serde(default)]
    pub grammar: Option<GrammarType>,
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
239
240
}

241
fn default_max_new_tokens() -> Option<u32> {
242
    Some(100)
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
243
244
245
246
}

fn default_parameters() -> GenerateParameters {
    GenerateParameters {
247
        best_of: None,
248
249
        temperature: None,
        repetition_penalty: None,
250
        frequency_penalty: None,
251
252
        top_k: None,
        top_p: None,
253
        typical_p: None,
254
        do_sample: true,
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
255
        max_new_tokens: default_max_new_tokens(),
256
        return_full_text: None,
257
        stop: Vec::new(),
258
        truncate: None,
259
        watermark: false,
OlivierDehaene's avatar
OlivierDehaene committed
260
        details: false,
261
        decoder_input_details: false,
262
        seed: None,
Nicolas Patry's avatar
Nicolas Patry committed
263
        top_n_tokens: None,
drbh's avatar
drbh committed
264
        grammar: None,
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
265
266
267
    }
}

268
#[derive(Clone, Deserialize, Serialize, ToSchema)]
269
270
271
pub(crate) struct ChatCompletion {
    pub id: String,
    pub object: String,
272
    #[schema(example = "1706270835")]
273
    pub created: u64,
274
    #[schema(example = "mistralai/Mistral-7B-Instruct-v0.2")]
275
276
277
278
279
280
    pub model: String,
    pub system_fingerprint: String,
    pub choices: Vec<ChatCompletionComplete>,
    pub usage: Usage,
}

281
#[derive(Clone, Deserialize, Serialize, ToSchema)]
282
283
284
pub(crate) struct ChatCompletionComplete {
    pub index: u32,
    pub message: Message,
285
    pub logprobs: Option<ChatCompletionLogprobs>,
286
287
288
    pub finish_reason: String,
}

289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
#[derive(Clone, Deserialize, Serialize, ToSchema)]
pub(crate) struct ChatCompletionLogprobs {
    content: Vec<ChatCompletionLogprob>,
}

impl From<(Token, Vec<Token>)> for ChatCompletionLogprobs {
    fn from(value: (Token, Vec<Token>)) -> Self {
        let (token, top_tokens) = value;

        Self {
            content: vec![ChatCompletionLogprob {
                token: token.text,
                logprob: token.logprob,
                top_logprobs: top_tokens
                    .into_iter()
                    .map(|t| ChatCompletionTopLogprob {
                        token: t.text,
                        logprob: t.logprob,
                    })
                    .collect(),
            }],
        }
    }
}

impl From<(Vec<Token>, Vec<Vec<Token>>)> for ChatCompletionLogprobs {
    fn from(value: (Vec<Token>, Vec<Vec<Token>>)) -> Self {
        let (tokens, top_tokens) = value;
        Self {
            content: tokens
                .into_iter()
                .zip(top_tokens)
                .map(|(t, top_t)| ChatCompletionLogprob {
                    token: t.text,
                    logprob: t.logprob,
                    top_logprobs: top_t
                        .into_iter()
                        .map(|t| ChatCompletionTopLogprob {
                            token: t.text,
                            logprob: t.logprob,
                        })
                        .collect(),
                })
                .collect(),
        }
    }
}

#[derive(Clone, Deserialize, Serialize, ToSchema)]
pub(crate) struct ChatCompletionLogprob {
    token: String,
    logprob: f32,
    top_logprobs: Vec<ChatCompletionTopLogprob>,
}

#[derive(Clone, Deserialize, Serialize, ToSchema)]
pub(crate) struct ChatCompletionTopLogprob {
    token: String,
    logprob: f32,
}

350
#[derive(Clone, Deserialize, Serialize, ToSchema)]
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
pub(crate) struct Usage {
    pub prompt_tokens: u32,
    pub completion_tokens: u32,
    pub total_tokens: u32,
}

impl ChatCompletion {
    pub(crate) fn new(
        model: String,
        system_fingerprint: String,
        output: String,
        created: u64,
        details: Details,
        return_logprobs: bool,
    ) -> Self {
        Self {
            id: String::new(),
            object: "text_completion".into(),
            created,
            model,
            system_fingerprint,
            choices: vec![ChatCompletionComplete {
                index: 0,
                message: Message {
                    role: "assistant".into(),
                    content: output,
377
                    name: None,
378
379
                },
                logprobs: return_logprobs
380
                    .then(|| ChatCompletionLogprobs::from((details.tokens, details.top_tokens))),
381
382
383
384
385
386
387
388
389
390
391
                finish_reason: details.finish_reason.to_string(),
            }],
            usage: Usage {
                prompt_tokens: details.prefill.len() as u32,
                completion_tokens: details.generated_tokens,
                total_tokens: details.prefill.len() as u32 + details.generated_tokens,
            },
        }
    }
}

392
#[derive(Clone, Deserialize, Serialize, ToSchema)]
393
394
395
pub(crate) struct ChatCompletionChunk {
    pub id: String,
    pub object: String,
396
    #[schema(example = "1706270978")]
397
    pub created: u64,
398
    #[schema(example = "mistralai/Mistral-7B-Instruct-v0.2")]
399
400
401
402
403
    pub model: String,
    pub system_fingerprint: String,
    pub choices: Vec<ChatCompletionChoice>,
}

404
#[derive(Clone, Deserialize, Serialize, ToSchema)]
405
406
407
pub(crate) struct ChatCompletionChoice {
    pub index: u32,
    pub delta: ChatCompletionDelta,
408
    pub logprobs: Option<ChatCompletionLogprobs>,
409
410
411
    pub finish_reason: Option<String>,
}

412
#[derive(Clone, Debug, Deserialize, Serialize, ToSchema)]
413
pub(crate) struct ChatCompletionDelta {
414
    #[schema(example = "user")]
415
    pub role: String,
416
    #[schema(example = "What is Deep Learning?")]
417
418
419
420
421
422
423
424
425
426
    pub content: String,
}

impl ChatCompletionChunk {
    pub(crate) fn new(
        model: String,
        system_fingerprint: String,
        delta: String,
        created: u64,
        index: u32,
427
        logprobs: Option<ChatCompletionLogprobs>,
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
        finish_reason: Option<String>,
    ) -> Self {
        Self {
            id: String::new(),
            object: "text_completion".to_string(),
            created,
            model,
            system_fingerprint,
            choices: vec![ChatCompletionChoice {
                index,
                delta: ChatCompletionDelta {
                    role: "assistant".to_string(),
                    content: delta,
                },
                logprobs,
                finish_reason,
            }],
        }
    }
}

#[derive(Clone, Deserialize, ToSchema, Serialize)]
pub(crate) struct ChatRequest {
451
    #[schema(example = "mistralai/Mistral-7B-Instruct-v0.2")]
drbh's avatar
drbh committed
452
    /// [UNUSED] ID of the model to use. See the model endpoint compatibility table for details on which models work with the Chat API.
453
    pub model: String,
drbh's avatar
drbh committed
454

455
    /// A list of messages comprising the conversation so far.
drbh's avatar
drbh committed
456
    #[schema(example = "[{\"role\": \"user\", \"content\": \"What is Deep Learning?\"}]")]
457
458
459
460
461
    pub messages: Vec<Message>,

    /// Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far,
    /// decreasing the model's likelihood to repeat the same line verbatim.
    #[serde(default)]
462
    #[schema(example = "1.0")]
463
464
465
466
467
468
469
470
471
472
473
474
475
476
    pub frequency_penalty: Option<f32>,

    /// UNUSED
    /// Modify the likelihood of specified tokens appearing in the completion. Accepts a JSON object that maps tokens
    /// (specified by their token ID in the tokenizer) to an associated bias value from -100 to 100. Mathematically,
    /// the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model,
    /// but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should
    /// result in a ban or exclusive selection of the relevant token.
    #[serde(default)]
    pub logit_bias: Option<Vec<f32>>,

    /// Whether to return log probabilities of the output tokens or not. If true, returns the log probabilities of each
    /// output token returned in the content of message.
    #[serde(default)]
477
    #[schema(example = "false")]
478
479
480
481
482
    pub logprobs: Option<bool>,

    /// An integer between 0 and 5 specifying the number of most likely tokens to return at each token position, each with
    /// an associated log probability. logprobs must be set to true if this parameter is used.
    #[serde(default)]
483
    #[schema(example = "5")]
484
485
486
487
    pub top_logprobs: Option<u32>,

    /// The maximum number of tokens that can be generated in the chat completion.
    #[serde(default)]
488
    #[schema(example = "32")]
489
490
491
492
493
494
    pub max_tokens: Option<u32>,

    /// UNUSED
    /// How many chat completion choices to generate for each input message. Note that you will be charged based on the
    /// number of generated tokens across all of the choices. Keep n as 1 to minimize costs.
    #[serde(default)]
495
    #[schema(nullable = true, example = "2")]
496
497
498
499
500
    pub n: Option<u32>,

    /// Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far,
    /// increasing the model's likelihood to talk about new topics
    #[serde(default)]
501
    #[schema(nullable = true, example = 0.1)]
502
503
504
505
506
507
508
    pub presence_penalty: Option<f32>,

    #[serde(default = "bool::default")]
    pub stream: bool,

    #[schema(nullable = true, example = 42)]
    pub seed: Option<u64>,
509
510
511
512
513
514

    /// What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while
    /// lower values like 0.2 will make it more focused and deterministic.
    ///
    /// We generally recommend altering this or `top_p` but not both.
    #[serde(default)]
515
    #[schema(nullable = true, example = 1.0)]
516
517
518
519
520
    pub temperature: Option<f32>,

    /// An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the
    /// tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered.
    #[serde(default)]
521
    #[schema(nullable = true, example = 0.95)]
522
    pub top_p: Option<f32>,
523
524
}

525
526
527
528
529
#[derive(Clone, Serialize, Deserialize)]
pub(crate) struct ChatTemplateInputs<'a> {
    messages: Vec<Message>,
    bos_token: Option<&'a str>,
    eos_token: Option<&'a str>,
530
    add_generation_prompt: bool,
531
532
}

533
534
535
536
537
538
#[derive(Clone, Deserialize, ToSchema, Serialize)]
pub(crate) struct Message {
    #[schema(example = "user")]
    pub role: String,
    #[schema(example = "My name is David and I")]
    pub content: String,
drbh's avatar
drbh committed
539
    #[serde(default, skip_serializing_if = "Option::is_none")]
540
541
    #[schema(example = "\"David\"")]
    pub name: Option<String>,
542
543
}

544
#[derive(Clone, Debug, Deserialize, ToSchema)]
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
545
pub(crate) struct GenerateRequest {
546
    #[schema(example = "My name is Olivier and I")]
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
547
548
549
550
551
    pub inputs: String,
    #[serde(default = "default_parameters")]
    pub parameters: GenerateParameters,
}

552
553
554
555
556
557
558
#[derive(Clone, Debug, Deserialize, ToSchema)]
pub(crate) struct CompatGenerateRequest {
    #[schema(example = "My name is Olivier and I")]
    pub inputs: String,
    #[serde(default = "default_parameters")]
    pub parameters: GenerateParameters,
    #[serde(default)]
OlivierDehaene's avatar
OlivierDehaene committed
559
    #[schema(default = "false")]
560
561
562
563
564
565
566
567
568
569
570
571
    pub stream: bool,
}

impl From<CompatGenerateRequest> for GenerateRequest {
    fn from(req: CompatGenerateRequest) -> Self {
        Self {
            inputs: req.inputs,
            parameters: req.parameters,
        }
    }
}

572
573
574
575
576
577
#[derive(Debug, Serialize, ToSchema)]
pub struct PrefillToken {
    #[schema(example = 0)]
    id: u32,
    #[schema(example = "test")]
    text: String,
578
    #[schema(nullable = true, example = - 0.34)]
579
580
581
    logprob: f32,
}

582
#[derive(Debug, Serialize, ToSchema, Clone)]
583
584
585
586
587
pub struct Token {
    #[schema(example = 0)]
    id: u32,
    #[schema(example = "test")]
    text: String,
588
    #[schema(nullable = true, example = - 0.34)]
589
    logprob: f32,
590
591
    #[schema(example = "false")]
    special: bool,
592
593
}

594
595
596
597
598
599
600
601
602
603
604
605
#[derive(Debug, Serialize, ToSchema)]
pub struct SimpleToken {
    #[schema(example = 0)]
    id: u32,
    #[schema(example = "test")]
    text: String,
    #[schema(example = 0)]
    start: usize,
    #[schema(example = 2)]
    stop: usize,
}

606
607
#[derive(Serialize, ToSchema)]
#[serde(rename_all(serialize = "snake_case"))]
608
#[schema(example = "Length")]
609
610
611
612
613
614
615
616
617
pub(crate) enum FinishReason {
    #[schema(rename = "length")]
    Length,
    #[serde(rename = "eos_token")]
    #[schema(rename = "eos_token")]
    EndOfSequenceToken,
    #[schema(rename = "stop_sequence")]
    StopSequence,
}
618

619
620
621
622
623
624
625
626
627
628
impl std::fmt::Display for FinishReason {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            FinishReason::Length => write!(f, "length"),
            FinishReason::EndOfSequenceToken => write!(f, "eos_token"),
            FinishReason::StopSequence => write!(f, "stop_sequence"),
        }
    }
}

629
630
631
632
633
634
635
636
637
638
639
640
#[derive(Serialize, ToSchema)]
pub(crate) struct BestOfSequence {
    #[schema(example = "test")]
    pub generated_text: String,
    #[schema(example = "length")]
    pub finish_reason: FinishReason,
    #[schema(example = 1)]
    pub generated_tokens: u32,
    #[schema(nullable = true, example = 42)]
    pub seed: Option<u64>,
    pub prefill: Vec<PrefillToken>,
    pub tokens: Vec<Token>,
Nicolas Patry's avatar
Nicolas Patry committed
641
642
    #[serde(skip_serializing_if = "Vec::is_empty")]
    pub top_tokens: Vec<Vec<Token>>,
643
644
}

645
#[derive(Serialize, ToSchema)]
OlivierDehaene's avatar
OlivierDehaene committed
646
pub(crate) struct Details {
647
648
649
    #[schema(example = "length")]
    pub finish_reason: FinishReason,
    #[schema(example = 1)]
OlivierDehaene's avatar
OlivierDehaene committed
650
    pub generated_tokens: u32,
651
    #[schema(nullable = true, example = 42)]
652
    pub seed: Option<u64>,
653
654
    pub prefill: Vec<PrefillToken>,
    pub tokens: Vec<Token>,
655
656
    #[serde(skip_serializing_if = "Option::is_none")]
    pub best_of_sequences: Option<Vec<BestOfSequence>>,
Nicolas Patry's avatar
Nicolas Patry committed
657
658
    #[serde(skip_serializing_if = "Vec::is_empty")]
    pub top_tokens: Vec<Vec<Token>>,
OlivierDehaene's avatar
OlivierDehaene committed
659
660
}

661
#[derive(Serialize, ToSchema)]
662
pub(crate) struct GenerateResponse {
663
    #[schema(example = "test")]
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
664
    pub generated_text: String,
OlivierDehaene's avatar
OlivierDehaene committed
665
666
    #[serde(skip_serializing_if = "Option::is_none")]
    pub details: Option<Details>,
Olivier Dehaene's avatar
v0.1.0  
Olivier Dehaene committed
667
}
668

669
670
671
672
#[derive(Serialize, ToSchema)]
#[serde(transparent)]
pub(crate) struct TokenizeResponse(Vec<SimpleToken>);

673
674
675
676
677
678
#[derive(Serialize, ToSchema)]
pub(crate) struct StreamDetails {
    #[schema(example = "length")]
    pub finish_reason: FinishReason,
    #[schema(example = 1)]
    pub generated_tokens: u32,
679
    #[schema(nullable = true, example = 42)]
680
681
682
683
    pub seed: Option<u64>,
}

#[derive(Serialize, ToSchema)]
684
pub(crate) struct StreamResponse {
685
    pub index: u32,
686
    pub token: Token,
Nicolas Patry's avatar
Nicolas Patry committed
687
688
    #[serde(skip_serializing_if = "Vec::is_empty")]
    pub top_tokens: Vec<Token>,
689
    #[schema(nullable = true, default = "null", example = "test")]
690
    pub generated_text: Option<String>,
691
692
    #[schema(nullable = true, default = "null")]
    pub details: Option<StreamDetails>,
693
694
}

695
#[derive(Serialize, ToSchema)]
696
697
pub(crate) struct ErrorResponse {
    pub error: String,
698
    pub error_type: String,
699
}
700
701

#[cfg(test)]
702
mod tests {
703
704
    use super::*;

705
706
    use tokenizers::Tokenizer;

707
    pub(crate) async fn get_tokenizer() -> Tokenizer {
708
709
710
711
        let api = hf_hub::api::sync::Api::new().unwrap();
        let repo = api.model("gpt2".to_string());
        let filename = repo.get("tokenizer.json").unwrap();
        Tokenizer::from_file(filename).unwrap()
712
    }
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765

    #[test]
    fn test_hub_nested_tokens_tokenizer_config() {
        // this is a subset of the tokenizer.json file
        // in this case we expect the tokens to be encoded as simple strings
        let json_content = r#"{
            "chat_template": "test",
            "bos_token": "<|begin▁of▁sentence|>",
            "eos_token": "<|end▁of▁sentence|>"
        }"#;

        let config: HubTokenizerConfig = serde_json::from_str(json_content).unwrap();

        // check that we successfully parsed the tokens
        assert_eq!(config.chat_template, Some("test".to_string()));
        assert_eq!(
            config.bos_token,
            Some("<|begin▁of▁sentence|>".to_string())
        );
        assert_eq!(config.eos_token, Some("<|end▁of▁sentence|>".to_string()));

        // in this case we expect the tokens to be encoded as structured tokens
        // we want the content of the structured token
        let json_content = r#"{
            "chat_template": "test",
            "bos_token": {
              "__type": "AddedToken",
              "content": "<|begin▁of▁sentence|>",
              "lstrip": false,
              "normalized": true,
              "rstrip": false,
              "single_word": false
            },
            "eos_token": {
              "__type": "AddedToken",
              "content": "<|end▁of▁sentence|>",
              "lstrip": false,
              "normalized": true,
              "rstrip": false,
              "single_word": false
            }
        }"#;

        let config: HubTokenizerConfig = serde_json::from_str(json_content).unwrap();

        // check that we successfully parsed the tokens
        assert_eq!(config.chat_template, Some("test".to_string()));
        assert_eq!(
            config.bos_token,
            Some("<|begin▁of▁sentence|>".to_string())
        );
        assert_eq!(config.eos_token, Some("<|end▁of▁sentence|>".to_string()));
    }
766
}