1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
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
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
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
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
//! API Request/Response Types (PMAT-COMPLY)
//!
//! Extracted from mod.rs for file health compliance.
//! Contains all basic API data structures.
use crate::registry::ModelInfo;
use serde::{Deserialize, Serialize};
/// Health check response
#[derive(Serialize, Deserialize)]
pub struct HealthResponse {
/// Service status
pub status: String,
/// Service version
pub version: String,
/// Compute mode: "cpu" or "gpu"
pub compute_mode: String,
}
/// Tokenize request
#[derive(Serialize, Deserialize)]
pub struct TokenizeRequest {
/// Text to tokenize
pub text: String,
/// Model ID (optional, uses default if not specified)
pub model_id: Option<String>,
}
/// Tokenize response
#[derive(Serialize, Deserialize)]
pub struct TokenizeResponse {
/// Token IDs
pub token_ids: Vec<u32>,
/// Number of tokens
pub num_tokens: usize,
}
/// Generate request
#[derive(Serialize, Deserialize)]
pub struct GenerateRequest {
/// Input prompt (token IDs or text)
pub prompt: String,
/// Maximum tokens to generate
#[serde(default = "default_max_tokens")]
pub max_tokens: usize,
/// Sampling temperature
#[serde(default = "default_temperature")]
pub temperature: f32,
/// Sampling strategy: "greedy", "`top_k`", or "`top_p`"
#[serde(default = "default_strategy")]
pub strategy: String,
/// Top-k value (if strategy is "`top_k`")
#[serde(default = "default_top_k")]
pub top_k: usize,
/// Top-p value (if strategy is "`top_p`")
#[serde(default = "default_top_p")]
pub top_p: f32,
/// Random seed for reproducibility
pub seed: Option<u64>,
/// Model ID (optional, uses default if not specified)
pub model_id: Option<String>,
}
/// Default max tokens for generation requests.
pub fn default_max_tokens() -> usize {
50
}
pub(crate) fn default_temperature() -> f32 {
1.0
}
pub(crate) fn default_strategy() -> String {
"greedy".to_string()
}
/// Default top-k value for sampling.
pub fn default_top_k() -> usize {
50
}
pub(crate) fn default_top_p() -> f32 {
0.9
}
/// Generate response
#[derive(Serialize, Deserialize)]
pub struct GenerateResponse {
/// Generated token IDs
pub token_ids: Vec<u32>,
/// Decoded text
pub text: String,
/// Number of generated tokens
pub num_generated: usize,
}
/// Error response
#[derive(Serialize, Deserialize)]
pub struct ErrorResponse {
/// Error message
pub error: String,
}
/// Batch tokenize request
#[derive(Serialize, Deserialize)]
pub struct BatchTokenizeRequest {
/// Texts to tokenize
pub texts: Vec<String>,
}
/// Batch tokenize response
#[derive(Serialize, Deserialize)]
pub struct BatchTokenizeResponse {
/// Results for each text in the same order
pub results: Vec<TokenizeResponse>,
}
/// Batch generate request
#[derive(Serialize, Deserialize)]
pub struct BatchGenerateRequest {
/// Input prompts
pub prompts: Vec<String>,
/// Maximum tokens to generate (shared across all prompts)
#[serde(default = "default_max_tokens")]
pub max_tokens: usize,
/// Sampling temperature (shared)
#[serde(default = "default_temperature")]
pub temperature: f32,
/// Sampling strategy (shared)
#[serde(default = "default_strategy")]
pub strategy: String,
/// Top-k value (shared)
#[serde(default = "default_top_k")]
pub top_k: usize,
/// Top-p value (shared)
#[serde(default = "default_top_p")]
pub top_p: f32,
/// Random seed for reproducibility
pub seed: Option<u64>,
}
/// Batch generate response
#[derive(Serialize, Deserialize)]
pub struct BatchGenerateResponse {
/// Results for each prompt in the same order
pub results: Vec<GenerateResponse>,
}
/// Stream token event (SSE)
#[derive(Serialize, Deserialize)]
pub struct StreamTokenEvent {
/// Token ID
pub token_id: u32,
/// Decoded text for this token
pub text: String,
}
/// Stream done event (SSE)
#[derive(Serialize, Deserialize)]
pub struct StreamDoneEvent {
/// Total number of tokens generated
pub num_generated: usize,
}
/// Models list response
#[derive(Serialize, Deserialize)]
pub struct ModelsResponse {
/// List of available models
pub models: Vec<ModelInfo>,
}