blob: bafa7b18a0a1c018f65ea1a2fcd4e34baeb6d0f7 (
plain) (
blame)
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
167
168
169
170
|
syntax = "proto3";
package yandex.cloud.ai.llm.v1alpha;
import "yandex/cloud/ai/llm/v1alpha/llm.proto";
import "google/api/annotations.proto";
import "yandex/cloud/validation.proto";
import "yandex/cloud/api/operation.proto";
import "yandex/cloud/operation/operation.proto";
option go_package = "github.com/yandex-cloud/go-genproto/yandex/cloud/ai/llm/v1alpha;llm";
option java_package = "yandex.cloud.api.ai.llm.v1alpha";
// Request for instructing the model to generate text.
message InstructRequest {
// The name or identifier of the model to be used for text generation.
// Possible value for now: `general`.
string model = 1 [(length) = "<=50"];
// Configuration options for text generation.
GenerationOptions generation_options = 2;
// Text precondition or context of the request.
// For example, if the instruction is "You are the youngest Nobel laureate",
// the request text might be "Tell us about your daily routine".
oneof Instruction {
// The text-based instruction for text generation.
string instruction_text = 3;
// A URI containing instructions for text generation.
string instruction_uri = 5;
}
// Request for text generation.
oneof Request {
// The text-based request for text generation.
string request_text = 4;
}
}
// Response containing generated text alternatives and token count.
message InstructResponse {
// A list of alternative text responses.
repeated Alternative alternatives = 1;
// The number of tokens used in the prompt, including both the [instruction_text] and [request_text].
int64 num_prompt_tokens = 2;
}
// Request to engage in a chat conversation with a text generation model.
message ChatRequest {
// The name or identifier of the model to be used for the chat.
// Possible value for now: `general`.
string model = 1 [(length) = "<=50"];
// Configuration options for text generation.
GenerationOptions generation_options = 2;
// Text precondition or context of the request.
// For example, the instruction may be "You are a helpful assistant".
oneof Instruction {
// The text-based instruction for the conversation.
string instruction_text = 3;
}
// A list of messages in the conversation.
repeated Message messages = 4;
}
// Contains a model-generated response for a chat query.
message ChatResponse {
// The assistant's message in the chat conversation.
Message message = 1;
// Total number of tokens used in both the chat request and chat response.
int64 num_tokens = 2;
}
// Service for text generation and conversation.
service TextGenerationService {
// RPC method for instructing the model to generate text.
rpc Instruct (InstructRequest) returns (stream InstructResponse) {
option (google.api.http) = {post: "/llm/v1alpha/instruct" body: "*"};
}
// RPC method for engaging in a chat conversation with the model.
rpc Chat (ChatRequest) returns (stream ChatResponse) {
option (google.api.http) = {post: "/llm/v1alpha/chat" body: "*"};
}
}
// Request to tokenize input text.
message TokenizeRequest {
// The name or identifier of the model to be used for tokenization.
// Possible values for now: `general`, `general:embedding`.
string model = 1 [(length) = "<=50"];
// The input text to tokenize.
string text = 2;
}
// Tokenization response.
message TokenizeResponse {
// A list of tokens obtained from tokenization.
repeated Token tokens = 1;
}
// Service for tokenizing input text.
service TokenizerService {
// RPC method for tokenizing input text.
rpc Tokenize (TokenizeRequest) returns (TokenizeResponse) {
option (google.api.http) = {post: "/llm/v1alpha/tokenize" body: "*"};
}
}
// Represents a request to obtain embeddings for text data.
message EmbeddingRequest {
// Enum to specify the type of embedding to be generated.
enum EmbeddingType {
// Unspecified embedding type.
EMBEDDING_TYPE_UNSPECIFIED = 0;
// Embedding for a query. Use this when you have a short query or search term
// that you want to obtain an embedding for. Query embeddings are typically
// used in information retrieval and search applications.
EMBEDDING_TYPE_QUERY = 1;
// Embedding for a document. Use this when you have a longer document or a piece
// of text that you want to obtain an embedding for. Document embeddings are often
// used in natural language understanding and document similarity tasks.
EMBEDDING_TYPE_DOCUMENT = 2;
}
// The type of embedding to be generated.
EmbeddingType embedding_type = 1;
// The name or identifier of the model to be used for embedding. Possible value for now: `general:embedding`.
string model = 2 [(length) = "<=50"];
// The input text for which the embedding is requested.
string text = 3;
}
// Represents a response containing embeddings for input text data.
message EmbeddingResponse {
// A repeated list of double values representing the embedding.
repeated double embedding = 1;
// The number of tokens in the input text.
int64 num_tokens = 2;
}
// Service for obtaining embeddings for text data.
service EmbeddingsService {
// RPC method to obtain embeddings for input text data.
rpc Embedding (EmbeddingRequest) returns (EmbeddingResponse) {
option (google.api.http) = {post: "/llm/v1alpha/embedding" body: "*"};
}
}
// Service for asynchronous text generation.
service TextGenerationAsyncService {
// RPC method for instructing the model to generate text.
rpc Instruct (InstructRequest) returns (operation.Operation) {
option (yandex.cloud.api.operation) = {
response: "InstructResponse"
};
option (google.api.http) = {post: "/llm/v1alpha/instructAsync" body: "*"};
}
}
|
