package orca

Get desktop application:
View/edit binary Protocol Buffers messages

• rpc Generate (V1CompletionsRequest, stream V1CompletionsResponse)

  completions.proto:94

  message V1CompletionsRequest

  completions.proto:19

  • optional bool stream = 1

  • optional string model = 57

  • optional string prompt = 3

  • repeated int32 tokens = 4

  • optional int32 timeout_microseconds = 5

  • optional int32 max_tokens = 6

  • optional int32 max_total_tokens = 7

  • optional int32 min_tokens = 8

  • optional int32 min_total_tokens = 9

  • optional int32 n = 10

  • optional int32 num_beams = 11

  • optional float length_penalty = 12

  • optional bool early_stopping = 15

  • optional int32 no_repeat_ngram = 17

  • optional int32 encoder_no_repeat_ngram = 18

  • optional float repetition_penalty = 19

  • optional float encoder_repetition_penalty = 34

  • optional float frequency_penalty = 53

  • optional float presence_penalty = 54

  • optional float temperature = 20

  • optional int32 top_k = 21

  • optional float top_p = 22

  • repeated string stop = 23

  • repeated V1CompletionsRequest.TokenSequence stop_tokens = 24

  • repeated uint64 seed = 26

  • repeated int32 token_index_to_replace = 27

  • repeated float embedding_to_replace = 28

  • optional V1CompletionsRequest.BeamSearchType beam_search_type = 29

  • optional bool beam_compat_pre_normalization = 30

  • optional bool beam_compat_no_post_normalization = 31

  • repeated string bad_words = 32

  • repeated V1CompletionsRequest.TokenSequence bad_word_tokens = 33

  • optional bool include_output_logits = 47

  • optional bool include_output_logprobs = 50

  • repeated int32 forced_output_tokens = 51

  • repeated int32 eos_token = 46

  • optional ResponseFormat response_format = 61

  • repeated SoftPrompt soft_prompts = 68

  • repeated int32 position_ids = 70

  • optional int32 generation_offset = 71

  message V1CompletionsResponse

  completions.proto:82

  • V1CompletionsResponse.Event event = 1

  • repeated int32 token = 2

  • optional string text = 3

  • repeated uint64 soft_prompt_ids = 5

Used in: V1ChatCompletionsRequest

• optional string content = 1

• string role = 2

• optional string name = 3

• optional string tool_call_id = 4

• repeated ToolCall tool_calls = 5

Used in: V1ChatCompletionsRequest, V1CompletionsRequest

• ResponseFormat.Type type = 1

• optional string schema = 2

  Json schema or regex

Used in: ResponseFormat

• text = 0

• json_object = 1

• regex = 2

Used in: V1CompletionsRequest

• int32 token_index_start = 1

• int32 token_index_end = 2

• repeated float embeddings = 3

  embeddings of length end - start, or empty if uuid is provided

• optional uint64 id = 4

Used in: V1ChatCompletionsRequest

• string type = 1

• optional Tool.Function function = 2

Used in: Tool

• string name = 1

• optional string description = 2

• optional google.protobuf.Struct parameters = 3

  Json schema

Used in: Message

• string id = 1

• string type = 2

• optional ToolCall.Function function = 3

Used in: ToolCall

• string name = 1

• string arguments = 2

• repeated Message messages = 1

• optional string model = 2

• optional float frequency_penalty = 3

• map<int32, float> logit_bias = 4

• optional int32 min_tokens = 5

• optional int32 max_tokens = 6

• optional int32 n = 7

• optional float presence_penalty = 8

• repeated string stop = 9

• optional bool stream = 10

• optional float temperature = 11

• optional float top_p = 12

• optional int32 timeout_microseconds = 13

• optional bool logprobs = 14

• optional int32 top_logprobs = 15

• optional int32 top_k = 19

• optional float repetition_penalty = 20

• repeated uint64 seed = 21

• repeated int32 eos_token = 22

• repeated Tool tools = 23

• optional ResponseFormat response_format = 24

• optional google.protobuf.Value tool_choice = 25

  "auto", "none", "required" or {"type": "function", "function": {"name": "my_function"}}

• optional bool parallel_tool_calls = 26

Used in: V1CompletionsRequest

• DETERMINISTIC = 0

  Use the standard beam search

• STOCHASTIC = 1

  Stochastic beam search by Kool et al. (2019)

• NAIVE_SAMPLING = 2

  Huggingface's beam sampling

Used in: V1CompletionsRequest

• repeated int32 tokens = 1

Used in: V1CompletionsResponse

• TOKEN_SAMPLED = 0

• COMPLETE = 1