cognee-llm 0.1.0

LLM client abstraction (OpenAI-compatible) for the cognee AI-memory pipeline.
Documentation 
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

//! Common types for LLM operations.

use serde::{Deserialize, Serialize};

/// Message role in a conversation.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
#[serde(rename_all = "lowercase")]
pub enum MessageRole {
    System,
    User,
    Assistant,
}

/// A message in a conversation.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct Message {
    pub role: MessageRole,
    pub content: String,
}

impl Message {
    pub fn system(content: impl Into<String>) -> Self {
        Self {
            role: MessageRole::System,
            content: content.into(),
        }
    }

    pub fn user(content: impl Into<String>) -> Self {
        Self {
            role: MessageRole::User,
            content: content.into(),
        }
    }

    pub fn assistant(content: impl Into<String>) -> Self {
        Self {
            role: MessageRole::Assistant,
            content: content.into(),
        }
    }
}

/// Options for LLM generation.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct GenerationOptions {
    /// Temperature for sampling (0.0 = deterministic, 1.0 = creative).
    #[serde(skip_serializing_if = "Option::is_none")]
    pub temperature: Option<f32>,

    /// Maximum number of tokens to generate.
    #[serde(skip_serializing_if = "Option::is_none")]
    pub max_tokens: Option<u32>,

    /// Top-p sampling parameter.
    #[serde(skip_serializing_if = "Option::is_none")]
    pub top_p: Option<f32>,

    /// Frequency penalty.
    #[serde(skip_serializing_if = "Option::is_none")]
    pub frequency_penalty: Option<f32>,

    /// Presence penalty.
    #[serde(skip_serializing_if = "Option::is_none")]
    pub presence_penalty: Option<f32>,

    /// Stop sequences.
    #[serde(skip_serializing_if = "Option::is_none")]
    pub stop: Option<Vec<String>>,
}

impl Default for GenerationOptions {
    fn default() -> Self {
        Self {
            temperature: Some(0.0),
            max_tokens: Some(16384),
            top_p: None,
            frequency_penalty: None,
            presence_penalty: None,
            stop: None,
        }
    }
}

/// Response from LLM generation.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct GenerationResponse {
    /// Generated text content.
    pub content: String,

    /// Model used for generation.
    pub model: String,

    /// Token usage information.
    pub usage: Option<TokenUsage>,

    /// Finish reason (e.g., "stop", "length", "content_filter").
    pub finish_reason: Option<String>,
}

/// Token usage statistics.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TokenUsage {
    pub prompt_tokens: u32,
    pub completion_tokens: u32,
    pub total_tokens: u32,
}