aiha 0.0.4

🦉 AIHA helps you to find the minimal requirements for any model on the 🤗 Hub.
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
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

//! Config metadata struct
use serde::Deserialize;

use crate::models::{
    BertModelConfig, BloomModelConfig, GPT2ModelConfig, GPTJModelConfig, GPTNeoModelConfig,
    LlamaModelConfig, ModelConfigTrait, ModelError, OPTModelConfig, T5ModelConfig,
};

/// Enum all the possible model types
#[derive(Clone, Debug, Deserialize)]
pub enum ModelConfig {
    /// Bert model config
    Bert(BertModelConfig),
    /// Bloom model config
    Bloom(BloomModelConfig),
    /// GPT2 model config
    Gpt2(GPT2ModelConfig),
    /// GPTJ model config
    GptJ(GPTJModelConfig),
    /// GPTNeo model config
    GPTNeo(GPTNeoModelConfig),
    /// Llama model config
    Llama(LlamaModelConfig),
    /// OPT model config
    Opt(OPTModelConfig),
    /// T5 model config
    T5(T5ModelConfig),
}

/// Model config implementation
impl ModelConfigTrait for ModelConfig {
    fn hidden_size(&self) -> i32 {
        match self {
            ModelConfig::Bert(config) => config.hidden_size(),
            ModelConfig::Bloom(config) => config.hidden_size(),
            ModelConfig::Gpt2(config) => config.hidden_size(),
            ModelConfig::GptJ(config) => config.hidden_size(),
            ModelConfig::GPTNeo(config) => config.hidden_size(),
            ModelConfig::Llama(config) => config.hidden_size(),
            ModelConfig::Opt(config) => config.hidden_size(),
            ModelConfig::T5(config) => config.hidden_size(),
        }
    }
    fn intermediate_size(&self) -> i32 {
        match self {
            ModelConfig::Bert(config) => config.intermediate_size(),
            ModelConfig::Bloom(config) => config.intermediate_size(),
            ModelConfig::Gpt2(config) => config.intermediate_size(),
            ModelConfig::GptJ(config) => config.intermediate_size(),
            ModelConfig::GPTNeo(config) => config.intermediate_size(),
            ModelConfig::Llama(config) => config.intermediate_size(),
            ModelConfig::Opt(config) => config.intermediate_size(),
            ModelConfig::T5(config) => config.intermediate_size(),
        }
    }
    fn max_position_embeddings(&self) -> i32 {
        match self {
            ModelConfig::Bert(config) => config.max_position_embeddings(),
            ModelConfig::Bloom(config) => config.max_position_embeddings(),
            ModelConfig::Gpt2(config) => config.max_position_embeddings(),
            ModelConfig::GptJ(config) => config.max_position_embeddings(),
            ModelConfig::GPTNeo(config) => config.max_position_embeddings(),
            ModelConfig::Llama(config) => config.max_position_embeddings(),
            ModelConfig::Opt(config) => config.max_position_embeddings(),
            ModelConfig::T5(config) => config.max_position_embeddings(),
        }
    }
    fn num_attention_heads(&self) -> i32 {
        match self {
            ModelConfig::Bert(config) => config.num_attention_heads(),
            ModelConfig::Bloom(config) => config.num_attention_heads(),
            ModelConfig::Gpt2(config) => config.num_attention_heads(),
            ModelConfig::GptJ(config) => config.num_attention_heads(),
            ModelConfig::GPTNeo(config) => config.num_attention_heads(),
            ModelConfig::Llama(config) => config.num_attention_heads(),
            ModelConfig::Opt(config) => config.num_attention_heads(),
            ModelConfig::T5(config) => config.num_attention_heads(),
        }
    }
    fn num_hidden_layers(&self) -> i32 {
        match self {
            ModelConfig::Bert(config) => config.num_hidden_layers(),
            ModelConfig::Bloom(config) => config.num_hidden_layers(),
            ModelConfig::Gpt2(config) => config.num_hidden_layers(),
            ModelConfig::GptJ(config) => config.num_hidden_layers(),
            ModelConfig::GPTNeo(config) => config.num_hidden_layers(),
            ModelConfig::Llama(config) => config.num_hidden_layers(),
            ModelConfig::Opt(config) => config.num_hidden_layers(),
            ModelConfig::T5(config) => config.num_hidden_layers(),
        }
    }
    fn model_type(&self) -> &str {
        match self {
            ModelConfig::Bert(config) => config.model_type(),
            ModelConfig::Bloom(config) => config.model_type(),
            ModelConfig::Gpt2(config) => config.model_type(),
            ModelConfig::GptJ(config) => config.model_type(),
            ModelConfig::GPTNeo(config) => config.model_type(),
            ModelConfig::Llama(config) => config.model_type(),
            ModelConfig::Opt(config) => config.model_type(),
            ModelConfig::T5(config) => config.model_type(),
        }
    }
    fn available_libraries(&self) -> &[crate::ModelLibraries] {
        match self {
            ModelConfig::Bert(config) => config.available_libraries(),
            ModelConfig::Bloom(config) => config.available_libraries(),
            ModelConfig::Gpt2(config) => config.available_libraries(),
            ModelConfig::GptJ(config) => config.available_libraries(),
            ModelConfig::GPTNeo(config) => config.available_libraries(),
            ModelConfig::Llama(config) => config.available_libraries(),
            ModelConfig::Opt(config) => config.available_libraries(),
            ModelConfig::T5(config) => config.available_libraries(),
        }
    }
    fn from_json(value: serde_json::Value) -> Result<Self, ModelError>
    where
        Self: Sized,
    {
        let model_type = value["model_type"]
            .as_str()
            .ok_or(ModelError::MissingField("model_type".to_string()))?;
        match model_type {
            "bert" => Ok(ModelConfig::Bert(BertModelConfig::from_json(value)?)),
            "bloom" => Ok(ModelConfig::Bloom(BloomModelConfig::from_json(value)?)),
            "gpt2" => Ok(ModelConfig::Gpt2(GPT2ModelConfig::from_json(value)?)),
            "gptj" => Ok(ModelConfig::GptJ(GPTJModelConfig::from_json(value)?)),
            "gpt_neo" => Ok(ModelConfig::GPTNeo(GPTNeoModelConfig::from_json(value)?)),
            "gpt_neox" => Ok(ModelConfig::GPTNeo(GPTNeoModelConfig::from_json(value)?)),
            "llama" => Ok(ModelConfig::Llama(LlamaModelConfig::from_json(value)?)),
            "opt" => Ok(ModelConfig::Opt(OPTModelConfig::from_json(value)?)),
            "t5" => Ok(ModelConfig::T5(T5ModelConfig::from_json(value)?)),
            _ => Err(ModelError::ModelNotImplemented(model_type.to_string())),
        }
    }
}

#[cfg(test)]
mod tests {}