File size: 6,882 Bytes
c120a1c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
import { t } from './i18n.js';

// the hash can be obtained from command line e.g. via: MODEL=path_to_model; python -c "import json, hashlib, sys; print(hashlib.sha256(json.load(open('"$MODEL"/tokenizer_config.json'))['chat_template'].encode()).hexdigest())"
// note that chat templates must be trimmed to match the llama.cpp metadata value
const hash_derivations = {
    // Meta
    'e10ca381b1ccc5cf9db52e371f3b6651576caee0a630b452e2816b2d404d4b65':
        // Meta-Llama-3.1-8B-Instruct
        // Meta-Llama-3.1-70B-Instruct
        'Llama 3 Instruct'
    ,
    '5816fce10444e03c2e9ee1ef8a4a1ea61ae7e69e438613f3b17b69d0426223a4':
        // Llama-3.2-1B-Instruct
        // Llama-3.2-3B-Instruct
        'Llama 3 Instruct'
    ,
    '73e87b1667d87ab7d7b579107f01151b29ce7f3ccdd1018fdc397e78be76219d':
        // Nemotron 70B
        'Llama 3 Instruct'
    ,

    // Mistral
    // Mistral Reference: https://github.com/mistralai/mistral-common
    'e16746b40344d6c5b5265988e0328a0bf7277be86f1c335156eae07e29c82826':
        // Mistral-Small-Instruct-2409
        // Mistral-Large-Instruct-2407
        'Mistral V2 & V3'
    ,
    '26a59556925c987317ce5291811ba3b7f32ec4c647c400c6cc7e3a9993007ba7':
        // Mistral-7B-Instruct-v0.3
        'Mistral V2 & V3'
    ,
    'e4676cb56dffea7782fd3e2b577cfaf1e123537e6ef49b3ec7caa6c095c62272':
        // Mistral-Nemo-Instruct-2407
        'Mistral V3-Tekken'
    ,
    '3c4ad5fa60dd8c7ccdf82fa4225864c903e107728fcaf859fa6052cb80c92ee9':
        // Mistral-Large-Instruct-2411
        'Mistral V7'
    ,
    '3934d199bfe5b6fab5cba1b5f8ee475e8d5738ac315f21cb09545b4e665cc005':
        // Mistral Small 24B
        'Mistral V7'
    ,

    // Gemma
    'ecd6ae513fe103f0eb62e8ab5bfa8d0fe45c1074fa398b089c93a7e70c15cfd6':
        // gemma-2-9b-it
        // gemma-2-27b-it
        'Gemma 2'
    ,
    '87fa45af6cdc3d6a9e4dd34a0a6848eceaa73a35dcfe976bd2946a5822a38bf3':
        // gemma-2-2b-it
        'Gemma 2'
    ,
    '7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4':
        // gemma-3
        'Gemma 2'
    ,

    // Cohere
    '3b54f5c219ae1caa5c0bb2cdc7c001863ca6807cf888e4240e8739fa7eb9e02e':
        // command-r-08-2024
        'Command R'
    ,

    // Tulu
    'ac7498a36a719da630e99d48e6ebc4409de85a77556c2b6159eeb735bcbd11df':
        // Tulu-3-8B
        // Tulu-3-70B
        'Tulu'
    ,

    // DeepSeek V2.5
    '54d400beedcd17f464e10063e0577f6f798fa896266a912d8a366f8a2fcc0bca':
        'DeepSeek-V2.5'
    ,

    // DeepSeek R1
    'b6835114b7303ddd78919a82e4d9f7d8c26ed0d7dfc36beeb12d524f6144eab1':
        'DeepSeek-V2.5'
    ,

    // THUDM-GLM 4
    '854b703e44ca06bdb196cc471c728d15dbab61e744fe6cdce980086b61646ed1':
        'GLM-4'
    ,

    // Kimi K2, ...
    'aab20feb9bc6881f941ea649356130ffbc4943b3c2577c0991e1fba90de5a0fc':
        'Moonshot AI'
    ,

    // gpt-oss (unsloth)
    '70da0d2348e40aaf8dad05f04a316835fd10547bd7e3392ce337e4c79ba91c01':
        'OpenAI Harmony'
    ,

    // gpt-oss (ggml-org)
    'a4c9919cbbd4acdd51ccffe22da049264b1b73e59055fa58811a99efbd7c8146':
        'OpenAI Harmony'
    ,
};

const substr_derivations = [
    ['Moonshot AI', ['<|im_user|>user<|im_middle|>', '<|im_assistant|>assistant<|im_middle|>', '<|im_end|>']],
    ['OpenAI Harmony', ['<|start|>user<|message|>', '<|start|>assistant<|channel|>final<|message|>', '<|end|>']],

    // Generic cases
    ['ChatML', ['<|im_start|>user', '<|im_start|>assistant', '<|im_end|>']],
];

const parse_derivation = derivation => (typeof derivation === 'string') ? {
    'context': derivation,
    'instruct': derivation,
} : derivation;

const not_found = { context: null, instruct: null };

export async function deriveTemplatesFromChatTemplate(chat_template, hash) {
    if (chat_template.trim() === '') {
        console.log('Missing chat template.');
        return not_found;
    }

    if (hash in hash_derivations) {
        return parse_derivation(hash_derivations[hash]);
    }

    // heuristics
    for (const [derivation, substr] of substr_derivations) {
        if ([substr].flat().every(str => chat_template.includes(str))) {
            return parse_derivation(derivation);
        }
    }

    console.warn(`Unknown chat template hash: ${hash} for [${chat_template}]`);
    return not_found;
}

export async function bindModelTemplates(power_user, online_status) {
    if (online_status === 'no_connection') {
        return false;
    }

    const chatTemplateHash = power_user.chat_template_hash;
    const bindModelTemplates = power_user.model_templates_mappings[online_status]
        ?? power_user.model_templates_mappings[chatTemplateHash]
        ?? {};
    const bindingsMatch = bindModelTemplates
        && power_user.context.preset == bindModelTemplates['context']
        && (!power_user.instruct.enabled || power_user.instruct.preset === bindModelTemplates['instruct']);

    const bound = [];

    if (bindingsMatch) {
        // unmap current preset
        delete power_user.model_templates_mappings[chatTemplateHash];
        delete power_user.model_templates_mappings[online_status];
        toastr.info(t`Context preset for ${online_status} will use defaults when loaded the next time.`);
    } else {
        if (power_user.context_derived) {
            if (power_user.context.preset !== bindModelTemplates['context']) {
                bound.push(`${power_user.context.preset} context preset`);
                // toastr.info(`Bound ${power_user.context.preset} preset to currently loaded model and all models that share its chat template.`);

                // map current preset to current chat template hash
                bindModelTemplates['context'] = power_user.context.preset;
            }
        } else {
            toastr.warning(t`Note: Context derivation is disabled. Not including context preset.`);
        }
        if (power_user.instruct.enabled) {
            if (power_user.instruct_derived) {
                if (power_user.instruct.preset !== bindModelTemplates['instruct']) {
                    bound.push(`${power_user.instruct.preset} instruct preset`);
                    bindModelTemplates['instruct'] = power_user.instruct.preset;
                }
            } else {
                toastr.warning(t`Note: Instruct derivation is disabled. Not including instruct preset.`);
            }
        }
        if (bound.length == 0) {
            toastr.warning(t`No applicable presets available.`);
            return false;
        }

        toastr.info(t`Bound ${online_status} to ${bound.join(', ')}.`);
        if (!online_status.startsWith('koboldcpp/ggml-model-')) {
            power_user.model_templates_mappings[online_status] = bindModelTemplates;
        }
        if (chatTemplateHash !== '') {
            power_user.model_templates_mappings[chatTemplateHash] = bindModelTemplates;
        }
    }

    return true;
}