File size: 17,584 Bytes
355bae1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5684fd5
355bae1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5684fd5
355bae1
 
5684fd5
355bae1
 
5684fd5
355bae1
 
 
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
import os
import json
import base64
import quickjs
import queue
import threading

from curl_cffi import requests
from ...typing import sha256, Dict, get_type_hints

url = 'https://play.vercel.ai'
model = None
supports_stream = True

models = {
    'claude-instant-v1': 'anthropic:claude-instant-v1',
    'claude-v1': 'anthropic:claude-v1',
    'alpaca-7b': 'replicate:replicate/alpaca-7b',
    'stablelm-tuned-alpha-7b': 'replicate:stability-ai/stablelm-tuned-alpha-7b',
    'bloom': 'huggingface:bigscience/bloom',
    'bloomz': 'huggingface:bigscience/bloomz',
    'flan-t5-xxl': 'huggingface:google/flan-t5-xxl',
    'flan-ul2': 'huggingface:google/flan-ul2',
    'gpt-neox-20b': 'huggingface:EleutherAI/gpt-neox-20b',
    'oasst-sft-4-pythia-12b-epoch-3.5': 'huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5',
    'santacoder': 'huggingface:bigcode/santacoder',
    'command-medium-nightly': 'cohere:command-medium-nightly',
    'command-xlarge-nightly': 'cohere:command-xlarge-nightly',
    'gpt-4': 'openai:gpt-4',
    'code-cushman-001': 'openai:code-cushman-001',
    'code-davinci-002': 'openai:code-davinci-002',
    'gpt-3.5-turbo': 'openai:gpt-3.5-turbo',
    'text-ada-001': 'openai:text-ada-001',
    'text-babbage-001': 'openai:text-babbage-001',
    'text-curie-001': 'openai:text-curie-001',
    'text-davinci-002': 'openai:text-davinci-002',
    'text-davinci-003': 'openai:text-davinci-003'
}

vercel_models = {'anthropic:claude-instant-v1': {'id': 'anthropic:claude-instant-v1', 'provider': 'anthropic', 'providerHumanName': 'Anthropic', 'makerHumanName': 'Anthropic', 'minBillingTier': 'hobby', 'parameters': {'temperature': {'value': 1, 'range': [0, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'topK': {'value': 1, 'range': [1, 500]}, 'presencePenalty': {'value': 1, 'range': [0, 1]}, 'frequencyPenalty': {'value': 1, 'range': [0, 1]}, 'stopSequences': {'value': ['\n\nHuman:'], 'range': []}}, 'name': 'claude-instant-v1'}, 'anthropic:claude-v1': {'id': 'anthropic:claude-v1', 'provider': 'anthropic', 'providerHumanName': 'Anthropic', 'makerHumanName': 'Anthropic', 'minBillingTier': 'hobby', 'parameters': {'temperature': {'value': 1, 'range': [0, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'topK': {'value': 1, 'range': [1, 500]}, 'presencePenalty': {'value': 1, 'range': [0, 1]}, 'frequencyPenalty': {'value': 1, 'range': [0, 1]}, 'stopSequences': {'value': ['\n\nHuman:'], 'range': []}}, 'name': 'claude-v1'}, 'replicate:replicate/alpaca-7b': {'id': 'replicate:replicate/alpaca-7b', 'provider': 'replicate', 'providerHumanName': 'Replicate', 'makerHumanName': 'Stanford', 'parameters': {'temperature': {'value': 0.75, 'range': [0.01, 5]}, 'maximumLength': {'value': 200, 'range': [50, 512]}, 'topP': {'value': 0.95, 'range': [0.01, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'repetitionPenalty': {'value': 1.1765, 'range': [0.01, 5]}, 'stopSequences': {'value': [], 'range': []}}, 'version': '2014ee1247354f2e81c0b3650d71ca715bc1e610189855f134c30ecb841fae21', 'name': 'alpaca-7b'}, 'replicate:stability-ai/stablelm-tuned-alpha-7b': {'id': 'replicate:stability-ai/stablelm-tuned-alpha-7b', 'provider': 'replicate', 'makerHumanName': 'StabilityAI', 'providerHumanName': 'Replicate', 'parameters': {'temperature': {'value': 0.75, 'range': [0.01, 5]}, 'maximumLength': {'value': 200, 'range': [50, 512]}, 'topP': {'value': 0.95, 'range': [0.01, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'repetitionPenalty': {'value': 1.1765, 'range': [0.01, 5]}, 'stopSequences': {'value': [], 'range': []}}, 'version': '4a9a32b4fd86c2d047f1d271fa93972683ec6ef1cf82f402bd021f267330b50b', 'name': 'stablelm-tuned-alpha-7b'}, 'huggingface:bigscience/bloom': {'id': 'huggingface:bigscience/bloom', 'provider': 'huggingface', 'providerHumanName': 'HuggingFace', 'makerHumanName': 'BigScience', 'instructions': "Do NOT talk to Bloom as an entity, it's not a chatbot but a webpage/blog/article completion model. For the best results: mimic a few words of a webpage similar to the content you want to generate. Start a sentence as if YOU were writing a blog, webpage, math post, coding article and Bloom will generate a coherent follow-up.", 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 0.95, 'range': [0.01, 0.99]}, 'topK': {'value': 4, 'range': [1, 500]}, 'repetitionPenalty': {'value': 1.03, 'range': [0.1, 2]}}, 'name': 'bloom'}, 'huggingface:bigscience/bloomz': {'id': 'huggingface:bigscience/bloomz', 'provider': 'huggingface', 'providerHumanName': 'HuggingFace', 'makerHumanName': 'BigScience', 'instructions': 'We recommend using the model to perform tasks expressed in natural language. For example, given the prompt "Translate to English: Je t\'aime.", the model will most likely answer "I love you.".', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 0.95, 'range': [0.01, 0.99]}, 'topK': {'value': 4, 'range': [1, 500]}, 'repetitionPenalty': {'value': 1.03, 'range': [0.1, 2]}}, 'name': 'bloomz'}, 'huggingface:google/flan-t5-xxl': {'id': 'huggingface:google/flan-t5-xxl', 'provider': 'huggingface', 'makerHumanName': 'Google', 'providerHumanName': 'HuggingFace', 'name': 'flan-t5-xxl', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 0.95, 'range': [0.01, 0.99]}, 'topK': {'value': 4, 'range': [1, 500]}, 'repetitionPenalty': {'value': 1.03, 'range': [0.1, 2]}}}, 'huggingface:google/flan-ul2': {'id': 'huggingface:google/flan-ul2', 'provider': 'huggingface', 'providerHumanName': 'HuggingFace', 'makerHumanName': 'Google', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 0.95, 'range': [0.01, 0.99]}, 'topK': {'value': 4, 'range': [1, 500]}, 'repetitionPenalty': {'value': 1.03, 'range': [0.1, 2]}}, 'name': 'flan-ul2'}, 'huggingface:EleutherAI/gpt-neox-20b': {'id': 'huggingface:EleutherAI/gpt-neox-20b', 'provider': 'huggingface', 'providerHumanName': 'HuggingFace', 'makerHumanName': 'EleutherAI', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 0.95, 'range': [0.01, 0.99]}, 'topK': {'value': 4, 'range': [1, 500]}, 'repetitionPenalty': {'value': 1.03, 'range': [0.1, 2]}, 'stopSequences': {'value': [], 'range': []}}, 'name': 'gpt-neox-20b'}, 'huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5': {'id': 'huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5', 'provider': 'huggingface', 'providerHumanName': 'HuggingFace', 'makerHumanName': 'OpenAssistant', 'parameters': {'maximumLength': {'value': 200, 'range': [50, 1024]}, 'typicalP': {'value': 0.2, 'range': [0.1, 0.99]}, 'repetitionPenalty': {'value': 1, 'range': [0.1, 2]}}, 'name': 'oasst-sft-4-pythia-12b-epoch-3.5'}, 'huggingface:bigcode/santacoder': {
    'id': 'huggingface:bigcode/santacoder', 'provider': 'huggingface', 'providerHumanName': 'HuggingFace', 'makerHumanName': 'BigCode', 'instructions': 'The model was trained on GitHub code. As such it is not an instruction model and commands like "Write a function that computes the square root." do not work well. You should phrase commands like they occur in source code such as comments (e.g. # the following function computes the sqrt) or write a function signature and docstring and let the model complete the function body.', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 0.95, 'range': [0.01, 0.99]}, 'topK': {'value': 4, 'range': [1, 500]}, 'repetitionPenalty': {'value': 1.03, 'range': [0.1, 2]}}, 'name': 'santacoder'}, 'cohere:command-medium-nightly': {'id': 'cohere:command-medium-nightly', 'provider': 'cohere', 'providerHumanName': 'Cohere', 'makerHumanName': 'Cohere', 'name': 'command-medium-nightly', 'parameters': {'temperature': {'value': 0.9, 'range': [0, 2]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0, 1]}, 'topK': {'value': 0, 'range': [0, 500]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'cohere:command-xlarge-nightly': {'id': 'cohere:command-xlarge-nightly', 'provider': 'cohere', 'providerHumanName': 'Cohere', 'makerHumanName': 'Cohere', 'name': 'command-xlarge-nightly', 'parameters': {'temperature': {'value': 0.9, 'range': [0, 2]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0, 1]}, 'topK': {'value': 0, 'range': [0, 500]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'openai:gpt-4': {'id': 'openai:gpt-4', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'name': 'gpt-4', 'minBillingTier': 'pro', 'parameters': {'temperature': {'value': 0.7, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'openai:code-cushman-001': {'id': 'openai:code-cushman-001', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}, 'name': 'code-cushman-001'}, 'openai:code-davinci-002': {'id': 'openai:code-davinci-002', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}, 'name': 'code-davinci-002'}, 'openai:gpt-3.5-turbo': {'id': 'openai:gpt-3.5-turbo', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'parameters': {'temperature': {'value': 0.7, 'range': [0, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'topK': {'value': 1, 'range': [1, 500]}, 'presencePenalty': {'value': 1, 'range': [0, 1]}, 'frequencyPenalty': {'value': 1, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}, 'name': 'gpt-3.5-turbo'}, 'openai:text-ada-001': {'id': 'openai:text-ada-001', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'name': 'text-ada-001', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'openai:text-babbage-001': {'id': 'openai:text-babbage-001', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'name': 'text-babbage-001', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'openai:text-curie-001': {'id': 'openai:text-curie-001', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'name': 'text-curie-001', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'openai:text-davinci-002': {'id': 'openai:text-davinci-002', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'name': 'text-davinci-002', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}, 'openai:text-davinci-003': {'id': 'openai:text-davinci-003', 'provider': 'openai', 'providerHumanName': 'OpenAI', 'makerHumanName': 'OpenAI', 'name': 'text-davinci-003', 'parameters': {'temperature': {'value': 0.5, 'range': [0.1, 1]}, 'maximumLength': {'value': 200, 'range': [50, 1024]}, 'topP': {'value': 1, 'range': [0.1, 1]}, 'presencePenalty': {'value': 0, 'range': [0, 1]}, 'frequencyPenalty': {'value': 0, 'range': [0, 1]}, 'stopSequences': {'value': [], 'range': []}}}}


# based on https://github.com/ading2210/vercel-llm-api // modified
class Client:
    def __init__(self):
        self.session = requests.Session(impersonate='chrome110')
        self.headers = {
            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110 Safari/537.36',
            'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8',
            'Accept-Encoding': 'gzip, deflate, br',
            'Accept-Language': 'en-US,en;q=0.5',
            'Te': 'trailers',
            'Upgrade-Insecure-Requests': '1'
        }
        self.session.headers.update(self.headers)

    def get_token(self):
        b64 = self.session.get('https://play.vercel.ai/openai.jpeg').text
        data = json.loads(base64.b64decode(b64))

        script = 'const globalThis = {data: `sentinel`}; (%s)(%s)' % (
            data['c'], data['a'])

        token_data = json.loads(quickjs.Context().eval(script).json())
        token_string = json.dumps(separators=(',', ':'),
                                  obj={'r': token_data, 't': data['t']})

        return base64.b64encode(token_string.encode()).decode()

    def get_default_params(self, model_id):
        return {key: param['value'] for key, param in vercel_models[model_id]['parameters'].items()}

    def generate(self, model_id: str, prompt: str, params: dict = {}):
        if ':' not in model_id:
            model_id = models[model_id]

        defaults = self.get_default_params(model_id)

        payload = defaults | params | {
            'prompt': prompt,
            'model': model_id,
        }

        headers = self.headers | {
            'Accept-Encoding': 'gzip, deflate, br',
            'Custom-Encoding': self.get_token(),
            'Host': 'play.vercel.ai',
            'Origin': 'https://play.vercel.ai',
            'Referrer': 'https://play.vercel.ai',
            'Sec-Fetch-Dest': 'empty',
            'Sec-Fetch-Mode': 'cors',
            'Sec-Fetch-Site': 'same-origin',
        }

        chunks_queue = queue.Queue()
        error = None
        response = None

        def callback(data):
            chunks_queue.put(data.decode())

        def request_thread():
            nonlocal response, error
            for _ in range(3):
                try:
                    response = self.session.post('https://play.vercel.ai/api/generate',
                                                 json=payload, headers=headers, content_callback=callback)
                    response.raise_for_status()

                except Exception as e:
                    if _ == 2:
                        error = e

                    else:
                        continue

        thread = threading.Thread(target=request_thread, daemon=True)
        thread.start()

        text = ''
        index = 0
        while True:
            try:
                chunk = chunks_queue.get(block=True, timeout=0.1)

            except queue.Empty:
                if error:
                    raise error

                elif response:
                    break

                else:
                    continue

            text += chunk
            lines = text.split('\n')

            if len(lines) - 1 > index:
                new = lines[index:-1]
                for word in new:
                    yield json.loads(word)
                index = len(lines) - 1

def _create_completion(model: str, messages: list, stream: bool, **kwargs):
    conversation = 'This is a conversation between a human and a language model, respond to the last message accordingly, referring to the past history of messages if needed.\n'

    for message in messages:
        conversation += '%s: %s\n' % (message['role'], message['content'])

    conversation += 'assistant: '

    yield from Client().generate(model, conversation)

params = f'g4f.Providers.{os.path.basename(__file__)[:-3]} supports: ' + \
    '(%s)' % ', '.join([f"{name}: {get_type_hints(_create_completion)[name].__name__}" for name in _create_completion.__code__.co_varnames[:_create_completion.__code__.co_argcount]])