gpt4free/g4f/Provider/HuggingChat.py

143 lines
5.3 KiB
Python
Raw Normal View History

from __future__ import annotations
2024-06-26 11:55:03 +00:00
import json, requests, re
2024-06-26 11:55:03 +00:00
from curl_cffi import requests as cf_reqs
from ..typing import CreateResult, Messages
from .base_provider import ProviderModelMixin, AbstractProvider
2024-06-29 14:00:50 +00:00
from .helper import format_prompt
2024-06-26 11:55:03 +00:00
class HuggingChat(AbstractProvider, ProviderModelMixin):
url = "https://huggingface.co/chat"
working = True
supports_stream = True
default_model = "mistralai/Mixtral-8x7B-Instruct-v0.1"
2024-01-23 04:02:14 +00:00
models = [
'meta-llama/Meta-Llama-3.1-70B-Instruct',
'meta-llama/Meta-Llama-3.1-405B-Instruct-FP8',
2024-04-13 02:09:11 +00:00
'CohereForAI/c4ai-command-r-plus',
'mistralai/Mixtral-8x7B-Instruct-v0.1',
'NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO',
2024-07-08 20:49:38 +00:00
'01-ai/Yi-1.5-34B-Chat',
'mistralai/Mistral-7B-Instruct-v0.2',
'microsoft/Phi-3-mini-4k-instruct',
2024-01-23 04:02:14 +00:00
]
2024-06-26 11:55:03 +00:00
model_aliases = {
"mistralai/Mistral-7B-Instruct-v0.1": "mistralai/Mistral-7B-Instruct-v0.2"
}
@classmethod
2024-06-26 11:55:03 +00:00
def create_completion(
cls,
model: str,
messages: Messages,
2024-06-26 11:55:03 +00:00
stream: bool,
**kwargs
2024-06-26 11:55:03 +00:00
) -> CreateResult:
if (model in cls.models) :
session = requests.Session()
headers = {
2024-06-29 14:00:50 +00:00
'accept' : '*/*',
'accept-language' : 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
'cache-control' : 'no-cache',
'origin' : 'https://huggingface.co',
'pragma' : 'no-cache',
'priority' : 'u=1, i',
'referer' : 'https://huggingface.co/chat/',
'sec-ch-ua' : '"Not/A)Brand";v="8", "Chromium";v="126", "Google Chrome";v="126"',
'sec-ch-ua-mobile' : '?0',
2024-06-26 11:55:03 +00:00
'sec-ch-ua-platform': '"macOS"',
2024-06-29 14:00:50 +00:00
'sec-fetch-dest' : 'empty',
'sec-fetch-mode' : 'cors',
'sec-fetch-site' : 'same-origin',
'user-agent' : 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
2024-06-26 11:55:03 +00:00
}
2024-04-13 02:09:11 +00:00
2024-06-26 11:55:03 +00:00
json_data = {
2024-06-29 14:00:50 +00:00
'searchEnabled' : True,
'activeModel' : 'CohereForAI/c4ai-command-r-plus', # doesn't matter
2024-06-26 11:55:03 +00:00
'hideEmojiOnSidebar': False,
2024-06-29 14:00:50 +00:00
'customPrompts' : {},
'assistants' : [],
'tools' : {},
'disableStream' : False,
'recentlySaved' : False,
'ethicsModalAccepted' : True,
'ethicsModalAcceptedAt' : None,
2024-06-26 11:55:03 +00:00
'shareConversationsWithModelAuthors': False,
}
response = cf_reqs.post('https://huggingface.co/chat/settings', headers=headers, json=json_data)
session.cookies.update(response.cookies)
response = session.post('https://huggingface.co/chat/conversation',
headers=headers, json={'model': model})
conversationId = response.json()['conversationId']
response = session.get(f'https://huggingface.co/chat/conversation/{conversationId}/__data.json?x-sveltekit-invalidated=11',
headers=headers,
)
messageId = extract_id(response.json())
settings = {
2024-06-29 14:00:50 +00:00
"inputs" : format_prompt(messages),
"id" : messageId,
"is_retry" : False,
"is_continue" : False,
"web_search" : False,
2024-06-26 11:55:03 +00:00
2024-06-29 14:00:50 +00:00
# TODO // add feature to enable/disable tools
2024-06-26 11:55:03 +00:00
"tools": {
2024-06-29 14:00:50 +00:00
"websearch" : True,
"document_parser" : False,
"query_calculator" : False,
"image_generation" : False,
"image_editing" : False,
"fetch_url" : False,
2024-06-26 11:55:03 +00:00
}
}
payload = {
"data": json.dumps(settings),
}
2024-04-22 18:02:17 +00:00
2024-06-26 11:55:03 +00:00
response = session.post(f"https://huggingface.co/chat/conversation/{conversationId}",
headers=headers, data=payload, stream=True,
)
first_token = True
for line in response.iter_lines():
line = json.loads(line)
if "type" not in line:
raise RuntimeError(f"Response: {line}")
elif line["type"] == "stream":
token = line["token"]
if first_token:
token = token.lstrip().replace('\u0000', '')
first_token = False
2024-06-29 14:00:50 +00:00
else:
token = token.replace('\u0000', '')
yield token
2024-06-26 11:55:03 +00:00
elif line["type"] == "finalAnswer":
break
def extract_id(response: dict) -> str:
data = response["nodes"][1]["data"]
uuid_pattern = re.compile(
r"^[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}$"
)
for item in data:
if type(item) == str and uuid_pattern.match(item):
return item
return None