diff options
| -rw-r--r-- | LEGAL_NOTICE.md | 6 | ||||
| -rw-r--r-- | README.md | 21 | ||||
| -rw-r--r-- | g4f/Provider/DeepInfra.py | 63 | ||||
| -rw-r--r-- | g4f/Provider/Llama2.py | 17 | ||||
| -rw-r--r-- | g4f/Provider/__init__.py | 3 | ||||
| -rw-r--r-- | g4f/models.py | 22 |
6 files changed, 111 insertions, 21 deletions
diff --git a/LEGAL_NOTICE.md b/LEGAL_NOTICE.md index 6a5bbab8..910ecd61 100644 --- a/LEGAL_NOTICE.md +++ b/LEGAL_NOTICE.md @@ -4,12 +4,12 @@ This repository is _not_ associated with or endorsed by providers of the APIs co Please note the following: -1. **Disclaimer**: The APIs, services, and trademarks mentioned in this repository belong to their respective owners. This project is _not_ claiming any right over them nor is it affiliated with or endorsed by any of the providers mentioned. +1. **Disclaimer**: The APIs, services, and trademarks mentioned in this repository belong to their respective owners. This project is _not_ claiming any rights over them nor is it affiliated with or endorsed by any of the providers mentioned. -2. **Responsibility**: The author of this repository is _not_ responsible for any consequences, damages, or losses arising from the use or misuse of this repository or the content provided by the third-party APIs. Users are solely responsible for their actions and any repercussions that may follow. We strongly recommend the users to follow the TOS of the each Website. +2. **Responsibility**: The author of this repository is _not_ responsible for any consequences, damages, or losses arising from the use or misuse of this repository or the content provided by the third-party APIs. Users are solely responsible for their actions and any repercussions that may follow. We strongly recommend the users to follow the TOS of each Website. 3. **Educational Purposes Only**: This repository and its content are provided strictly for educational purposes. By using the information and code provided, users acknowledge that they are using the APIs and models at their own risk and agree to comply with any applicable laws and regulations. 4. **Indemnification**: Users agree to indemnify, defend, and hold harmless the author of this repository from and against any and all claims, liabilities, damages, losses, or expenses, including legal fees and costs, arising out of or in any way connected with their use or misuse of this repository, its content, or related third-party APIs. -5. **Updates and Changes**: The author reserves the right to modify, update, or remove any content, information, or features in this repository at any time without prior notice. Users are responsible for regularly reviewing the content and any changes made to this repository.
\ No newline at end of file +5. **Updates and Changes**: The author reserves the right to modify, update, or remove any content, information, or features in this repository at any time without prior notice. Users are responsible for regularly reviewing the content and any changes made to this repository. @@ -11,7 +11,7 @@ pip install -U g4f ``` ## New features -* Telegram Channel: [t.me/g4f_channel](https://t.me/g4f_channel) +* Telegram Channel: [t.me/g4f_channel](https://telegram.me/g4f_channel) * g4f documentation (unfinished): [g4f.mintlify.app](https://g4f.mintlify.app) | Contribute to the docs via: [github.com/xtekky/gpt4free-docs](https://github.com/xtekky/gpt4free-docs) ## Table of Contents @@ -331,9 +331,12 @@ python -m g4f.api ```py import openai -openai.api_key = " Leave Empty if you don't use embeddings, otherwise your Hugging Face token" -openai.api_base = "http://localhost:1337/v1" +# Set your Hugging Face token as the API key if you use embeddings +# If you don't use embeddings, leave it empty +openai.api_key = "YOUR_HUGGING_FACE_TOKEN" # Replace with your actual token +# Set the API base URL if needed, e.g., for a local development environment +openai.api_base = "http://localhost:1337/v1" def main(): chat_completion = openai.ChatCompletion.create( @@ -343,18 +346,18 @@ def main(): ) if isinstance(chat_completion, dict): - # not stream + # Not streaming print(chat_completion.choices[0].message.content) else: - # stream + # Streaming for token in chat_completion: content = token["choices"][0]["delta"].get("content") - if content != None: + if content is not None: print(content, end="", flush=True) - -if __name__ == "__main__": +if __name__ == "__main": main() + ``` ## Models @@ -543,7 +546,7 @@ Call in your terminal the "create_provider" script: python etc/tool/create_provider.py ``` 1. Enter your name for the new provider. -2. Copy and paste a cURL command from your browser developer tools. +2. Copy and paste the `URL` command from your browser developer tools. 3. Let the AI create the provider for you. 4. Customize the provider according to your needs. diff --git a/g4f/Provider/DeepInfra.py b/g4f/Provider/DeepInfra.py new file mode 100644 index 00000000..70045cae --- /dev/null +++ b/g4f/Provider/DeepInfra.py @@ -0,0 +1,63 @@ +from __future__ import annotations + +import json +from aiohttp import ClientSession + +from ..typing import AsyncResult, Messages +from .base_provider import AsyncGeneratorProvider + + +class DeepInfra(AsyncGeneratorProvider): + url = "https://deepinfra.com" + working = True + + @classmethod + async def create_async_generator( + cls, + model: str, + messages: Messages, + proxy: str = None, + **kwargs + ) -> AsyncResult: + if not model: + model = "meta-llama/Llama-2-70b-chat-hf" + headers = { + "User-Agent": "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/118.0", + "Accept": "text/event-stream", + "Accept-Language": "de,en-US;q=0.7,en;q=0.3", + "Accept-Encoding": "gzip, deflate, br", + "Referer": f"{cls.url}/", + "Content-Type": "application/json", + "X-Deepinfra-Source": "web-page", + "Origin": cls.url, + "Connection": "keep-alive", + "Sec-Fetch-Dest": "empty", + "Sec-Fetch-Mode": "cors", + "Sec-Fetch-Site": "same-site", + "Pragma": "no-cache", + "Cache-Control": "no-cache", + } + async with ClientSession(headers=headers) as session: + data = { + "model": model, + "messages": messages, + "stream": True, + } + async with session.post( + "https://api.deepinfra.com/v1/openai/chat/completions", + json=data, + proxy=proxy + ) as response: + response.raise_for_status() + first = True + async for line in response.content: + if line.startswith(b"data: [DONE]"): + break + elif line.startswith(b"data: "): + chunk = json.loads(line[6:])["choices"][0]["delta"].get("content") + if chunk: + if first: + chunk = chunk.lstrip() + if chunk: + first = False + yield chunk
\ No newline at end of file diff --git a/g4f/Provider/Llama2.py b/g4f/Provider/Llama2.py index b59fde12..1b332f86 100644 --- a/g4f/Provider/Llama2.py +++ b/g4f/Provider/Llama2.py @@ -6,15 +6,14 @@ from ..typing import AsyncResult, Messages from .base_provider import AsyncGeneratorProvider models = { - "7B": {"name": "Llama 2 7B", "version": "d24902e3fa9b698cc208b5e63136c4e26e828659a9f09827ca6ec5bb83014381", "shortened":"7B"}, - "13B": {"name": "Llama 2 13B", "version": "9dff94b1bed5af738655d4a7cbcdcde2bd503aa85c94334fe1f42af7f3dd5ee3", "shortened":"13B"}, - "70B": {"name": "Llama 2 70B", "version": "2796ee9483c3fd7aa2e171d38f4ca12251a30609463dcfd4cd76703f22e96cdf", "shortened":"70B"}, + "meta-llama/Llama-2-7b-chat-hf": {"name": "Llama 2 7B", "version": "d24902e3fa9b698cc208b5e63136c4e26e828659a9f09827ca6ec5bb83014381", "shortened":"7B"}, + "meta-llama/Llama-2-13b-chat-hf": {"name": "Llama 2 13B", "version": "9dff94b1bed5af738655d4a7cbcdcde2bd503aa85c94334fe1f42af7f3dd5ee3", "shortened":"13B"}, + "meta-llama/Llama-2-70b-chat-hf": {"name": "Llama 2 70B", "version": "2796ee9483c3fd7aa2e171d38f4ca12251a30609463dcfd4cd76703f22e96cdf", "shortened":"70B"}, "Llava": {"name": "Llava 13B", "version": "6bc1c7bb0d2a34e413301fee8f7cc728d2d4e75bfab186aa995f63292bda92fc", "shortened":"Llava"} } class Llama2(AsyncGeneratorProvider): url = "https://www.llama2.ai" - supports_gpt_35_turbo = True working = True @classmethod @@ -26,8 +25,8 @@ class Llama2(AsyncGeneratorProvider): **kwargs ) -> AsyncResult: if not model: - model = "70B" - if model not in models: + model = "meta-llama/Llama-2-70b-chat-hf" + elif model not in models: raise ValueError(f"Model are not supported: {model}") version = models[model]["version"] headers = { @@ -54,7 +53,7 @@ class Llama2(AsyncGeneratorProvider): "systemPrompt": kwargs.get("system_message", "You are a helpful assistant."), "temperature": kwargs.get("temperature", 0.75), "topP": kwargs.get("top_p", 0.9), - "maxTokens": kwargs.get("max_tokens", 1024), + "maxTokens": kwargs.get("max_tokens", 8000), "image": None } started = False @@ -68,9 +67,9 @@ class Llama2(AsyncGeneratorProvider): def format_prompt(messages: Messages): messages = [ - f"[INST]{message['content']}[/INST]" + f"[INST] {message['content']} [/INST]" if message["role"] == "user" else message["content"] for message in messages ] - return "\n".join(messages)
\ No newline at end of file + return "\n".join(messages) + "\n"
\ No newline at end of file diff --git a/g4f/Provider/__init__.py b/g4f/Provider/__init__.py index 653b6026..60d3bd25 100644 --- a/g4f/Provider/__init__.py +++ b/g4f/Provider/__init__.py @@ -17,6 +17,7 @@ from .ChatgptFree import ChatgptFree from .ChatgptLogin import ChatgptLogin from .ChatgptX import ChatgptX from .Cromicle import Cromicle +from .DeepInfra import DeepInfra from .FakeGpt import FakeGpt from .FreeGpt import FreeGpt from .GPTalk import GPTalk @@ -70,6 +71,7 @@ class ProviderUtils: 'ChatgptX': ChatgptX, 'CodeLinkAva': CodeLinkAva, 'Cromicle': Cromicle, + 'DeepInfra': DeepInfra, 'DfeHub': DfeHub, 'EasyChat': EasyChat, 'Equing': Equing, @@ -144,6 +146,7 @@ __all__ = [ 'ChatgptLogin', 'ChatgptX', 'Cromicle', + 'DeepInfra', 'CodeLinkAva', 'DfeHub', 'EasyChat', diff --git a/g4f/models.py b/g4f/models.py index 7eee917a..5e4d9b40 100644 --- a/g4f/models.py +++ b/g4f/models.py @@ -6,12 +6,14 @@ from .Provider import ( GptForLove, ChatgptAi, GptChatly, + DeepInfra, ChatgptX, ChatBase, GeekGpt, FakeGpt, FreeGpt, NoowAi, + Llama2, Vercel, Aichat, GPTalk, @@ -74,6 +76,21 @@ gpt_4 = Model( ]) ) +llama2_7b = Model( + name = "meta-llama/Llama-2-7b-chat-hf", + base_provider = 'huggingface', + best_provider = RetryProvider([Llama2, DeepInfra])) + +llama2_13b = Model( + name ="meta-llama/Llama-2-13b-chat-hf", + base_provider = 'huggingface', + best_provider = RetryProvider([Llama2, DeepInfra])) + +llama2_70b = Model( + name = "meta-llama/Llama-2-70b-chat-hf", + base_provider = "huggingface", + best_provider = RetryProvider([Llama2, DeepInfra])) + # Bard palm = Model( name = 'palm', @@ -246,6 +263,11 @@ class ModelUtils: 'gpt-4-0613' : gpt_4_0613, 'gpt-4-32k' : gpt_4_32k, 'gpt-4-32k-0613' : gpt_4_32k_0613, + + # Llama 2 + 'llama2-7b' : llama2_7b, + 'llama2-13b': llama2_13b, + 'llama2-70b': llama2_70b, # Bard 'palm2' : palm, |