Merge branch 'master' into feature/insight-agent

CONTRIBUTING.md

@@ -1,6 +1,6 @@
 🐫 **Welcome to CAMEL!** 🐫
 
-Thank you for your interest in contributing to the CAMEL project! 🎉 We're excited to have your support. As an open-source initiative in a rapidly evolving and open-ended field, we wholeheartedly welcome contributions of all kinds. Whether you want to introduce new features, enhance the infrastructure, improve documentation, asking issues, add more examples, implement state-of-the-art research ideas, or fix bugs, we appreciate your enthusiasm and efforts. 🙌  You are welcome to join our [slack](https://join.slack.com/t/camel-kwr1314/shared_invite/zt-1vy8u9lbo-ZQmhIAyWSEfSwLCl2r2eKA) for more efficient communication. 💬
+Thank you for your interest in contributing to the CAMEL project! 🎉 We're excited to have your support. As an open-source initiative in a rapidly evolving and open-ended field, we wholeheartedly welcome contributions of all kinds. Whether you want to introduce new features, enhance the infrastructure, improve documentation, asking issues, add more examples, implement state-of-the-art research ideas, or fix bugs, we appreciate your enthusiasm and efforts. 🙌  You are welcome to join our [slack](https://join.slack.com/t/camel-ai/shared_invite/zt-2g7xc41gy-_7rcrNNAArIP6sLQqldkqQ) for more efficient communication. 💬
 
 ## Guidelines 📝
 

README.md

@@ -122,7 +122,7 @@ pip install -e .[all] # (Optional)
 
 ## Example
 
-You can find a list of tasks for different set of assistant and user role pairs [here](https://drive.google.com/file/d/194PPaSTBR07m-PzjS-Ty6KlPLdFIPQDd/view?usp=share_link).
+You can find a list of tasks for different sets of assistant and user role pairs [here](https://drive.google.com/file/d/194PPaSTBR07m-PzjS-Ty6KlPLdFIPQDd/view?usp=share_link).
 
 As an example, to run the `role_playing.py` script:
 
@@ -282,7 +282,7 @@ For more information please contact camel.ai.team@gmail.com.
 [colab-image]: https://colab.research.google.com/assets/colab-badge.svg
 [huggingface-url]: https://huggingface.co/camel-ai
 [huggingface-image]: https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-CAMEL--AI-ffc107?color=ffc107&logoColor=white
-[slack-url]: https://join.slack.com/t/camel-kwr1314/shared_invite/zt-1vy8u9lbo-ZQmhIAyWSEfSwLCl2r2eKA
+[slack-url]: https://join.slack.com/t/camel-ai/shared_invite/zt-2g7xc41gy-_7rcrNNAArIP6sLQqldkqQ
 [slack-image]: https://img.shields.io/badge/Slack-CAMEL--AI-blueviolet?logo=slack
 [discord-url]: https://discord.gg/CNcNpquyDc
 [discord-image]: https://img.shields.io/badge/Discord-CAMEL--AI-7289da?logo=discord&logoColor=white&color=7289da

camel/configs.py

@@ -17,6 +17,8 @@
 from dataclasses import asdict, dataclass, field
 from typing import TYPE_CHECKING, Any, Dict, List, Optional, Sequence, Union
 
+from anthropic._types import NOT_GIVEN, NotGiven
+
 if TYPE_CHECKING:
     from camel.functions import OpenAIFunction
 
@@ -211,3 +213,55 @@ class OpenSourceConfig(BaseConfig):
     param
     for param in asdict(FunctionCallingConfig()).keys()
 }
+
+
+@dataclass(frozen=True)
+class AnthropicConfig(BaseConfig):
+    r"""Defines the parameters for generating chat completions using the
+    Anthropic API.
+
+    See: https://docs.anthropic.com/claude/reference/complete_post
+    Args:
+        max_tokens_to_sample (int, optional): The maximum number of tokens to
+            generate before stopping. Note that Anthropic models may stop
+            before reaching this maximum. This parameter only specifies the
+            absolute maximum number of tokens to generate.
+            (default: :obj:`256`)
+        stop_sequences (List[str], optional): Sequences that will cause the
+            model to stop generating completion text. Anthropic models stop
+            on "\n\nHuman:", and may include additional built-in stop sequences
+            in the future. By providing the stop_sequences parameter, you may
+            include additional strings that will cause the model to stop
+            generating.
+        temperature (float, optional): Amount of randomness injected into the
+            response. Defaults to 1. Ranges from 0 to 1. Use temp closer to 0
+            for analytical / multiple choice, and closer to 1 for creative
+            and generative tasks.
+            (default: :obj:`1`)
+        top_p (float, optional): Use nucleus sampling. In nucleus sampling, we
+            compute the cumulative distribution over all the options for each
+            subsequent token in decreasing probability order and cut it off
+            once it reaches a particular probability specified by `top_p`.
+            You should either alter `temperature` or `top_p`,
+            but not both.
+            (default: :obj:`0.7`)
+        top_k (int, optional): Only sample from the top K options for each
+            subsequent token. Used to remove "long tail" low probability
+            responses.
+            (default: :obj:`5`)
+        metadata: An object describing metadata about the request.
+        stream (bool, optional): Whether to incrementally stream the response
+          using server-sent events.
+            (default: :obj:`False`)
+
+    """
+    max_tokens: int = 256
+    stop_sequences: Union[List[str], NotGiven] = NOT_GIVEN
+    temperature: float = 1
+    top_p: Union[float, NotGiven] = NOT_GIVEN
+    top_k: Union[int, NotGiven] = NOT_GIVEN
+    metadata: NotGiven = NOT_GIVEN
+    stream: bool = False
+
+
+ANTHROPIC_API_PARAMS = {param for param in asdict(AnthropicConfig()).keys()}

camel/embeddings/openai_embedding.py

@@ -17,7 +17,7 @@
 
 from camel.embeddings import BaseEmbedding
 from camel.types import EmbeddingModelType
-from camel.utils import openai_api_key_required
+from camel.utils import api_key_required
 
 
 class OpenAIEmbedding(BaseEmbedding[str]):
@@ -41,7 +41,7 @@ def __init__(
         self.output_dim = model_type.output_dim
         self.client = OpenAI()
 
-    @openai_api_key_required
+    @api_key_required
     def embed_list(
         self,
         objs: List[str],

camel/models/__init__.py

@@ -15,11 +15,13 @@
 from .openai_model import OpenAIModel
 from .stub_model import StubModel
 from .open_source_model import OpenSourceModel
+from .anthropic_model import AnthropicModel
 from .model_factory import ModelFactory
 
 __all__ = [
     'BaseModelBackend',
     'OpenAIModel',
+    'AnthropicModel',
     'StubModel',
     'OpenSourceModel',
     'ModelFactory',

camel/models/anthropic_model.py

@@ -0,0 +1,118 @@
+# =========== Copyright 2023 @ CAMEL-AI.org. All Rights Reserved. ===========
+# Licensed under the Apache License, Version 2.0 (the “License”);
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an “AS IS” BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# =========== Copyright 2023 @ CAMEL-AI.org. All Rights Reserved. ===========
+import os
+from typing import Any, Dict, Optional
+
+from anthropic import Anthropic
+from anthropic._types import NOT_GIVEN
+
+from camel.configs import ANTHROPIC_API_PARAMS
+from camel.models import BaseModelBackend
+from camel.types import ChatCompletion, ModelType
+from camel.utils import AnthropicTokenCounter, BaseTokenCounter
+
+
+class AnthropicModel(BaseModelBackend):
+    r"""Anthropic API in a unified BaseModelBackend interface."""
+
+    def __init__(self, model_type: ModelType,
+                 model_config_dict: Dict[str, Any]) -> None:
+        super().__init__(model_type, model_config_dict)
+        self.client = Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
+        self._token_counter: Optional[BaseTokenCounter] = None
+
+    def _convert_response_from_anthropic_to_openai(self, response):
+        # openai ^1.0.0 format, reference openai/types/chat/chat_completion.py
+        obj = ChatCompletion.construct(
+            id=None, choices=[
+                dict(
+                    index=0, message={
+                        "role": "assistant",
+                        "content": response.content[0].text
+                    }, finish_reason=response.stop_reason)
+            ], created=None, model=response.model, object="chat.completion")
+        return obj
+
+    @property
+    def token_counter(self) -> BaseTokenCounter:
+        r"""Initialize the token counter for the model backend.
+
+        Returns:
+            BaseTokenCounter: The token counter following the model's
+                tokenization style.
+        """
+        if not self._token_counter:
+            self._token_counter = AnthropicTokenCounter(self.model_type)
+        return self._token_counter
+
+    def count_tokens_from_prompt(self, prompt: str) -> int:
+        r"""Count the number of tokens from a prompt.
+
+        Args:
+            prompt (str): The prompt string.
+
+        Returns:
+            int: The number of tokens in the prompt.
+        """
+        return self.client.count_tokens(prompt)
+
+    def run(
+        self,
+        messages,
+    ):
+        r"""Run inference of Anthropic chat completion.
+
+        Args:
+            messages (List[Dict]): Message list with the chat history
+                in OpenAI API format.
+
+        Returns:
+            Dict[str, Any]: Response in the OpenAI API format.
+        """
+
+        if messages[0]["role"] == "system":
+            sys_msg = messages.pop(0)["content"]  # type: ignore
+        else:
+            sys_msg = NOT_GIVEN
+        response = self.client.messages.create(  # type: ignore
+            model=self.model_type.value, system=sys_msg, messages=messages,
+            **self.model_config_dict)
+
+        # format response to openai format
+        response = self._convert_response_from_anthropic_to_openai(response)
+
+        return response
+
+    def check_model_config(self):
+        r"""Check whether the model configuration is valid for anthropic
+        model backends.
+
+        Raises:
+            ValueError: If the model configuration dictionary contains any
+                unexpected arguments to OpenAI API, or it does not contain
+                :obj:`model_path` or :obj:`server_url`.
+        """
+        for param in self.model_config_dict:
+            if param not in ANTHROPIC_API_PARAMS:
+                raise ValueError(f"Unexpected argument `{param}` is "
+                                 "input into Anthropic model backend.")
+
+    @property
+    def stream(self) -> bool:
+        r"""Returns whether the model is in stream mode,
+            which sends partial results each time.
+        Returns:
+            bool: Whether the model is in stream mode.
+        """
+        return self.model_config_dict.get("stream", False)

camel/models/model_factory.py

@@ -14,6 +14,7 @@
 from typing import Any, Dict
 
 from camel.models import (
+    AnthropicModel,
     BaseModelBackend,
     OpenAIModel,
     OpenSourceModel,
@@ -52,6 +53,8 @@ def create(model_type: ModelType,
             model_class = StubModel
         elif model_type.is_open_source:
             model_class = OpenSourceModel
+        elif model_type.is_anthropic:
+            model_class = AnthropicModel
         else:
             raise ValueError(f"Unknown model type `{model_type}` is input")
 

camel/models/openai_model.py

@@ -20,11 +20,7 @@
 from camel.messages import OpenAIMessage
 from camel.models import BaseModelBackend
 from camel.types import ChatCompletion, ChatCompletionChunk, ModelType
-from camel.utils import (
-    BaseTokenCounter,
-    OpenAITokenCounter,
-    openai_api_key_required,
-)
+from camel.utils import BaseTokenCounter, OpenAITokenCounter, api_key_required
 
 
 class OpenAIModel(BaseModelBackend):
@@ -57,7 +53,7 @@ def token_counter(self) -> BaseTokenCounter:
             self._token_counter = OpenAITokenCounter(self.model_type)
         return self._token_counter
 
-    @openai_api_key_required
+    @api_key_required
     def run(
         self,
         messages: List[OpenAIMessage],

camel/types/enums.py

@@ -37,6 +37,17 @@ class ModelType(Enum):
     VICUNA = "vicuna"
     VICUNA_16K = "vicuna-16k"
 
+    # Legacy anthropic models
+    # NOTE: anthropic lagecy models only Claude 2.1 has system prompt support
+    CLAUDE_2_1 = "claude-2.1"
+    CLAUDE_2_0 = "claude-2.0"
+    CLAUDE_INSTANT_1_2 = "claude-instant-1.2"
+
+    #  3 models
+    CLAUDE_3_OPUS = "claude-3-opus-20240229"
+    CLAUDE_3_SONNET = "claude-3-sonnet-20240229"
+    CLAUDE_3_HAIKU = "claude-3-haiku-20240307"
+
     @property
     def value_for_tiktoken(self) -> str:
         return self.value if self is not ModelType.STUB else "gpt-3.5-turbo"
@@ -62,6 +73,22 @@ def is_open_source(self) -> bool:
             ModelType.VICUNA_16K,
         }
 
+    @property
+    def is_anthropic(self) -> bool:
+        r"""Returns whether this type of models is Anthropic-released model.
+
+        Returns:
+            bool: Whether this type of models is anthropic.
+        """
+        return self in {
+            ModelType.CLAUDE_INSTANT_1_2,
+            ModelType.CLAUDE_2_0,
+            ModelType.CLAUDE_2_1,
+            ModelType.CLAUDE_3_OPUS,
+            ModelType.CLAUDE_3_SONNET,
+            ModelType.CLAUDE_3_HAIKU,
+        }
+
     @property
     def token_limit(self) -> int:
         r"""Returns the maximum token limit for a given model.
@@ -89,6 +116,15 @@ def token_limit(self) -> int:
             return 2048
         elif self is ModelType.VICUNA_16K:
             return 16384
+        if self in {ModelType.CLAUDE_2_0, ModelType.CLAUDE_INSTANT_1_2}:
+            return 100_000
+        elif self in {
+                ModelType.CLAUDE_2_1,
+                ModelType.CLAUDE_3_OPUS,
+                ModelType.CLAUDE_3_SONNET,
+                ModelType.CLAUDE_3_HAIKU,
+        }:
+            return 200_000
         else:
             raise ValueError("Unknown model type")
 

camel/utils/__init__.py

@@ -12,7 +12,7 @@
 # limitations under the License.
 # =========== Copyright 2023 @ CAMEL-AI.org. All Rights Reserved. ===========
 from .commons import (
-    openai_api_key_required,
+    api_key_required,
     print_text_animated,
     get_prompt_template_key_words,
     get_first_int,
@@ -24,21 +24,19 @@
     to_pascal,
     PYDANTIC_V2,
 )
-from .token_counting import (
-    get_model_encoding,
-    BaseTokenCounter,
-    OpenAITokenCounter,
-    OpenSourceTokenCounter,
-)
+from .token_counting import (get_model_encoding, BaseTokenCounter,
+                             OpenAITokenCounter, OpenSourceTokenCounter,
+                             AnthropicTokenCounter)
 
 __all__ = [
-    'openai_api_key_required',
+    'api_key_required',
     'print_text_animated',
     'get_prompt_template_key_words',
     'get_first_int',
     'download_tasks',
     'get_task_list',
     'check_server_running',
+    'AnthropicTokenCounter',
     'get_system_information',
     'to_pascal',
     'PYDANTIC_V2',