Spaces:
Runtime error
Runtime error
File size: 10,959 Bytes
ed4d993 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 |
import re
from collections import defaultdict
from typing import Any, Dict, Iterator, List, Optional, Tuple, Union
from langchain_core._api.deprecation import deprecated
from langchain_core.callbacks import (
CallbackManagerForLLMRun,
)
from langchain_core.language_models.chat_models import BaseChatModel
from langchain_core.messages import (
AIMessage,
AIMessageChunk,
BaseMessage,
ChatMessage,
HumanMessage,
SystemMessage,
)
from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult
from langchain_core.pydantic_v1 import Extra
from langchain_community.chat_models.anthropic import (
convert_messages_to_prompt_anthropic,
)
from langchain_community.chat_models.meta import convert_messages_to_prompt_llama
from langchain_community.llms.bedrock import BedrockBase
from langchain_community.utilities.anthropic import (
get_num_tokens_anthropic,
get_token_ids_anthropic,
)
def _convert_one_message_to_text_mistral(message: BaseMessage) -> str:
if isinstance(message, ChatMessage):
message_text = f"\n\n{message.role.capitalize()}: {message.content}"
elif isinstance(message, HumanMessage):
message_text = f"[INST] {message.content} [/INST]"
elif isinstance(message, AIMessage):
message_text = f"{message.content}"
elif isinstance(message, SystemMessage):
message_text = f"<<SYS>> {message.content} <</SYS>>"
else:
raise ValueError(f"Got unknown type {message}")
return message_text
def convert_messages_to_prompt_mistral(messages: List[BaseMessage]) -> str:
"""Convert a list of messages to a prompt for mistral."""
return "\n".join(
[_convert_one_message_to_text_mistral(message) for message in messages]
)
def _format_image(image_url: str) -> Dict:
"""
Formats an image of format data:image/jpeg;base64,{b64_string}
to a dict for anthropic api
{
"type": "base64",
"media_type": "image/jpeg",
"data": "/9j/4AAQSkZJRg...",
}
And throws an error if it's not a b64 image
"""
regex = r"^data:(?P<media_type>image/.+);base64,(?P<data>.+)$"
match = re.match(regex, image_url)
if match is None:
raise ValueError(
"Anthropic only supports base64-encoded images currently."
" Example: data:image/png;base64,'/9j/4AAQSk'..."
)
return {
"type": "base64",
"media_type": match.group("media_type"),
"data": match.group("data"),
}
def _format_anthropic_messages(
messages: List[BaseMessage],
) -> Tuple[Optional[str], List[Dict]]:
"""Format messages for anthropic."""
"""
[
{
"role": _message_type_lookups[m.type],
"content": [_AnthropicMessageContent(text=m.content).dict()],
}
for m in messages
]
"""
system: Optional[str] = None
formatted_messages: List[Dict] = []
for i, message in enumerate(messages):
if message.type == "system":
if i != 0:
raise ValueError("System message must be at beginning of message list.")
if not isinstance(message.content, str):
raise ValueError(
"System message must be a string, "
f"instead was: {type(message.content)}"
)
system = message.content
continue
role = _message_type_lookups[message.type]
content: Union[str, List[Dict]]
if not isinstance(message.content, str):
# parse as dict
assert isinstance(
message.content, list
), "Anthropic message content must be str or list of dicts"
# populate content
content = []
for item in message.content:
if isinstance(item, str):
content.append(
{
"type": "text",
"text": item,
}
)
elif isinstance(item, dict):
if "type" not in item:
raise ValueError("Dict content item must have a type key")
if item["type"] == "image_url":
# convert format
source = _format_image(item["image_url"]["url"])
content.append(
{
"type": "image",
"source": source,
}
)
else:
content.append(item)
else:
raise ValueError(
f"Content items must be str or dict, instead was: {type(item)}"
)
else:
content = message.content
formatted_messages.append(
{
"role": role,
"content": content,
}
)
return system, formatted_messages
class ChatPromptAdapter:
"""Adapter class to prepare the inputs from Langchain to prompt format
that Chat model expects.
"""
@classmethod
def convert_messages_to_prompt(
cls, provider: str, messages: List[BaseMessage]
) -> str:
if provider == "anthropic":
prompt = convert_messages_to_prompt_anthropic(messages=messages)
elif provider == "meta":
prompt = convert_messages_to_prompt_llama(messages=messages)
elif provider == "mistral":
prompt = convert_messages_to_prompt_mistral(messages=messages)
elif provider == "amazon":
prompt = convert_messages_to_prompt_anthropic(
messages=messages,
human_prompt="\n\nUser:",
ai_prompt="\n\nBot:",
)
else:
raise NotImplementedError(
f"Provider {provider} model does not support chat."
)
return prompt
@classmethod
def format_messages(
cls, provider: str, messages: List[BaseMessage]
) -> Tuple[Optional[str], List[Dict]]:
if provider == "anthropic":
return _format_anthropic_messages(messages)
raise NotImplementedError(
f"Provider {provider} not supported for format_messages"
)
_message_type_lookups = {
"human": "user",
"ai": "assistant",
"AIMessageChunk": "assistant",
"HumanMessageChunk": "user",
"function": "user",
}
@deprecated(
since="0.0.34", removal="0.3", alternative_import="langchain_aws.ChatBedrock"
)
class BedrockChat(BaseChatModel, BedrockBase):
"""Chat model that uses the Bedrock API."""
@property
def _llm_type(self) -> str:
"""Return type of chat model."""
return "amazon_bedrock_chat"
@classmethod
def is_lc_serializable(cls) -> bool:
"""Return whether this model can be serialized by Langchain."""
return True
@classmethod
def get_lc_namespace(cls) -> List[str]:
"""Get the namespace of the langchain object."""
return ["langchain", "chat_models", "bedrock"]
@property
def lc_attributes(self) -> Dict[str, Any]:
attributes: Dict[str, Any] = {}
if self.region_name:
attributes["region_name"] = self.region_name
return attributes
class Config:
"""Configuration for this pydantic object."""
extra = Extra.forbid
def _stream(
self,
messages: List[BaseMessage],
stop: Optional[List[str]] = None,
run_manager: Optional[CallbackManagerForLLMRun] = None,
**kwargs: Any,
) -> Iterator[ChatGenerationChunk]:
provider = self._get_provider()
prompt, system, formatted_messages = None, None, None
if provider == "anthropic":
system, formatted_messages = ChatPromptAdapter.format_messages(
provider, messages
)
else:
prompt = ChatPromptAdapter.convert_messages_to_prompt(
provider=provider, messages=messages
)
for chunk in self._prepare_input_and_invoke_stream(
prompt=prompt,
system=system,
messages=formatted_messages,
stop=stop,
run_manager=run_manager,
**kwargs,
):
delta = chunk.text
yield ChatGenerationChunk(message=AIMessageChunk(content=delta))
def _generate(
self,
messages: List[BaseMessage],
stop: Optional[List[str]] = None,
run_manager: Optional[CallbackManagerForLLMRun] = None,
**kwargs: Any,
) -> ChatResult:
completion = ""
llm_output: Dict[str, Any] = {"model_id": self.model_id}
if self.streaming:
for chunk in self._stream(messages, stop, run_manager, **kwargs):
completion += chunk.text
else:
provider = self._get_provider()
prompt, system, formatted_messages = None, None, None
params: Dict[str, Any] = {**kwargs}
if provider == "anthropic":
system, formatted_messages = ChatPromptAdapter.format_messages(
provider, messages
)
else:
prompt = ChatPromptAdapter.convert_messages_to_prompt(
provider=provider, messages=messages
)
if stop:
params["stop_sequences"] = stop
completion, usage_info = self._prepare_input_and_invoke(
prompt=prompt,
stop=stop,
run_manager=run_manager,
system=system,
messages=formatted_messages,
**params,
)
llm_output["usage"] = usage_info
return ChatResult(
generations=[ChatGeneration(message=AIMessage(content=completion))],
llm_output=llm_output,
)
def _combine_llm_outputs(self, llm_outputs: List[Optional[dict]]) -> dict:
final_usage: Dict[str, int] = defaultdict(int)
final_output = {}
for output in llm_outputs:
output = output or {}
usage = output.get("usage", {})
for token_type, token_count in usage.items():
final_usage[token_type] += token_count
final_output.update(output)
final_output["usage"] = final_usage
return final_output
def get_num_tokens(self, text: str) -> int:
if self._model_is_anthropic:
return get_num_tokens_anthropic(text)
else:
return super().get_num_tokens(text)
def get_token_ids(self, text: str) -> List[int]:
if self._model_is_anthropic:
return get_token_ids_anthropic(text)
else:
return super().get_token_ids(text)
|