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 | class Solar(OpenAI):
api_key: str = Field(default=None, description="The SOLAR API key.")
api_base: str = Field(default="", description="The base URL for SOLAR API.")
model: str = Field(
default="solar-1-mini-chat", description="The SOLAR model to use."
)
context_window: int = Field(
default=DEFAULT_CONTEXT_WINDOW,
description=LLMMetadata.__fields__["context_window"].field_info.description,
)
is_chat_model: bool = Field(
default=False,
description=LLMMetadata.__fields__["is_chat_model"].field_info.description,
)
is_function_calling_model: bool = Field(
default=False,
description=LLMMetadata.__fields__[
"is_function_calling_model"
].field_info.description,
)
tokenizer: Union[Tokenizer, str, None] = Field(
default=None,
description=(
"An instance of a tokenizer object that has an encode method, or the name"
" of a tokenizer model from Hugging Face. If left as None, then this"
" disables inference of max_tokens."
),
)
def __init__(
self,
model: str = DEFAULT_SOLAR_MODEL,
temperature: float = 0.1,
max_tokens: Optional[int] = None,
additional_kwargs: Optional[Dict[str, Any]] = None,
max_retries: int = 3,
timeout: float = 60.0,
reuse_client: bool = True,
api_key: Optional[str] = None,
api_base: Optional[str] = None,
api_version: Optional[str] = None,
callback_manager: Optional[CallbackManager] = None,
default_headers: Optional[Dict[str, str]] = None,
http_client: Optional[httpx.Client] = None,
# base class
system_prompt: Optional[str] = None,
messages_to_prompt: Optional[Callable[[Sequence[ChatMessage]], str]] = None,
completion_to_prompt: Optional[Callable[[str], str]] = None,
pydantic_program_mode: PydanticProgramMode = PydanticProgramMode.DEFAULT,
output_parser: Optional[BaseOutputParser] = None,
**kwargs: Any,
) -> None:
# add warning for this class is deprecated
warnings.warn(
"""Solar LLM is deprecated. Please use Upstage LLM instead.
Install the package using `pip install llama-index-llms-upstage`
""",
)
api_key, api_base = resolve_solar_credentials(
api_key=api_key,
api_base=api_base,
)
super().__init__(
model=model,
temperature=temperature,
max_tokens=max_tokens,
additional_kwargs=additional_kwargs,
max_retries=max_retries,
callback_manager=callback_manager,
api_key=api_key,
api_version=api_version,
api_base=api_base,
timeout=timeout,
reuse_client=reuse_client,
default_headers=default_headers,
system_prompt=system_prompt,
messages_to_prompt=messages_to_prompt,
completion_to_prompt=completion_to_prompt,
pydantic_program_mode=pydantic_program_mode,
output_parser=output_parser,
**kwargs,
)
@property
def metadata(self) -> LLMMetadata:
return LLMMetadata(
context_window=self.context_window,
num_output=self.max_tokens or -1,
is_chat_model=self.is_chat_model,
is_function_calling_model=self.is_function_calling_model,
model_name=self.model,
)
@property
def _tokenizer(self) -> Optional[Tokenizer]:
if isinstance(self.tokenizer, str):
return AutoTokenizer.from_pretrained(self.tokenizer)
return self.tokenizer
@classmethod
def class_name(cls) -> str:
return "Solar"
def complete(
self, prompt: str, formatted: bool = False, **kwargs: Any
) -> CompletionResponse:
"""Complete the prompt."""
if not formatted:
prompt = self.completion_to_prompt(prompt)
return super().complete(prompt, **kwargs)
def stream_complete(
self, prompt: str, formatted: bool = False, **kwargs: Any
) -> CompletionResponseGen:
"""Stream complete the prompt."""
if not formatted:
prompt = self.completion_to_prompt(prompt)
return super().stream_complete(prompt, **kwargs)
def chat(self, messages: Sequence[ChatMessage], **kwargs: Any) -> ChatResponse:
"""Chat with the model."""
if not self.metadata.is_chat_model:
prompt = self.messages_to_prompt(messages)
completion_response = self.complete(prompt, formatted=True, **kwargs)
return completion_response_to_chat_response(completion_response)
return super().chat(messages, **kwargs)
def stream_chat(
self, messages: Sequence[ChatMessage], **kwargs: Any
) -> ChatResponseGen:
if not self.metadata.is_chat_model:
prompt = self.messages_to_prompt(messages)
completion_response = self.stream_complete(prompt, formatted=True, **kwargs)
return stream_completion_response_to_chat_response(completion_response)
return super().stream_chat(messages, **kwargs)
# -- Async methods --
async def acomplete(
self, prompt: str, formatted: bool = False, **kwargs: Any
) -> CompletionResponse:
"""Complete the prompt."""
if not formatted:
prompt = self.completion_to_prompt(prompt)
return await super().acomplete(prompt, **kwargs)
async def astream_complete(
self, prompt: str, formatted: bool = False, **kwargs: Any
) -> CompletionResponseAsyncGen:
"""Stream complete the prompt."""
if not formatted:
prompt = self.completion_to_prompt(prompt)
return await super().astream_complete(prompt, **kwargs)
async def achat(
self, messages: Sequence[ChatMessage], **kwargs: Any
) -> ChatResponse:
"""Chat with the model."""
if not self.metadata.is_chat_model:
prompt = self.messages_to_prompt(messages)
completion_response = await self.acomplete(prompt, formatted=True, **kwargs)
return completion_response_to_chat_response(completion_response)
return await super().achat(messages, **kwargs)
async def astream_chat(
self, messages: Sequence[ChatMessage], **kwargs: Any
) -> ChatResponseAsyncGen:
if not self.metadata.is_chat_model:
prompt = self.messages_to_prompt(messages)
completion_response = await self.astream_complete(
prompt, formatted=True, **kwargs
)
return async_stream_completion_response_to_chat_response(
completion_response
)
return await super().astream_chat(messages, **kwargs)
|
