SystemSh0cker/private-gpt
1
0
Fork 0
mirror of https://github.com/zylon-ai/private-gpt.git synced 2025-12-22 20:12:55 +01:00
Code Issues Projects Releases Packages Activity

WIP more prompt format, and more maintainable

Browse source
This commit is contained in:
Louis 2023-12-03 00:48:43 +01:00
parent 3d301d0c6f
commit 76faffb269
11 changed files with 476 additions and 217 deletions
Download patch file Download diff file Expand all files Collapse all files

33
private_gpt/settings/settings.py
View file

@ -98,13 +98,33 @@ class LocalSettings(BaseModel):
embedding_hf_model_name: str = Field(
description="Name of the HuggingFace model to use for embeddings"
)
prompt_style: Literal["default", "llama2", "tag"] = Field(
prompt_style: Literal[
"llama_cpp.llama-2",
"llama_cpp.alpaca",
"llama_cpp.vicuna",
"llama_cpp.oasst_llama",
"llama_cpp.baichuan-2",
"llama_cpp.baichuan",
"llama_cpp.openbuddy",
"llama_cpp.redpajama-incite",
"llama_cpp.snoozy",
"llama_cpp.phind",
"llama_cpp.intel",
"llama_cpp.open-orca",
"llama_cpp.mistrallite",
"llama_cpp.zephyr",
"llama_cpp.chatml",
"llama_cpp.openchat",
"llama2",
"vigogne",
"template",
] | None = Field(
None,
description=(
"The prompt style to use for the chat engine. "
"If `default` - use the default prompt style from the llama_index. It should look like `role: message`.\n"
"If None is given - use the default prompt style from the llama_index. It should look like `role: message`.\n"
"If `llama2` - use the llama2 prompt style from the llama_index. Based on `<s>`, `[INST]` and `<<SYS>>`.\n"
"If `tag` - use the `tag` prompt style. It should look like `<|role|>: message`. \n"
"If `llama_cpp.<name>` - use the `<name>` prompt style, implemented by `llama-cpp-python`. \n"
"`llama2` is the historic behaviour. `default` might work better with your custom models."
),
)
@ -118,6 +138,13 @@ class LocalSettings(BaseModel):
),
)
template_name: str | None = Field(
None,
description=(
"The name of the template to use for the chat engine, if the `prompt_style` is `template`."
),
)
class EmbeddingSettings(BaseModel):
mode: Literal["local", "openai", "sagemaker", "mock"]