github/home-llm
1
1
Fork 0
mirror of https://github.com/acon96/home-llm.git synced 2026-01-10 14:18:00 -05:00
Code Issues Packages Projects Releases Wiki Activity

hook up prompt caching to config flow

Browse Source
This commit is contained in:
Alex O'Connell
2024-03-30 16:50:24 -04:00
parent aa3f82d56f
commit 0008f6bb5e
4 changed files with 128 additions and 85 deletions
Download Patch File Download Diff File Expand all files Collapse all files

65
custom_components/llama_conversation/__init__.py
View File

@@ -41,13 +41,17 @@ from .const import (
CONF_ALLOWED_SERVICE_CALL_ARGUMENTS,
CONF_PROMPT_TEMPLATE,
CONF_USE_GBNF_GRAMMAR,
CONF_GBNF_GRAMMAR_FILE,
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES,
CONF_IN_CONTEXT_EXAMPLES_FILE,
CONF_TEXT_GEN_WEBUI_PRESET,
CONF_OPENAI_API_KEY,
CONF_TEXT_GEN_WEBUI_ADMIN_KEY,
CONF_REFRESH_SYSTEM_PROMPT,
CONF_REMEMBER_CONVERSATION,
CONF_REMEMBER_NUM_INTERACTIONS,
CONF_PROMPT_CACHING_ENABLED,
CONF_PROMPT_CACHING_INTERVAL,
CONF_SERVICE_CALL_REGEX,
CONF_REMOTE_USE_CHAT_ENDPOINT,
CONF_TEXT_GEN_WEBUI_CHAT_MODE,
@@ -63,9 +67,14 @@ from .const import (
DEFAULT_ALLOWED_SERVICE_CALL_ARGUMENTS,
DEFAULT_PROMPT_TEMPLATE,
DEFAULT_USE_GBNF_GRAMMAR,
DEFAULT_GBNF_GRAMMAR_FILE,
DEFAULT_USE_IN_CONTEXT_LEARNING_EXAMPLES,
DEFAULT_IN_CONTEXT_EXAMPLES_FILE,
DEFAULT_REFRESH_SYSTEM_PROMPT,
DEFAULT_REMEMBER_CONVERSATION,
DEFAULT_REMEMBER_NUM_INTERACTIONS,
DEFAULT_PROMPT_CACHING_ENABLED,
DEFAULT_PROMPT_CACHING_INTERVAL,
DEFAULT_SERVICE_CALL_REGEX,
DEFAULT_REMOTE_USE_CHAT_ENDPOINT,
DEFAULT_TEXT_GEN_WEBUI_CHAT_MODE,
@@ -81,8 +90,6 @@ from .const import (
TEXT_GEN_WEBUI_CHAT_MODE_INSTRUCT,
TEXT_GEN_WEBUI_CHAT_MODE_CHAT_INSTRUCT,
DOMAIN,
GBNF_GRAMMAR_FILE,
IN_CONTEXT_EXAMPLES_FILE,
PROMPT_TEMPLATE_DESCRIPTIONS,
)
@@ -90,19 +97,24 @@ _LOGGER = logging.getLogger(__name__)
CONFIG_SCHEMA = cv.config_entry_only_config_schema(DOMAIN)
async def update_listener(hass, entry):
async def update_listener(hass: HomeAssistant, entry: ConfigEntry):
"""Handle options update."""
hass.data[DOMAIN][entry.entry_id] = entry
# call update handler
agent = await ha_conversation._get_agent_manager(hass).async_get_agent(entry.entry_id)
agent: LLaMAAgent = await ha_conversation._get_agent_manager(hass).async_get_agent(entry.entry_id)
agent._update_options()
backend_type = entry.data.get(CONF_BACKEND_TYPE, DEFAULT_BACKEND_TYPE)
if backend_type in [ BACKEND_TYPE_LLAMA_HF, BACKEND_TYPE_LLAMA_EXISTING ]:
if entry.options.get(CONF_PROMPT_CACHING_ENABLED, DEFAULT_PROMPT_CACHING_ENABLED):
hass.async_create_task(agent._async_cache_prompt(None, None, None))
return True
async def async_setup_entry(hass: HomeAssistant, entry: ConfigEntry) -> bool:
"""Set up Local LLaMA Conversation from a config entry."""
def create_agent(backend_type):
agent_cls = None
@@ -132,8 +144,8 @@ async def async_setup_entry(hass: HomeAssistant, entry: ConfigEntry) -> bool:
hass.data[DOMAIN][entry.entry_id] = entry
if backend_type in [ BACKEND_TYPE_LLAMA_HF, BACKEND_TYPE_LLAMA_EXISTING ]:
# if entry.options.get(CONF_PROMPT_CACHING_ENABLED, DEFAULT_PROMPT_CACHING_ENABLED):
hass.async_create_task(agent._async_cache_prompt(None, None, None))
if entry.options.get(CONF_PROMPT_CACHING_ENABLED, DEFAULT_PROMPT_CACHING_ENABLED):
hass.async_create_task(agent._async_cache_prompt(None, None, None))
return True
@@ -190,7 +202,8 @@ class LLaMAAgent(AbstractConversationAgent):
def _load_icl_examples(self):
try:
icl_filename = os.path.join(os.path.dirname(__file__), IN_CONTEXT_EXAMPLES_FILE)
icl_file = self.entry.options.get(CONF_IN_CONTEXT_EXAMPLES_FILE, DEFAULT_IN_CONTEXT_EXAMPLES_FILE)
icl_filename = os.path.join(os.path.dirname(__file__), icl_file)
with open(icl_filename) as f:
self.in_context_examples = list(csv.DictReader(f))
@@ -210,7 +223,7 @@ class LLaMAAgent(AbstractConversationAgent):
self.in_context_examples = None
@property
def entry(self):
def entry(self) -> ConfigEntry:
return self.hass.data[DOMAIN][self.entry_id]
@property
@@ -239,7 +252,7 @@ class LLaMAAgent(AbstractConversationAgent):
template_desc = PROMPT_TEMPLATE_DESCRIPTIONS[prompt_template]
refresh_system_prompt = self.entry.options.get(CONF_REFRESH_SYSTEM_PROMPT, DEFAULT_REFRESH_SYSTEM_PROMPT)
remember_conversation = self.entry.options.get(CONF_REMEMBER_CONVERSATION, DEFAULT_REMEMBER_CONVERSATION)
remember_num_interactions = self.entry.options.get(CONF_REMEMBER_NUM_INTERACTIONS, False)
remember_num_interactions = self.entry.options.get(CONF_REMEMBER_NUM_INTERACTIONS, DEFAULT_REMEMBER_NUM_INTERACTIONS)
service_call_regex = self.entry.options.get(CONF_SERVICE_CALL_REGEX, DEFAULT_SERVICE_CALL_REGEX)
allowed_service_call_arguments = self.entry.options \
.get(CONF_ALLOWED_SERVICE_CALL_ARGUMENTS, DEFAULT_ALLOWED_SERVICE_CALL_ARGUMENTS)
@@ -439,9 +452,6 @@ class LLaMAAgent(AbstractConversationAgent):
"""Generate a prompt for the user."""
entities_to_expose, domains = self._async_get_exposed_entities()
if True: # TODO: change this to check if prompt caching is enabled
self.last_updated_entities
extra_attributes_to_expose = self.entry.options \
.get(CONF_EXTRA_ATTRIBUTES_TO_EXPOSE, DEFAULT_EXTRA_ATTRIBUTES_TO_EXPOSE)
allowed_service_call_arguments = self.entry.options \
@@ -541,9 +551,8 @@ class LocalLLaMAAgent(LLaMAAgent):
llama_cpp_module: Any
remove_prompt_caching_listener: Callable
model_lock: threading.Lock
fastest_cache_prime_interval: int
last_cache_prime: float
last_updated_entities: list[str]
# last_updated_entities: list[str]
def _load_model(self, entry: ConfigEntry) -> None:
self.model_path = entry.data.get(CONF_DOWNLOADED_MODEL_FILE)
@@ -580,7 +589,7 @@ class LocalLLaMAAgent(LLaMAAgent):
self.grammar = None
if entry.options.get(CONF_USE_GBNF_GRAMMAR, DEFAULT_USE_GBNF_GRAMMAR):
self._load_grammar()
self._load_grammar(entry.options.get(CONF_GBNF_GRAMMAR_FILE, DEFAULT_GBNF_GRAMMAR_FILE))
# TODO: check about disk caching
# self.llm.set_cache(self.llama_cpp_module.LlamaDiskCache(
@@ -590,19 +599,16 @@ class LocalLLaMAAgent(LLaMAAgent):
self.remove_prompt_caching_listener = None
self.last_cache_prime = None
self.fastest_cache_prime_interval = 30
# self.fastest_cache_prime_interval = entry.options.get(CONF_PROMPT_CACHING_INTERVAL, DEFAULT_PROMPT_CACHING_INTERVAL)
self.model_lock = threading.Lock()
# if entry.options.get(CONF_PROMPT_CACHING_ENABLED, DEFAULT_PROMPT_CACHING_ENABLED):
self._set_prompt_caching(enabled=True)
if entry.options.get(CONF_PROMPT_CACHING_ENABLED, DEFAULT_PROMPT_CACHING_ENABLED):
self._set_prompt_caching(enabled=True)
def _load_grammar(self):
def _load_grammar(self, filename: str):
LlamaGrammar = getattr(self.llama_cpp_module, "LlamaGrammar")
_LOGGER.debug("Loading grammar...")
try:
# TODO: make grammar configurable
with open(os.path.join(os.path.dirname(__file__), GBNF_GRAMMAR_FILE)) as f:
with open(os.path.join(os.path.dirname(__file__), filename)) as f:
grammar_str = "".join(f.readlines())
self.grammar = LlamaGrammar.from_string(grammar_str)
_LOGGER.debug("Loaded grammar")
@@ -611,18 +617,20 @@ class LocalLLaMAAgent(LLaMAAgent):
self.grammar = None
def _update_options(self):
LLaMAAgent._update_options()
LLaMAAgent._update_options(self)
if self.entry.options.get(CONF_USE_GBNF_GRAMMAR, DEFAULT_USE_GBNF_GRAMMAR):
self._load_grammar()
self._load_grammar(self.entry.options.get(CONF_GBNF_GRAMMAR_FILE, DEFAULT_GBNF_GRAMMAR_FILE))
else:
self.grammar = None
self._set_prompt_caching(enabled=self.entry.options.get(CONF_PROMPT_CACHING_ENABLED, DEFAULT_PROMPT_CACHING_ENABLED))
@callback
async def _async_cache_prompt(self, entity, old_state, new_state):
raw_prompt = self.entry.options.get(CONF_PROMPT, DEFAULT_PROMPT)
if self.last_updated_entities:
self.last_updated_entities
# if self.last_updated_entities:
# self.last_updated_entities
# TODO: track which entities are updated most often and sort them to the bottom of the system prompt
prompt = self._format_prompt([
@@ -651,7 +659,8 @@ class LocalLLaMAAgent(LLaMAAgent):
def _prime_llamacpp_kv_cache(self, prompt: str) -> None:
current_time = time.time()
if self.last_cache_prime and current_time - self.last_cache_prime < self.fastest_cache_prime_interval:
fastest_prime_interval = self.entry.options.get(CONF_PROMPT_CACHING_INTERVAL, DEFAULT_PROMPT_CACHING_INTERVAL)
if self.last_cache_prime and current_time - self.last_cache_prime < fastest_prime_interval:
return
with self.model_lock:

74
custom_components/llama_conversation/config_flow.py
View File

@@ -47,13 +47,17 @@ from .const import (
CONF_DOWNLOADED_MODEL_QUANTIZATION_OPTIONS,
CONF_PROMPT_TEMPLATE,
CONF_USE_GBNF_GRAMMAR,
CONF_GBNF_GRAMMAR_FILE,
CONF_EXTRA_ATTRIBUTES_TO_EXPOSE,
CONF_ALLOWED_SERVICE_CALL_ARGUMENTS,
CONF_TEXT_GEN_WEBUI_PRESET,
CONF_REFRESH_SYSTEM_PROMPT,
CONF_REMEMBER_CONVERSATION,
CONF_REMEMBER_NUM_INTERACTIONS,
CONF_PROMPT_CACHING_ENABLED,
CONF_PROMPT_CACHING_INTERVAL,
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES,
CONF_IN_CONTEXT_EXAMPLES_FILE,
CONF_OPENAI_API_KEY,
CONF_TEXT_GEN_WEBUI_ADMIN_KEY,
CONF_SERVICE_CALL_REGEX,
@@ -73,11 +77,16 @@ from .const import (
DEFAULT_DOWNLOADED_MODEL_QUANTIZATION,
DEFAULT_PROMPT_TEMPLATE,
DEFAULT_USE_GBNF_GRAMMAR,
DEFAULT_GBNF_GRAMMAR_FILE,
DEFAULT_EXTRA_ATTRIBUTES_TO_EXPOSE,
DEFAULT_ALLOWED_SERVICE_CALL_ARGUMENTS,
DEFAULT_REFRESH_SYSTEM_PROMPT,
DEFAULT_REMEMBER_CONVERSATION,
DEFAULT_REMEMBER_NUM_INTERACTIONS,
DEFAULT_PROMPT_CACHING_ENABLED,
DEFAULT_PROMPT_CACHING_INTERVAL,
DEFAULT_USE_IN_CONTEXT_LEARNING_EXAMPLES,
DEFAULT_IN_CONTEXT_EXAMPLES_FILE,
DEFAULT_SERVICE_CALL_REGEX,
DEFAULT_REMOTE_USE_CHAT_ENDPOINT,
DEFAULT_TEXT_GEN_WEBUI_CHAT_MODE,
@@ -570,9 +579,28 @@ class OptionsFlow(config_entries.OptionsFlow):
self, user_input: dict[str, Any] | None = None
) -> FlowResult:
"""Manage the options."""
errors = {}
description_placeholders = {}
if user_input is not None:
# TODO: validate that files exist (GBNF + ICL examples)
return self.async_create_entry(title="LLaMA Conversation", data=user_input)
if not user_input.get(CONF_REFRESH_SYSTEM_PROMPT) and user_input.get(CONF_PROMPT_CACHING_ENABLED):
errors["base"] = "sys_refresh_caching_enabled"
if user_input.get(CONF_USE_GBNF_GRAMMAR):
filename = user_input.get(CONF_GBNF_GRAMMAR_FILE, DEFAULT_GBNF_GRAMMAR_FILE)
if not os.path.isfile(os.path.join(os.path.dirname(__file__), filename)):
errors["base"] = "missing_gbnf_file"
description_placeholders["filename"] = filename
if user_input.get(CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES):
filename = user_input.get(CONF_IN_CONTEXT_EXAMPLES_FILE, DEFAULT_IN_CONTEXT_EXAMPLES_FILE)
if not os.path.isfile(os.path.join(os.path.dirname(__file__), filename)):
errors["base"] = "missing_icl_file"
description_placeholders["filename"] = filename
if len(errors) == 0:
return self.async_create_entry(title="LLaMA Conversation", data=user_input)
schema = local_llama_config_option_schema(
self.config_entry.options,
self.config_entry.data[CONF_BACKEND_TYPE],
@@ -580,6 +608,8 @@ class OptionsFlow(config_entries.OptionsFlow):
return self.async_show_form(
step_id="init",
data_schema=vol.Schema(schema),
errors=errors,
description_placeholders=description_placeholders,
)
@@ -616,6 +646,16 @@ def local_llama_config_option_schema(options: MappingProxyType[str, Any], backen
multiple=False,
mode=SelectSelectorMode.DROPDOWN,
)),
vol.Required(
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES,
description={"suggested_value": options.get(CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES)},
default=DEFAULT_USE_IN_CONTEXT_LEARNING_EXAMPLES,
): bool,
vol.Required(
CONF_IN_CONTEXT_EXAMPLES_FILE,
description={"suggested_value": options.get(CONF_IN_CONTEXT_EXAMPLES_FILE)},
default=DEFAULT_IN_CONTEXT_EXAMPLES_FILE,
): str,
vol.Required(
CONF_MAX_TOKENS,
description={"suggested_value": options.get(CONF_MAX_TOKENS)},
@@ -649,12 +689,8 @@ def local_llama_config_option_schema(options: MappingProxyType[str, Any], backen
vol.Optional(
CONF_REMEMBER_NUM_INTERACTIONS,
description={"suggested_value": options.get(CONF_REMEMBER_NUM_INTERACTIONS)},
default=DEFAULT_REMEMBER_NUM_INTERACTIONS,
): int,
vol.Required(
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES,
description={"suggested_value": options.get(CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES)},
default=DEFAULT_USE_IN_CONTEXT_LEARNING_EXAMPLES,
): bool,
}
if is_local_backend(backend_type):
@@ -674,11 +710,26 @@ def local_llama_config_option_schema(options: MappingProxyType[str, Any], backen
description={"suggested_value": options.get(CONF_TEMPERATURE)},
default=DEFAULT_TEMPERATURE,
): NumberSelector(NumberSelectorConfig(min=0, max=1, step=0.05)),
vol.Required(
CONF_PROMPT_CACHING_ENABLED,
description={"suggested_value": options.get(CONF_PROMPT_CACHING_ENABLED)},
default=DEFAULT_PROMPT_CACHING_ENABLED,
): bool,
vol.Required(
CONF_PROMPT_CACHING_INTERVAL,
description={"suggested_value": options.get(CONF_PROMPT_CACHING_INTERVAL)},
default=DEFAULT_PROMPT_CACHING_INTERVAL,
): NumberSelector(NumberSelectorConfig(min=1, max=60, step=1)),
vol.Required(
CONF_USE_GBNF_GRAMMAR,
description={"suggested_value": options.get(CONF_USE_GBNF_GRAMMAR)},
default=DEFAULT_USE_GBNF_GRAMMAR,
): bool
): bool,
vol.Required(
CONF_GBNF_GRAMMAR_FILE,
description={"suggested_value": options.get(CONF_GBNF_GRAMMAR_FILE)},
default=DEFAULT_GBNF_GRAMMAR_FILE,
): str
})
elif backend_type == BACKEND_TYPE_TEXT_GEN_WEBUI:
result = insert_after_key(result, CONF_MAX_TOKENS, {
@@ -761,7 +812,12 @@ def local_llama_config_option_schema(options: MappingProxyType[str, Any], backen
CONF_USE_GBNF_GRAMMAR,
description={"suggested_value": options.get(CONF_USE_GBNF_GRAMMAR)},
default=DEFAULT_USE_GBNF_GRAMMAR,
): bool
): bool,
vol.Required(
CONF_GBNF_GRAMMAR_FILE,
description={"suggested_value": options.get(CONF_GBNF_GRAMMAR_FILE)},
default=DEFAULT_GBNF_GRAMMAR_FILE,
): str
})
elif backend_type == BACKEND_TYPE_OLLAMA:
result = insert_after_key(result, CONF_MAX_TOKENS, {

57
custom_components/llama_conversation/const.py
View File

@@ -1,8 +1,5 @@
"""Constants for the LLaMa Conversation integration."""
import types
# import voluptuous as vol
# import homeassistant.helpers.config_validation as cv
# from homeassistant.const import CONF_HOST, CONF_PORT, CONF_SSL
DOMAIN = "llama_conversation"
CONF_PROMPT = "prompt"
@@ -57,8 +54,6 @@ CONF_EXTRA_ATTRIBUTES_TO_EXPOSE = "extra_attributes_to_expose"
DEFAULT_EXTRA_ATTRIBUTES_TO_EXPOSE = ["rgb_color", "brightness", "temperature", "humidity", "fan_mode", "media_title", "volume_level", "item"]
CONF_ALLOWED_SERVICE_CALL_ARGUMENTS = "allowed_service_call_arguments"
DEFAULT_ALLOWED_SERVICE_CALL_ARGUMENTS = ["rgb_color", "brightness", "temperature", "humidity", "fan_mode", "hvac_mode", "preset_mode", "item", "duration"]
GBNF_GRAMMAR_FILE = "output.gbnf"
IN_CONTEXT_EXAMPLES_FILE = "in_context_examples.csv"
CONF_PROMPT_TEMPLATE = "prompt_template"
PROMPT_TEMPLATE_CHATML = "chatml"
PROMPT_TEMPLATE_ALPACA = "alpaca"
@@ -107,16 +102,25 @@ PROMPT_TEMPLATE_DESCRIPTIONS = {
}
CONF_USE_GBNF_GRAMMAR = "gbnf_grammar"
DEFAULT_USE_GBNF_GRAMMAR = False
CONF_GBNF_GRAMMAR_FILE = "gbnf_grammar_file"
DEFAULT_GBNF_GRAMMAR_FILE = "output.gbnf"
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES = "in_context_examples"
DEFAULT_USE_IN_CONTEXT_LEARNING_EXAMPLES = True
CONF_IN_CONTEXT_EXAMPLES_FILE = "in_context_examples_file"
DEFAULT_IN_CONTEXT_EXAMPLES_FILE = "in_context_examples.csv"
CONF_TEXT_GEN_WEBUI_PRESET = "text_generation_webui_preset"
CONF_OPENAI_API_KEY = "openai_api_key"
CONF_TEXT_GEN_WEBUI_ADMIN_KEY = "text_generation_webui_admin_key"
CONF_REFRESH_SYSTEM_PROMPT = "refresh_prompt_per_tern"
CONF_REMEMBER_CONVERSATION = "remember_conversation"
CONF_REMEMBER_NUM_INTERACTIONS = "remember_num_interactions"
DEFAULT_REFRESH_SYSTEM_PROMPT = True
CONF_REMEMBER_CONVERSATION = "remember_conversation"
DEFAULT_REMEMBER_CONVERSATION = True
CONF_REMEMBER_NUM_INTERACTIONS = "remember_num_interactions"
DEFAULT_REMEMBER_NUM_INTERACTIONS = 5
CONF_PROMPT_CACHING_ENABLED = "prompt_caching"
DEFAULT_PROMPT_CACHING_ENABLED = False
CONF_PROMPT_CACHING_INTERVAL = "prompt_caching_interval"
DEFAULT_PROMPT_CACHING_INTERVAL = 30
CONF_SERVICE_CALL_REGEX = "service_call_regex"
DEFAULT_SERVICE_CALL_REGEX = r"({[\S \t]*?})"
FINE_TUNED_SERVICE_CALL_REGEX = r"```homeassistant\n([\S \t\n]*?)```"
@@ -194,42 +198,3 @@ OPTIONS_OVERRIDES = {
CONF_PROMPT_TEMPLATE: PROMPT_TEMPLATE_ZEPHYR,
}
}
# TODO: need to rewrite the internal config_entry key names so they actually make sense before we expose this
# method of configuring the component. doing so will require writing a config version upgrade migration
# MODEL_CONFIG_SCHEMA = vol.Schema(
# {
# vol.Required(CONF_BACKEND_TYPE): vol.All(
# vol.In([
# BACKEND_TYPE_LLAMA_EXISTING,
# BACKEND_TYPE_TEXT_GEN_WEBUI,
# BACKEND_TYPE_LLAMA_CPP_PYTHON_SERVER,
# BACKEND_TYPE_OLLAMA,
# BACKEND_TYPE_GENERIC_OPENAI,
# ])
# ),
# vol.Optional(CONF_HOST): cv.string,
# vol.Optional(CONF_PORT, default=DEFAULT_PORT): cv.port,
# vol.Optional(CONF_SSL, default=DEFAULT_SSL): cv.boolean,
# vol.Optional("options"): vol.Schema(
# {
# vol.Optional(CONF_PROMPT): cv.string,
# vol.Optional(CONF_PROMPT_TEMPLATE): vol.All(
# vol.In([
# PROMPT_TEMPLATE_ALPACA,
# PROMPT_TEMPLATE_CHATML,
# PROMPT_TEMPLATE_LLAMA2,
# PROMPT_TEMPLATE_MISTRAL,
# PROMPT_TEMPLATE_VICUNA,
# PROMPT_TEMPLATE_ZEPHYR,
# ])
# ),
# }
# )
# }
# )
# CONFIG_SCHEMA = vol.Schema(
# { DOMAIN: vol.All(cv.ensure_list, [MODEL_CONFIG_SCHEMA]) },
# extra=vol.ALLOW_EXTRA,
# )

17
custom_components/llama_conversation/translations/en.json
View File

@@ -60,6 +60,7 @@
"extra_attributes_to_expose": "Additional attribute to expose in the context",
"allowed_service_call_arguments": "Arguments allowed to be pass to service calls",
"gbnf_grammar": "Enable GBNF Grammar",
"gbnf_grammar_file": "GBNF Grammar Filename",
"openai_api_key": "API Key",
"text_generation_webui_admin_key": "Admin Key",
"service_call_regex": "Service Call Regex",
@@ -67,9 +68,12 @@
"remember_conversation": "Remember conversation",
"remember_num_interactions": "Number of past interactions to remember",
"in_context_examples": "Enable in context learning (ICL) examples",
"in_context_examples_file": "In context learning examples CSV filename",
"text_generation_webui_preset": "Generation Preset/Character Name",
"remote_use_chat_endpoint": "Use chat completions endpoint",
"text_generation_webui_chat_mode": "Chat Mode"
"text_generation_webui_chat_mode": "Chat Mode",
"prompt_caching": "Enable Prompt Caching",
"prompt_caching_interval": "Prompt Caching fastest refresh interval (sec)"
},
"data_description": {
"prompt": "See [here](https://github.com/acon96/home-llm/blob/develop/docs/Model%20Prompting.md) for more information on model prompting.",
@@ -98,6 +102,7 @@
"extra_attributes_to_expose": "Additional attribute to expose in the context",
"allowed_service_call_arguments": "Arguments allowed to be pass to service calls",
"gbnf_grammar": "Enable GBNF Grammar",
"gbnf_grammar_file": "GBNF Grammar Filename",
"openai_api_key": "API Key",
"text_generation_webui_admin_key": "Admin Key",
"service_call_regex": "Service Call Regex",
@@ -105,11 +110,19 @@
"remember_conversation": "Remember conversation",
"remember_num_interactions": "Number of past interactions to remember",
"in_context_examples": "Enable in context learning (ICL) examples",
"in_context_examples_file": "In context learning examples CSV filename",
"text_generation_webui_preset": "Generation Preset/Character Name",
"remote_use_chat_endpoint": "Use chat completions endpoint",
"text_generation_webui_chat_mode": "Chat Mode"
"text_generation_webui_chat_mode": "Chat Mode",
"prompt_caching": "Enable Prompt Caching",
"prompt_caching_interval": "Prompt Caching fastest refresh interval (sec)"
}
}
},
"error": {
"sys_refresh_caching_enabled": "System prompt refresh must be enabled for prompt caching to work!",
"missing_gbnf_file": "The GBNF file was not found: '{filename}'",
"missing_icl_file": "The in context learning example CSV file was not found: '{filename}'"
}
},
"selector": {