gavindiaz/NexQuant
1
0
Fork 0
mirror of https://github.com/NicolasBohn/NexQuant.git synced 2026-07-28 07:57:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

chore: move use_auto_chat_cache_seed_gen and init_chat_cache_seed to LLMSett… (#444)

Browse Source 
* move use_auto_chat_cache_seed_gen and init_chat_cache_seed to LLMSettings

* remove RD_AGENT_SETTINGS in llm_utils.py
This commit is contained in:
XianBW
2024-10-21 12:03:59 +08:00
committed by GitHub
parent 64df9bec37
commit 671d4f0600
5 changed files with 19 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rdagent/core/conf.py
-10
View File
@@ -15,16 +15,6 @@ class RDAgentSettings(BaseSettings):
# TODO: (xiao) think it can be a separate config.
log_trace_path: str | None = None
# Behavior of returning answers to the same question when caching is enabled
use_auto_chat_cache_seed_gen: bool = False
"""
`_create_chat_completion_inner_function` provdies a feature to pass in a seed to affect the cache hash key
We want to enable a auto seed generator to get different default seed for `_create_chat_completion_inner_function`
if seed is not given.
So the cache will only not miss you ask the same question on same round.
"""
init_chat_cache_seed: int = 42
# azure document intelligence configs
azure_document_intelligence_key: str = ""
azure_document_intelligence_endpoint: str = ""
rdagent/core/utils.py
+2 -1
View File
@@ -14,6 +14,7 @@ from filelock import FileLock
from fuzzywuzzy import fuzz # type: ignore[import-untyped]
from rdagent.core.conf import RD_AGENT_SETTINGS
from rdagent.oai.llm_conf import LLM_SETTINGS
class RDAgentException(Exception): # noqa: N818
@@ -98,7 +99,7 @@ class CacheSeedGen:
"""
def __init__(self) -> None:
self.set_seed(RD_AGENT_SETTINGS.init_chat_cache_seed)
self.set_seed(LLM_SETTINGS.init_chat_cache_seed)
def set_seed(self, seed: int) -> None:
random.seed(seed)
rdagent/oai/llm_conf.py
+10
View File
@@ -20,6 +20,16 @@ class LLMSettings(BaseSettings):
prompt_cache_path: str = str(Path.cwd() / "prompt_cache.db")
max_past_message_include: int = 10
# Behavior of returning answers to the same question when caching is enabled
use_auto_chat_cache_seed_gen: bool = False
"""
`_create_chat_completion_inner_function` provdies a feature to pass in a seed to affect the cache hash key
We want to enable a auto seed generator to get different default seed for `_create_chat_completion_inner_function`
if seed is not given.
So the cache will only not miss you ask the same question on same round.
"""
init_chat_cache_seed: int = 42
# Chat configs
openai_api_key: str = "" # TODO: simplify the key design.
chat_openai_api_key: str = ""
rdagent/oai/llm_utils.py
+1 -2
View File
@@ -17,7 +17,6 @@ from typing import Any, Optional
import numpy as np
import tiktoken
from rdagent.core.conf import RD_AGENT_SETTINGS
from rdagent.core.utils import LLM_CACHE_SEED_GEN, SingletonBaseClass
from rdagent.log import LogColors
from rdagent.log import rdagent_logger as logger
@@ -596,7 +595,7 @@ class APIBackend:
To make retries useful, we need to enable a seed.
This seed is different from `self.chat_seed` for GPT. It is for the local cache mechanism enabled by RD-Agent locally.
"""
if seed is None and RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen:
if seed is None and LLM_SETTINGS.use_auto_chat_cache_seed_gen:
seed = LLM_CACHE_SEED_GEN.get_next_seed()
# TODO: we can add this function back to avoid so much `self.cfg.log_llm_chat_content`
test/oai/test_completion.py
+6 -8
View File
@@ -60,7 +60,6 @@ class TestChatCompletion(unittest.TestCase):
- 2 pass
- cache is not missed & same question get different answer.
"""
from rdagent.core.conf import RD_AGENT_SETTINGS
from rdagent.core.utils import LLM_CACHE_SEED_GEN
from rdagent.oai.llm_conf import LLM_SETTINGS
@@ -68,7 +67,7 @@ class TestChatCompletion(unittest.TestCase):
user_prompt = f"Give me {2} random country names, list {2} cities in each country, and introduce them"
origin_value = (
RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_chat_cache,
LLM_SETTINGS.dump_chat_cache,
)
@@ -76,7 +75,7 @@ class TestChatCompletion(unittest.TestCase):
LLM_SETTINGS.use_chat_cache = True
LLM_SETTINGS.dump_chat_cache = True
RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen = True
LLM_SETTINGS.use_auto_chat_cache_seed_gen = True
LLM_CACHE_SEED_GEN.set_seed(10)
response1 = APIBackend().build_messages_and_create_chat_completion(
@@ -110,7 +109,7 @@ class TestChatCompletion(unittest.TestCase):
# Reset, for other tests
(
RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_chat_cache,
LLM_SETTINGS.dump_chat_cache,
) = origin_value
@@ -132,7 +131,6 @@ class TestChatCompletion(unittest.TestCase):
- 2 pass
- cache is not missed & same question get different answer.
"""
from rdagent.core.conf import RD_AGENT_SETTINGS
from rdagent.core.utils import LLM_CACHE_SEED_GEN, multiprocessing_wrapper
from rdagent.oai.llm_conf import LLM_SETTINGS
@@ -140,7 +138,7 @@ class TestChatCompletion(unittest.TestCase):
user_prompt = f"Give me {2} random country names, list {2} cities in each country, and introduce them"
origin_value = (
RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_chat_cache,
LLM_SETTINGS.dump_chat_cache,
)
@@ -148,7 +146,7 @@ class TestChatCompletion(unittest.TestCase):
LLM_SETTINGS.use_chat_cache = True
LLM_SETTINGS.dump_chat_cache = True
RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen = True
LLM_SETTINGS.use_auto_chat_cache_seed_gen = True
func_calls = [(_worker, (system_prompt, user_prompt)) for _ in range(4)]
@@ -161,7 +159,7 @@ class TestChatCompletion(unittest.TestCase):
# Reset, for other tests
(
RD_AGENT_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_auto_chat_cache_seed_gen,
LLM_SETTINGS.use_chat_cache,
LLM_SETTINGS.dump_chat_cache,
) = origin_value