idchlife/rag-solution
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix for openai compatibility model with needed parameters

Browse Source
This commit is contained in:
idchlife
2026-03-09 10:21:39 +03:00
parent 969d25209c
commit 5721bad117
4 changed files with 39 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
services/rag/llamaindex/config.py
View File

@@ -93,12 +93,19 @@ def get_llm_model():
return llm
elif strategy == "openai":
from llama_index.llms.openai_like import OpenAILike
# from helpers.openai_compatible_llm import OpenAICompatibleLLM
from helpers.openai_compatible_llm import OpenAICompatibleLLM
openai_chat_url = os.getenv("OPENAI_CHAT_URL", "https://api.openai.com/v1")
openai_chat_key = os.getenv("OPENAI_CHAT_KEY", "dummy_key_for_template")
openai_chat_model = os.getenv("OPENAI_CHAT_MODEL", "gpt-3.5-turbo")
openai_chat_temperature = float(os.getenv("OPENAI_CHAT_TEMPERATURE", "0.1"))
openai_chat_max_tokens_env = os.getenv("OPENAI_CHAT_MAX_TOKENS", "").strip()
openai_chat_max_tokens = (
int(openai_chat_max_tokens_env) if openai_chat_max_tokens_env else 1024
)
openai_reasoning_effort = (
os.getenv("OPENAI_CHAT_REASONING_EFFORT", "").strip() or None
)
openai_is_fc_model = (
os.getenv("OPENAI_CHAT_IS_FUNCTION_CALLING_MODEL", "false").lower()
== "true"
@@ -109,13 +116,19 @@ def get_llm_model():
logger.info(
f"Initializing OpenAI-compatible chat model: {openai_chat_model} "
f"(base={openai_chat_url}, function_calling={openai_is_fc_model})"
f"(base={openai_chat_url}, max_tokens={openai_chat_max_tokens}, "
f"reasoning_effort={openai_reasoning_effort}, function_calling={openai_is_fc_model})"
)
llm = OpenAILike(
llm = OpenAICompatibleLLM(
model=openai_chat_model,
api_base=openai_chat_url,
api_key=openai_chat_key,
temperature=openai_chat_temperature,
max_tokens=openai_chat_max_tokens,
reasoning_effort=openai_reasoning_effort,
timeout=120.0,
is_function_calling_model=openai_is_fc_model,
)
return llm