Merge_PR_418_refine_hf_fallback_policy_and_sif_low_cost_routing

This commit is contained in:
ajaysi
2026-03-12 16:19:19 +05:30
3 changed files with 126 additions and 6 deletions

View File

@@ -38,8 +38,12 @@ class SharedLLMWrapper:
return llm_text_gen(
prompt,
user_id=self.user_id,
<<<<<<< HEAD
preferred_hf_models=LOW_COST_SHARED_REMOTE_MODELS,
flow_type="sif_agent",
=======
preferred_hf_models=REMOTE_LOW_COST_HF_MODELS,
>>>>>>> pr-418
)
except Exception as e:
logger.error(f"SharedLLMWrapper failed to generate text: {e}")
@@ -50,7 +54,12 @@ class SharedLLMWrapper:
_local_llm_cache = {}
<<<<<<< HEAD
LOW_COST_SHARED_REMOTE_MODELS = [
=======
REMOTE_LOW_COST_HF_MODELS = [
>>>>>>> pr-418
"Qwen/Qwen2.5-1.5B-Instruct",
"Qwen/Qwen2.5-0.5B-Instruct",
"TinyLlama/TinyLlama-1.1B-Chat-v1.0",