phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-25 10:44:24 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix(alerting.py): fix datetime comparison logic

Browse source
This commit is contained in:
Krrish Dholakia 2024-05-14 22:09:33 -07:00
parent 0bac40b0f2
commit 54587db402
4 changed files with 37 additions and 18 deletions
Download patch file Download diff file Expand all files Collapse all files

9
litellm/utils.py
View file

@ -59,9 +59,7 @@ from importlib import resources
with resources.open_text("litellm.llms.tokenizers", "anthropic_tokenizer.json") as f:
json_data = json.load(f)
# Convert to str (if necessary)
json_str = json.dumps(json_data)
claude_tokenizer = Tokenizer.from_str(json_str)
cohere_tokenizer = Tokenizer.from_pretrained("Xenova/c4ai-command-r-v01-tokenizer")
claude_json_str = json.dumps(json_data)
import importlib.metadata
from ._logging import verbose_logger
from .types.router import LiteLLM_Params
@ -3856,12 +3854,15 @@ def get_replicate_completion_pricing(completion_response=None, total_time=0.0):
@lru_cache(maxsize=128)
def _select_tokenizer(model: str):
global claude_tokenizer, cohere_tokenizer
if model in litellm.cohere_models and "command-r" in model:
# cohere
cohere_tokenizer = Tokenizer.from_pretrained(
"Xenova/c4ai-command-r-v01-tokenizer"
)
return {"type": "huggingface_tokenizer", "tokenizer": cohere_tokenizer}
# anthropic
elif model in litellm.anthropic_models and "claude-3" not in model:
claude_tokenizer = Tokenizer.from_str(claude_json_str)
return {"type": "huggingface_tokenizer", "tokenizer": claude_tokenizer}
# llama2
elif "llama-2" in model.lower() or "replicate" in model.lower():