← back to Exo
Some session method cleanup
b22c21ac16fd9cb46a8d6a0d57e9a630f6408cba · 2024-12-08 02:35:45 -0800 · Nel Nibcord
Files touched
M exo/inference/inference_engine.py
Diff
commit b22c21ac16fd9cb46a8d6a0d57e9a630f6408cba
Author: Nel Nibcord <blindcrone@tuta.io>
Date: Sun Dec 8 02:35:45 2024 -0800
Some session method cleanup
---
exo/inference/inference_engine.py | 6 +-----
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/exo/inference/inference_engine.py b/exo/inference/inference_engine.py
index ad14bf8c..7c71e6ef 100644
--- a/exo/inference/inference_engine.py
+++ b/exo/inference/inference_engine.py
@@ -35,12 +35,8 @@ class InferenceEngine(ABC):
if hook is not None:
hook()
- async def ensure_session_match(self, key, check, value_gen):
- if key not in self.session or not check(self.session[key]):
- await self.save_session(key, value_gen())
-
async def clear_session(self):
- session.empty()
+ self.session.empty()
async def infer_prompt(self, request_id: str, shard: Shard, prompt: str) -> np.ndarray:
tokens = await self.encode(shard, prompt)
← 98edb393 Initialize inference engine session in base class
·
back to Exo
·
Do we need casting here? 59af2dd5 →