← back to Exo
dynamic halfway partition point in unit test
17065d879b0fcdfe90a09f58b5ce49f91c654cea · 2024-10-10 14:17:46 -0700 · Alex Cheema
Files touched
M exo/inference/test_inference_engine.py
Diff
commit 17065d879b0fcdfe90a09f58b5ce49f91c654cea
Author: Alex Cheema <alexcheema123@gmail.com>
Date: Thu Oct 10 14:17:46 2024 -0700
dynamic halfway partition point in unit test
---
exo/inference/test_inference_engine.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/exo/inference/test_inference_engine.py b/exo/inference/test_inference_engine.py
index 4e44a833..6a68237a 100644
--- a/exo/inference/test_inference_engine.py
+++ b/exo/inference/test_inference_engine.py
@@ -19,7 +19,7 @@ async def test_inference_engine(inference_engine_1: InferenceEngine, inference_e
inference_state=inference_state_full,
)
- pp = 15
+ pp = n_layers // 2
resp1, inference_state_1, _ = await inference_engine_1.infer_prompt("B", shard=Shard(model_id=model_id, start_layer=0, end_layer=pp, n_layers=n_layers), prompt=prompt)
resp2, inference_state_2, _ = await inference_engine_2.infer_tensor(
"B",
← ae74d2da run unit test on llama 3.2 1b for faster test
·
back to Exo
·
fix: tokenize 07305323 →