Skip to content

Automation foundation model for tiny devices: 2-bit, 8-29 MB, tool calls, ASR, structured extraction and embeddings on phones, wearables, smart homes, robots, cars and microcontrollers.

change post-training defaults

HenryNdubuaku646ebdcb35f1
diff --git a/src/cli.py b/src/cli.py
index 2103e2e..12c7186 100644
--- a/src/cli.py
+++ b/src/cli.py
@@ -94,10 +94,6 @@ def _install_xla_log_filter():
t.start()
-# Install the stderr filter BEFORE importing anything that might transitively
-# initialize jax/XLA. In particular, `.data` imports `datasets`, which probes
-# for jax at import time and fully loads it if present — that initializes XLA's
-# C++ absl logging and captures fd 2 before we get a chance to redirect it.
os.environ.setdefault("TF_CPP_MIN_LOG_LEVEL", "2")
os.environ.setdefault("GRPC_VERBOSITY", "ERROR")
_install_xla_log_filter()
@@ -123,7 +119,7 @@ def main():
"(e.g. needle_base.pkl). Uses CLI config; partial-loads "
"matching params, random-inits the rest.")
p.add_argument("--epochs", type=int, default=1)
- p.add_argument("--batch-size", type=int, default=32)
+ p.add_argument("--batch-size", type=int, default=64)
p.add_argument("--lr", type=float, default=3e-4)
p.add_argument("--muon-lr", type=float, default=0.02)
p.add_argument("--d-model", type=int, default=512)