Training in progress, step 2000

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:77372a6d52c655e4dece9a267b336001382a5ea2e303e1fc9b17fc92db04b3bb
+oid sha256:3d1467cc7d81b669faa608f96f0e48ecce3eb6d6824347594c97fcf770153c16
 size 1527847357

run.sh

@@ -12,7 +12,7 @@ deepspeed run-ba.py \
 	--torch_compile_mode="reduce-overhead" \
 	--torch_compile_mode="max-autotune" \
 	--logging_steps="25" \
-	--learning_rate="1e-5" \
+	--learning_rate="5e-6" \
 	--max_steps="5000" \
 	--output_dir="./" \
 	--per_device_train_batch_size="32" \

runs/Dec21_16-27-53_129-146-176-120/events.out.tfevents.1671640138.129-146-176-120.405220.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:797ce9ce2e24095627d5c6440c64d62b8158d481fae06ae23c654ded7e67f093
-size 10883
+oid sha256:f2a934041ca11bdc3b376be8151aa5626d848410d8e220cc849448be4d9a3815
+size 17481