test

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": ".",
+  "_name_or_path": "./",
   "architectures": [
     "XLMRobertaForMaskedLM"
   ],

events.out.tfevents.1664916736.t1v-n-b899420b-w-0.278543.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf1e85eff2353955c51a626f18500c01e7dec159facddefb6bb22529c33a40bd
+size 40

events.out.tfevents.1664917260.t1v-n-b899420b-w-0.281577.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ea148d2a1e3ef547244be3eea7fc176b551fc01c2e9700388697ed480fc3805
+size 367912

events.out.tfevents.1664925833.t1v-n-b899420b-w-0.291698.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d3ad12a8242e1c39bbf79009716cdcff41d66500383a388c68874029bd8aaf7
+size 1398158

events.out.tfevents.1664951079.t1v-n-b899420b-w-0.315648.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10a7aac4dddee5fdbd66befd292d0052ea2cc31baa879111903e5843f367d38c
+size 40

flax_model.msgpack

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e176e8c732d80403ad8c03cdc305b5a1897e92ea407df51ef1cb427415e52299
+size 1113187999

run.sh

@@ -13,7 +13,7 @@ python run_mlm_flax_stream.py \
     --learning_rate="3e-4" \
     --warmup_steps="1000" \
     --overwrite_output_dir \
-    --num_train_steps="10000" \
+    --num_train_steps="250000" \
     --adam_beta1="0.9" \
     --adam_beta2="0.98" \
     --logging_steps="500" \

run_mlm_flax_stream.py

@@ -628,13 +628,20 @@ if __name__ == "__main__":
                 print("*** Printing debug info")
                 print(training_args.output_dir)
                 print(training_args.push_to_hub)
-                breakpoint()
-                model.save_pretrained(
-                    training_args.output_dir,
-                    params=params,
-                    push_to_hub=training_args.push_to_hub,
-                    commit_message=f"Saving weights and logs of step {step+1}",
-                )
+                try:
+                    model.save_pretrained(
+                        training_args.output_dir,
+                        params=params,
+                        push_to_hub=training_args.push_to_hub,
+                        commit_message=f"Saving weights and logs of step {step+1}",
+                    )
+                except:
+
+                    model.save_pretrained(
+                        training_args.output_dir,
+                        params=params
+                    )
+                    print("probems pushing")
 
         # update tqdm bar
         steps.update(1)