Training in progress, step 500

ipu_config.json

@@ -7,7 +7,7 @@
   "execute_encoder_on_cpu_for_generation": false,
   "gradient_accumulation_steps": 16,
   "inference_device_iterations": 5,
-  "inference_replication_factor": 4,
+  "inference_replication_factor": 1,
   "ipus_per_replica": 4,
   "layers_per_ipu": [
     0,
@@ -22,8 +22,8 @@
   "optimum_version": "1.5.1",
   "output_mode": "final",
   "recompute_checkpoint_every_layer": true,
-  "replicated_tensor_sharding": true,
-  "replication_factor": 4,
+  "replicated_tensor_sharding": false,
+  "replication_factor": 1,
   "seed": 42,
   "sharded_execution_for_inference": false,
   "transformers_version": "4.20.1"

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:63db7b83a7ce5ef95c5e8aa4fc5630338e658f9f10a901f594e941cdd886007f
+oid sha256:15c9ed3a50283db3dd7eb9818765187ed5197cf8de7ee17741eeddb8c3581419
 size 249361645

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a31b3006c15e01af8dfe5181367d2c8d64c14d1529505cbc74c5d5ec38d13b7f
+oid sha256:a4c863f254a9cb5a89dd86e3f933d072dc799d9b540901a495b06ca5037bedcf
 size 2735