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": 1,
+  "inference_replication_factor": 4,
   "ipus_per_replica": 4,
   "layers_per_ipu": [
     0,
@@ -20,8 +20,8 @@
   "optimum_version": "1.5.1",
   "output_mode": "final",
   "recompute_checkpoint_every_layer": true,
-  "replicated_tensor_sharding": false,
-  "replication_factor": 1,
+  "replicated_tensor_sharding": true,
+  "replication_factor": 4,
   "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 sha256:67fb639ec8abb1464b958e3ffa80b1ba1584db82d204be07418971fbded4072e
+oid sha256:1c2f4d1a97f0dc32186e17f230187b2f81484292670021cb1ff3f5d62df52462
 size 217864753

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a9ab068517bde9174282280dc6d7503e098228625b30ac49a794d46f10868a2c
+oid sha256:fbd29fa5698fd2d8e75633cd7b2a84d6db0461aabe6afe006c6e1cef9a221fec
 size 2735