Training in progress, step 500

ipu_config.json

@@ -7,7 +7,7 @@
   "execute_encoder_on_cpu_for_generation": false,
   "gradient_accumulation_steps": 32,
   "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 sha256:7241eef3c21b772e82b9f75d2ccbc71d120e03db018f8e8d689a7a048a6a16f6
+oid sha256:3d31d8999032ce876523e18d2a975eba339d0963be6c0856c54b789752179c3f
 size 248181489

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:00bf4f2eec2d78114847e12ee7bbefe929fd0ac967682dc4811cfd2bf668a141
+oid sha256:0e59e09e32a5bd2a07f0bf6ed2e257a36d4722870b8bd19642798254c274c9d8
 size 2735