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": "1.5.0",
   "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:a9425cf264ac71692457adb7c930645eec53d8f58bfda5c9789d2c63a87d31c2
+oid sha256:e624c9769cbe8abfe9424c42d5ef9c75b35b7ccec2efe0b06de110f284441104
 size 249361645

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba57c7dea9da0bf3614cac07c3c31449da75de42e4e9968b6245ca65a5306f7b
+oid sha256:3851e04548301e40a03699c4c9d7a40849a200e81de3ff863127b5fb01d2e598
 size 2735