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,
@@ -25,8 +25,8 @@
   "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:a9a1dea4d32e9e39418ba64795b63a2dc1345f2258bd78c1b0e07db36eb8a6c9
+oid sha256:ecaa6bae30103480598cb6384b973b3906a10fcad937e514e043b8558d7f4ecc
 size 121055163

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:80d34a87fcbb8b1b8fa567b16ff21010861123cf54418503785de3cfa9575955
+oid sha256:24a1c9fbba01cbc9e5b164fc0e1941a7b562aef298f0618d1dbc6880d9a35631
 size 2863