Training in progress, step 500

config.json

@@ -1,7 +1,7 @@
 {
   "_name_or_path": "t5-small",
   "architectures": [
-    "PipelinedT5ForConditionalGeneration"
+    "PoptorchPipelinedT5ForConditionalGeneration"
   ],
   "d_ff": 2048,
   "d_kv": 64,

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 sha256:3ec335dab050e57402ca0bd3b4755afd707441afd03ec2d35509d971eb68e67d
+oid sha256:863f382c9c2e724d4fc0d8c86d61a2aa6242c9caa97d75a75efd548701fcc837
 size 121055163

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e4e84e695ee42cd997024486b034ec2de8a11062128e05a51dcb7af5eff00e6
+oid sha256:8ebf6d746d8d8fa30027fcac36f3bd697091e0dda956ff14af93a89411637049
 size 2863