Training in progress, step 500

config.json

@@ -5,7 +5,7 @@
   "add_bias_logits": false,
   "add_final_layer_norm": false,
   "architectures": [
-    "PipelinedBartForConditionalGeneration"
+    "PoptorchPipelinedBartForConditionalGeneration"
   ],
   "attention_dropout": 0.1,
   "bos_token_id": 0,

ipu_config.json

@@ -7,7 +7,7 @@
   "execute_encoder_on_cpu_for_generation": false,
   "gradient_accumulation_steps": 128,
   "inference_device_iterations": 6,
-  "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 sha256:4f1b528f7fdc8c91a7d4825ee56284e7c31ff7e3fcb03e9ba880d0319f09bd70
+oid sha256:c507c4050dfb4185cc0bb54322e5cfdb6be3660943af02f8b6c099fa45b91ddd
 size 279024057

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:802eefe4468c1e4ef5891245da228816c7e57129ea80c628441da04edaf49045
+oid sha256:d7e6e5147211ca70e7fac2b61c0965dfb393135d32bd52efb6e02dea5f4f0540
 size 2863