Training in progress, epoch 1

ipu_config.json

@@ -1,8 +1,7 @@
 {
-  "decompose_grad_sum": false,
+  "auto_loss_scaling": false,
   "device_iterations": 1,
   "embedding_serialization_factor": 0,
-  "enable_half_first_order_momentum": true,
   "enable_half_partials": true,
   "executable_cache_dir": "/tmp/exe_cache",
   "execute_encoder_on_cpu_for_generation": false,
@@ -20,14 +19,12 @@
     0.25
   ],
   "optimizer_state_offchip": true,
-  "optimum_version": "1.4.1",
+  "optimum_version": "1.5.0",
   "output_mode": "final",
-  "profile_dir": "",
   "recompute_checkpoint_every_layer": true,
   "replicated_tensor_sharding": false,
   "replication_factor": 1,
   "seed": 42,
   "sharded_execution_for_inference": false,
-  "transformers_version": "4.20.0",
-  "use_popdist": false
+  "transformers_version": "4.20.0"
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4aa74af1e92f29a5a60ec1c20ef472eac3bd6aa6bd7ff9285277b5d0156ca21f
+oid sha256:2e279809a3aaa8610d8699c0ddc52539a9303fbd18845cbf0b0c4d09a3987039
 size 249363181

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:220c979b94fe9594eb3e8fd25dc7e1a750b0ff38441ca60aa8eb0e747d80f1a6
-size 2671
+oid sha256:6270486d18e3477bcb79bfac8e41c729302b43d0de50ab4bac50d45c6cc74eb8
+size 2735