Training in progress, step 500

ipu_config.json

@@ -3,7 +3,7 @@
   "device_iterations": 1,
   "embedding_serialization_factor": 0,
   "enable_half_partials": true,
-  "executable_cache_dir": "/tmp/exe_cache",
+  "executable_cache_dir": "/tmp/exe_cache/multiple_choice",
   "execute_encoder_on_cpu_for_generation": false,
   "gradient_accumulation_steps": 16,
   "inference_device_iterations": 5,
@@ -19,7 +19,7 @@
     0.25
   ],
   "optimizer_state_offchip": true,
-  "optimum_version": "1.5.0",
+  "optimum_version": "1.5.1",
   "output_mode": "final",
   "recompute_checkpoint_every_layer": true,
   "replicated_tensor_sharding": false,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dff2e1da3933a8f97408c8ab958258391bc44a9d12e5eb5de7e04e5fdc57bc44
+oid sha256:f464647164547918c6e4315d28d75328ee9a8e6ec6a5d0f56bb7344df132251a
 size 249361645

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3851e04548301e40a03699c4c9d7a40849a200e81de3ff863127b5fb01d2e598
+oid sha256:f9d9c447cf4eabf1c5d2617323e853898ae124d4ad241965d2d1d3afc11ba4d0
 size 2735