Training in progress, step 500

config.json

@@ -2,7 +2,7 @@
   "_name_or_path": "gpt2",
   "activation_function": "gelu_new",
   "architectures": [
-    "PipelinedGPT2LMHeadModel"
+    "PoptorchPipelinedGPT2LMHeadModel"
   ],
   "attn_pdrop": 0.1,
   "bos_token_id": 50256,

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": 4,
+  "inference_replication_factor": 1,
   "ipus_per_replica": 4,
   "layers_per_ipu": [
     0,
@@ -20,8 +20,8 @@
   "optimum_version": "1.5.1",
   "output_mode": "final",
   "recompute_checkpoint_every_layer": true,
-  "replicated_tensor_sharding": true,
-  "replication_factor": 4,
+  "replicated_tensor_sharding": false,
+  "replication_factor": 1,
   "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 sha256:2c198b53f4c8c7a6525d3e27ecb5c7126b9c0916c930ad6cc8b357ec5a5cedb2
+oid sha256:98be7a37645b50ad99a01b1d44d9ab4abd27738bff597b76c13b02edde1f3ed6
 size 261514473

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c572f29b5f38b3164f2f5583aac51a73d15af505fe6dcee703da1e38943e77e0
+oid sha256:bf23742a20b2521133935ea0977f2af7132a835720c09d5b20461b7412099489
 size 2735