Training in progress, epoch 10

config.json

@@ -4,18 +4,18 @@
   ],
   "batch_size": 4,
   "bos_token_id": 1,
-  "d_ff": 1024,
+  "d_ff": 512,
   "d_model": 1024,
-  "dropout": 0.13129617833969415,
+  "dropout": 0.13619667425956658,
   "eos_token_id": 2,
-  "model_size": 81697026,
+  "model_size": 105627906,
   "model_type": "custom_code",
   "n_heads": 16,
-  "n_layers": 2,
+  "n_layers": 3,
   "num_train_epochs": 20,
   "pad_token_id": 0,
   "router_dim": 10,
-  "sequence_length": 128,
+  "sequence_length": 64,
   "source_sequence_dimension": 70,
   "target_sequence_dimension": 306,
   "torch_dtype": "float32",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5cdcd2e137e1b5fdf92d760c2be873e2b3f1aa7f37bde4b653a562dd8d2ef0b0
-size 326804928
+oid sha256:c401adcd171ceaac8bd9e0845b7661d85538d010a74c55ed814bc2d0873bf641
+size 422536616

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b59d643fb47105b5f85b598a0cb6beda4cd27a5bded38d9410678baf60529918
+oid sha256:159967394442ef4416aa90ecbee137ab3ec1bd4be825e4d6cb189ca9244b56f9
 size 5304