Training in progress, step 1000

config.json

@@ -7,11 +7,11 @@
   ],
   "attend_to_previous_input": false,
   "base_model_config": null,
-  "base_model_name": "meta-llama/Llama-3.2-1B",
+  "base_model_name": "HuggingFaceTB/SmolLM2-360M",
   "constant_depth": false,
   "correction": true,
   "d_mem": 64,
-  "dtype": "bfloat16",
+  "dtype": "float32",
   "freeze_mem": false,
   "gating": false,
   "layers_attr": "model.layers",
@@ -21,15 +21,12 @@
   "noisy_halting": false,
   "num_mem_tokens": 32,
   "segment_alignment": "left",
-  "segment_size": 1024,
-  "sliding_window": true,
+  "segment_size": 256,
+  "sliding_window": false,
   "time_penalty": 0.0,
   "transformers_version": "4.57.1",
   "use_denom": true,
-  "use_sink": true,
-  "wrap_pos": false,
-  "auto_map": {
-    "AutoConfig": "modeling_armt.ARMTConfig",
-    "AutoModelForCausalLM": "modeling_armt.InnerLoopARMTForCausalLM"
-  }
-}
+  "use_sink": false,
+  "wrap_layers": null,
+  "wrap_pos": false
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fc53384a64c72342150a186244c88b3b4777c9920420e0559fbc7807f30c4fd
+size 1581378582

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f611e06634e71a81b71d6dabce59e72a77202abdab67535b76daec6cb4fc1b4
-size 6904
+oid sha256:1368ed84377993a4b44fe26324df7269812af0c8ec550abe9eadd6fe320805ca
+size 5624