step 60

README.md

@@ -0,0 +1,10 @@
+---
+license: apache-2.0
+library_name: pytorch
+---
+
+# lobotomy-1b
+
+21-layer ~1B student from Qwen3.8-27B (64→21, d=5120→1536).
+
+step=60

config.json

@@ -0,0 +1,27 @@
+{
+  "step": 60,
+  "cfg": {
+    "vocab_size": 248320,
+    "hidden_size": 1536,
+    "n_layers": 21,
+    "n_heads": 12,
+    "n_kv_heads": 4,
+    "linear_heads": 12,
+    "head_dim": 128,
+    "ffn_hidden": 4096,
+    "hybrid_ratio": 3,
+    "linear": "gdn",
+    "rms_norm_eps": 1e-06,
+    "tie_embeddings": true,
+    "conv_size": 4,
+    "expand_v": 1.0,
+    "use_fla": false,
+    "logit_softcap": 15.0,
+    "mtp": 1
+  },
+  "progress": {
+    "next_shard_i": 3,
+    "step": 61,
+    "last_shard": "shard_00003.safetensors"
+  }
+}

downproj.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:029b356d1fbea598d518ba46738af11c03c11337368ba7d939f6662c4fc04566
+size 31457368

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70abbf8b309794bd11cdeba11c8f2e8207835584a8422c1d6d6959eba8d0c80a
+size 2698072840