Upload 10 files

config.json

@@ -33,11 +33,10 @@
   "esmc_config": {
     "attention_bias": false,
     "attention_dropout": 0.0,
-    "bos_token_id": null,
+    "bos_token_id": 0,
     "classifier_dropout": 0.1,
     "dtype": "float32",
-    "eos_token_id": null,
-    "expansion_ratio": 2.6666666666666665,
+    "eos_token_id": 2,
     "head_dim": 64,
     "hidden_act": "silu",
     "hidden_size": 2560,
@@ -55,7 +54,6 @@
       "rope_theta": 10000.0,
       "rope_type": "default"
     },
-    "scale_residue": true,
     "tie_word_embeddings": false,
     "vocab_size": 64
   },
@@ -72,7 +70,6 @@
   "max_chars": 4,
   "model_type": "esmfold2",
   "msa_encoder": {
-    "divide_outer_before_proj": false,
     "head_dim": 16,
     "hidden_size": 128,
     "intermediate_size": 512,
@@ -80,8 +77,7 @@
     "num_attention_heads": 8,
     "num_hidden_layers": 4,
     "outer_hidden_size": 32,
-    "outer_product_chunk_size": null,
-    "overwrite": true
+    "outer_product_chunk_size": null
   },
   "num_loops": 3,
   "num_relative_chain_bins": 2,
@@ -164,5 +160,5 @@
     "num_distogram_bins": 64,
     "step_scale": 1.5
   },
-  "transformers_version": "5.15.0.dev0"
+  "transformers_version": "5.16.0.dev0"
 }

model-00001-of-00006.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdd94e62ad71ed144ebdc5561d07dd8ace66aec0b195928792e15ef784ce8cbd
-size 4955672068
+oid sha256:a24f050e2a2f9746b43b435c53719afc575a374fa302731b6911356a3d8b5cac
+size 4955672788

model.safetensors.index.json

@@ -966,34 +966,34 @@
     "folding_trunk.layers.9.tri_mul_out.proj_bundle.weight": "model-00001-of-00006.safetensors",
     "folding_trunk.layers.9.tri_mul_out.proj_emit.weight": "model-00001-of-00006.safetensors",
     "folding_trunk.layers.9.tri_mul_out.proj_gate.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.atom_linear.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.atom_norm.bias": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.atom_norm.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.atom_to_token_linear.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.adaln_linear.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.adaln_linear.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.adaln_linear.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "inputs_atom_encoder.layers.2.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.atom_linear.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.atom_norm.bias": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.atom_norm.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.atom_to_token_linear.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.adaln_linear.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.adaln_linear.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.adaln_linear.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.atom_encoder.layers.2.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
     "language_model.layer_weights": "model-00001-of-00006.safetensors",
     "language_model.pair_input_norm.bias": "model-00001-of-00006.safetensors",
     "language_model.pair_input_norm.weight": "model-00001-of-00006.safetensors",
@@ -1250,7 +1250,7 @@
     "parcae.log_state_decay": "model-00001-of-00006.safetensors",
     "parcae.log_delta": "model-00001-of-00006.safetensors",
     "parcae.out_proj.weight": "model-00001-of-00006.safetensors",
-    "rel_pos.embed.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.rel_pos.embed.weight": "model-00001-of-00006.safetensors",
     "structure_head.atom_decoder.layers.0.adaln_linear.weight": "model-00001-of-00006.safetensors",
     "structure_head.atom_decoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
     "structure_head.atom_decoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
@@ -1307,7 +1307,7 @@
     "structure_head.atom_encoder.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
     "structure_head.atom_encoder.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
     "structure_head.atom_encoder.layers.2.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.atom_encoder.coords_linear.weight": "model-00001-of-00006.safetensors",
+    "structure_head.coords_linear.weight": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.fourier.phases": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.fourier.frequencies": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.noise_norm.bias": "model-00001-of-00006.safetensors",
@@ -1316,309 +1316,309 @@
     "structure_head.conditioning.single_input_norm.bias": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.single_input_norm.weight": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.single_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.0.norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.0.norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.1.norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.1.norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_0.norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_0.norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_1.norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_1.norm.weight": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.pair_input_norm.bias": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.pair_input_norm.weight": "model-00001-of-00006.safetensors",
     "structure_head.conditioning.pair_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.0.norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.0.norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.1.norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.1.norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_0.norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_0.norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_1.norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_1.norm.weight": "model-00001-of-00006.safetensors",
     "structure_head.single_step_norm.bias": "model-00001-of-00006.safetensors",
     "structure_head.single_step_norm.weight": "model-00001-of-00006.safetensors",
     "structure_head.single_to_token.weight": "model-00001-of-00006.safetensors",
     "structure_head.token_norm.bias": "model-00001-of-00006.safetensors",
     "structure_head.token_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.k_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.v_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.out_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.out_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.o_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.pair_norm.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.pair_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.q_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.attn.q_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.0.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.1.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.10.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.11.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.2.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.3.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.4.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.5.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.6.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.7.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.8.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.adaln.gate_proj.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.adaln.gate_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.adaln.cond_norm.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.adaln.shift_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.output_gate.bias": "model-00001-of-00006.safetensors",
-    "structure_head.token_transformer.layers.9.transition.output_gate.weight": "model-00001-of-00006.safetensors",
-    "token_bonds.weight": "model-00001-of-00006.safetensors",
-    "pair_init_1.weight": "model-00001-of-00006.safetensors",
-    "pair_init_2.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.single_transitions.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
-    "structure_head.conditioning.pair_transitions.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.input_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.input_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.input_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.input_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.self_attn.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.attn_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.attn_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.pair_bias_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.pair_norm.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.pair_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.self_attn.q_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.0.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.1.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.10.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.11.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.2.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.3.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.4.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.5.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.6.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.7.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.8.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.post_attention_layernorm.gate_proj.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.post_attention_layernorm.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.post_attention_layernorm.cond_norm.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.post_attention_layernorm.shift_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.mlp_gate.bias": "model-00001-of-00006.safetensors",
+    "structure_head.token_transformer.layers.9.mlp_gate.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.token_bonds.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.pair_init_1.weight": "model-00001-of-00006.safetensors",
+    "input_embedder.pair_init_2.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.single_transition_1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_0.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_1.mlp.gate_up_proj.weight": "model-00001-of-00006.safetensors",
+    "structure_head.conditioning.pair_transition_1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
     "esmc.embed_tokens.weight": "model-00001-of-00006.safetensors",
     "esmc.layers.0.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
     "esmc.layers.0.input_layernorm.bias": "model-00001-of-00006.safetensors",