Upload MultiHead_MultiLabel_XL

config.json

@@ -0,0 +1,109 @@
+{
+  "_name_or_path": "/home/vasters/touche24/touche-Sotiris/xlm-roberta-xl-38-layers-5e-6-all-dir/checkpoint-74580",
+  "architectures": [
+    "MultiHead_MultiLabel_XL"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 2560,
+  "id2label": {
+    "0": "Self-direction: thought attained",
+    "1": "Self-direction: thought constrained",
+    "2": "Self-direction: action attained",
+    "3": "Self-direction: action constrained",
+    "4": "Stimulation attained",
+    "5": "Stimulation constrained",
+    "6": "Hedonism attained",
+    "7": "Hedonism constrained",
+    "8": "Achievement attained",
+    "9": "Achievement constrained",
+    "10": "Power: dominance attained",
+    "11": "Power: dominance constrained",
+    "12": "Power: resources attained",
+    "13": "Power: resources constrained",
+    "14": "Face attained",
+    "15": "Face constrained",
+    "16": "Security: personal attained",
+    "17": "Security: personal constrained",
+    "18": "Security: societal attained",
+    "19": "Security: societal constrained",
+    "20": "Tradition attained",
+    "21": "Tradition constrained",
+    "22": "Conformity: rules attained",
+    "23": "Conformity: rules constrained",
+    "24": "Conformity: interpersonal attained",
+    "25": "Conformity: interpersonal constrained",
+    "26": "Humility attained",
+    "27": "Humility constrained",
+    "28": "Benevolence: caring attained",
+    "29": "Benevolence: caring constrained",
+    "30": "Benevolence: dependability attained",
+    "31": "Benevolence: dependability constrained",
+    "32": "Universalism: concern attained",
+    "33": "Universalism: concern constrained",
+    "34": "Universalism: nature attained",
+    "35": "Universalism: nature constrained",
+    "36": "Universalism: tolerance attained",
+    "37": "Universalism: tolerance constrained"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 10240,
+  "label2id": {
+    "Achievement attained": 8,
+    "Achievement constrained": 9,
+    "Benevolence: caring attained": 28,
+    "Benevolence: caring constrained": 29,
+    "Benevolence: dependability attained": 30,
+    "Benevolence: dependability constrained": 31,
+    "Conformity: interpersonal attained": 24,
+    "Conformity: interpersonal constrained": 25,
+    "Conformity: rules attained": 22,
+    "Conformity: rules constrained": 23,
+    "Face attained": 14,
+    "Face constrained": 15,
+    "Hedonism attained": 6,
+    "Hedonism constrained": 7,
+    "Humility attained": 26,
+    "Humility constrained": 27,
+    "Power: dominance attained": 10,
+    "Power: dominance constrained": 11,
+    "Power: resources attained": 12,
+    "Power: resources constrained": 13,
+    "Security: personal attained": 16,
+    "Security: personal constrained": 17,
+    "Security: societal attained": 18,
+    "Security: societal constrained": 19,
+    "Self-direction: action attained": 2,
+    "Self-direction: action constrained": 3,
+    "Self-direction: thought attained": 0,
+    "Self-direction: thought constrained": 1,
+    "Stimulation attained": 4,
+    "Stimulation constrained": 5,
+    "Tradition attained": 20,
+    "Tradition constrained": 21,
+    "Universalism: concern attained": 32,
+    "Universalism: concern constrained": 33,
+    "Universalism: nature attained": 34,
+    "Universalism: nature constrained": 35,
+    "Universalism: tolerance attained": 36,
+    "Universalism: tolerance constrained": 37
+  },
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "xlm-roberta-xl",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "problem_type": "multi_label_classification",
+  "tokenizer_class": "XLMRobertaTokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.40.2",
+  "type_vocab_size": 1,
+  "use_cache": true,
+  "vocab_size": 250041
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19e3393812f6797ad93f6d08af2a48b2c7504de103e8ab0d16a1f9f0c5e007b1
+size 4978468240

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb2975b866c4b6ecb736b8c58be1d79ac398e2cb4a207fd8c46558a271131a8d
+size 4930427680

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64bfd5595c5155cac062ece12c2b2dad41a5f302409e9a5df5b31703656f273b
+size 4904992464

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bba545d2f5b04307e08bbca856e80ccbc6b75da2e1812cc068d134de91f76c9
+size 2179438384

model.safetensors.index.json

@@ -0,0 +1,770 @@
+{
+  "metadata": {
+    "total_size": 16993230168
+  },
+  "weight_map": {
+    "language_heads.BG.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self.key.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self.key.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self.query.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self.query.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self.value.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self.value.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.intermediate.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.intermediate.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.BG.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.BG.out_proj.weight": "model-00004-of-00004.safetensors",
+    "language_heads.DE.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.DE.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.DE.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.DE.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.DE.out_proj.weight": "model-00004-of-00004.safetensors",
+    "language_heads.EL.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.output.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.extra_transformer_layer_1.output.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EL.out_proj.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EL.out_proj.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.output.dense.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.extra_transformer_layer_1.output.dense.weight": "model-00003-of-00004.safetensors",
+    "language_heads.EN.out_proj.bias": "model-00003-of-00004.safetensors",
+    "language_heads.EN.out_proj.weight": "model-00003-of-00004.safetensors",
+    "language_heads.FR.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self.key.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self.key.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self.query.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self.query.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self.value.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self.value.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.intermediate.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.intermediate.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.FR.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.FR.out_proj.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self.key.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self.key.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self.query.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self.query.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self.value.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self.value.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.intermediate.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.intermediate.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.HE.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.HE.out_proj.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self.key.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self.key.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self.query.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self.query.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self.value.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self.value.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.intermediate.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.intermediate.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.IT.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.IT.out_proj.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self.key.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self.key.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self.query.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self.query.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self.value.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self.value.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.intermediate.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.intermediate.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.NL.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.NL.out_proj.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.LayerNorm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.LayerNorm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self.key.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self.key.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self.query.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self.query.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self.value.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self.value.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self_attn_layer_norm.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.attention.self_attn_layer_norm.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.intermediate.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.intermediate.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.output.dense.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.extra_transformer_layer_1.output.dense.weight": "model-00004-of-00004.safetensors",
+    "language_heads.TR.out_proj.bias": "model-00004-of-00004.safetensors",
+    "language_heads.TR.out_proj.weight": "model-00004-of-00004.safetensors",
+    "roberta.embeddings.position_embeddings.weight": "model-00001-of-00004.safetensors",
+    "roberta.embeddings.token_type_embeddings.weight": "model-00001-of-00004.safetensors",
+    "roberta.embeddings.word_embeddings.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.0.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.0.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.1.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.10.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.10.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.11.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.12.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.13.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.14.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.15.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.16.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.17.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.18.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.19.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.2.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.2.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.20.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.20.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.21.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.22.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.23.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.23.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.23.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.23.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.23.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.24.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.25.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.26.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.27.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.28.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.29.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.3.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.3.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.30.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.30.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.31.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.32.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.33.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.34.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.LayerNorm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.LayerNorm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self.key.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self.key.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self.query.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self.query.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self.value.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self.value.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self_attn_layer_norm.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.attention.self_attn_layer_norm.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.intermediate.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.intermediate.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.output.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.35.output.dense.weight": "model-00003-of-00004.safetensors",
+    "roberta.encoder.layer.4.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.4.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.5.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.LayerNorm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.LayerNorm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.6.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.7.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.output.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.output.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self.key.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self.key.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self.query.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self.query.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self.value.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self.value.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self_attn_layer_norm.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.attention.self_attn_layer_norm.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.intermediate.dense.bias": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.intermediate.dense.weight": "model-00001-of-00004.safetensors",
+    "roberta.encoder.layer.7.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.7.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.8.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.LayerNorm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.LayerNorm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self.key.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self.key.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self.query.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self.query.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self.value.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self.value.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self_attn_layer_norm.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.attention.self_attn_layer_norm.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.intermediate.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.intermediate.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.output.dense.bias": "model-00002-of-00004.safetensors",
+    "roberta.encoder.layer.9.output.dense.weight": "model-00002-of-00004.safetensors",
+    "roberta.pooler.dense.bias": "model-00003-of-00004.safetensors",
+    "roberta.pooler.dense.weight": "model-00003-of-00004.safetensors"
+  }
+}