Upload folder using huggingface_hub

.ipynb_checkpoints/special_tokens_map-checkpoint.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

config.json

@@ -0,0 +1,26 @@
+{
+  "architectures": [
+    "MistralForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 32768,
+  "model_type": "mistral",
+  "num_attention_heads": 32,
+  "num_experts": 8,
+  "num_experts_per_token": 2,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-05,
+  "rope_theta": 100000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.36.0.dev0",
+  "use_cache": true,
+  "vocab_size": 32000
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "transformers_version": "4.36.0.dev0"
+}

pytorch_model-00001-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eafe4142da4abd8d783b5c2ded4a58abdcd3c5d54eca80a5cf9c0430875316b
+size 4892754403

pytorch_model-00002-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7eed9709d97f2da15e345b786496d4f4c2fa5fe7a9770e6d01569b7cd9e2fa3e
+size 4983015549

pytorch_model-00003-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66edf32649372dea0bb668d80069c253e9bbfbc6295d9ace3928745ee5484ccb
+size 4983015633

pytorch_model-00004-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d10c566dddc85bf4f1ed9599a91aa070cd60fd020893e33310ee3e86bf869dc
+size 4899045183

pytorch_model-00005-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6cd8a15e9f26fc70dce1c85f6e4f6f220fd4dee58d9d0efbd9a21c7db881318
+size 4983015585

pytorch_model-00006-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:235146f448670f9ac0df1a6580a7ffefac41a42cae034f097e2de74b4d442cca
+size 4983015549

pytorch_model-00007-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68283f855e17acc58348e799a67f6ceee7a537c8ca791bc03e3e9888c717c1c3
+size 4899045247

pytorch_model-00008-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:075b93d4498b0555b7fc4f50632808c9c7e452ca084440bc1b56a38ac0f43dd4
+size 4983015673

pytorch_model-00009-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e51955fec37ce020f1843dc6c1239c18387b35d3eebd4024a0e803f8105c132f
+size 4983015613

pytorch_model-00010-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bb0e72f3b662532cc496d6446e88c1aa74f94d4a2c492d32b43c2689a9aa7d6
+size 4899045247

pytorch_model-00011-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d19fce3f2d12226582ad39ae4ce6d7ad7a5ca4c8ce47ce04a33daa3b1dbf1fb
+size 4983015637

pytorch_model-00012-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79237d6d1541219e42fa89f3e272c80ec822ed166b9d54e1e28a771acaed53e6
+size 4983015637

pytorch_model-00013-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efe50b375252bdaf9d762ac0c5550ade4ec052d680cbe0f6ce3effb1f4f7b3b2
+size 4983015613

pytorch_model-00014-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa0a358291074ceff90f1420489bb8cf96e80a037e4206fd7c9fc4ea4a3c9cf0
+size 4899045247

pytorch_model-00015-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73fe5d3b8f1c5c4c406c993b5155effe888f805b322410546fee00201ca58e2b
+size 4983015673

pytorch_model-00016-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35f9633abf24dc4140d5598730caec249d17b2e225310d4117b060f168a4e547
+size 4983015613

pytorch_model-00017-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e08662c4e3fb7d732bcfc0d1380718e61401130f1b067f20182ca2fbc105049b
+size 4899045247

pytorch_model-00018-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a54230146d8bec9f8b09fb259c7c14932ecfef08fe4e329d0a11d784de042085
+size 4983015649

pytorch_model-00019-of-00019.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2200670b3ecfc30846fbb953ad1c6fa02596992d64796c22a1f8444e143a35b
+size 4221754105

pytorch_model.bin.index.json

@@ -0,0 +1,1002 @@
+{
+  "metadata": {
+    "total_size": 93405585408
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00019-of-00019.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.0.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.0.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.0.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.1.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.1.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.1.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.2.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.2.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.2.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.3.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.3.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.3.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.4.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.4.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.4.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.5.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.5.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.5.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.6.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.6.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.6.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.7.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.7.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.experts.7.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.mlp.gate.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.0.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.0.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.0.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.1.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.1.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.1.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.2.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.2.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.2.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.3.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.3.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.3.w3.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.4.w1.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.4.w2.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.mlp.experts.4.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.5.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.5.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.5.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.6.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.6.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.6.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.7.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.7.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.experts.7.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.mlp.gate.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00019.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.0.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.10.mlp.experts.0.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.10.mlp.experts.0.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.10.mlp.experts.1.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.1.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.1.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.2.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.2.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.2.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.3.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.3.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.3.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.4.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.4.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.4.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.5.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.5.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.5.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.6.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.6.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.6.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.7.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.7.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.experts.7.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.mlp.gate.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.mlp.experts.0.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.0.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.0.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.1.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.1.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.1.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.2.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.2.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.2.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.3.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.3.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.3.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.4.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.4.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.4.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.5.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.5.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.5.w3.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.6.w1.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.6.w2.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.mlp.experts.6.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.mlp.experts.7.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.mlp.experts.7.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.mlp.experts.7.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.mlp.gate.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00007-of-00019.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.0.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.0.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.0.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.1.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.1.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.1.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.2.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.2.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.2.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.3.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.3.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.3.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.4.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.4.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.4.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.5.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.5.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.5.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.6.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.6.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.6.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.7.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.7.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.experts.7.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.mlp.gate.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.0.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.0.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.0.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.1.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.1.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.1.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.2.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.2.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.2.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.3.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.3.w2.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.3.w3.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.4.w1.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.mlp.experts.4.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.4.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.5.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.5.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.5.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.6.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.6.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.6.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.7.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.7.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.experts.7.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.mlp.gate.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00008-of-00019.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.0.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.0.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.0.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.1.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.1.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.1.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.2.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.2.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.2.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.3.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.3.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.3.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.4.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.4.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.4.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.5.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.5.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.5.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.6.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.6.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.6.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.7.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.7.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.experts.7.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.mlp.gate.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.0.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.mlp.experts.0.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.mlp.experts.0.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.mlp.experts.1.w1.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.mlp.experts.1.w2.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.mlp.experts.1.w3.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.mlp.experts.2.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.2.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.2.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.3.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.3.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.3.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.4.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.4.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.4.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.5.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.5.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.5.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.6.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.6.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.6.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.7.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.7.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.experts.7.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.mlp.gate.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00009-of-00019.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.16.mlp.experts.0.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.0.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.0.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.1.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.1.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.1.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.2.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.2.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.2.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.3.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.3.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.3.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.4.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.4.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.4.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.5.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.5.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.5.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.6.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.6.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.6.w3.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.7.w1.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.7.w2.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.mlp.experts.7.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.16.mlp.gate.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00010-of-00019.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.0.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.0.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.0.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.1.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.1.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.1.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.2.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.2.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.2.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.3.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.3.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.3.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.4.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.4.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.4.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.5.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.5.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.5.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.6.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.6.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.6.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.7.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.7.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.experts.7.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.mlp.gate.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.0.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.0.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.0.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.1.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.1.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.1.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.2.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.2.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.2.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.3.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.3.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.3.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.4.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.4.w2.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.4.w3.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.5.w1.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.mlp.experts.5.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.5.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.6.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.6.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.6.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.7.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.7.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.experts.7.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.mlp.gate.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00011-of-00019.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.0.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.0.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.0.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.1.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.1.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.1.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.2.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.2.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.2.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.3.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.3.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.3.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.4.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.4.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.4.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.5.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.5.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.5.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.6.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.6.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.6.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.7.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.7.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.experts.7.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.mlp.gate.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.0.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.0.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.0.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.1.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.1.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.1.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.2.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.2.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.2.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.3.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.3.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.3.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.4.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.4.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.4.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.5.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.5.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.5.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.6.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.6.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.6.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.7.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.7.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.experts.7.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.mlp.gate.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.0.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.0.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.0.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.1.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.1.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.1.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.2.w1.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.2.w2.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.2.w3.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.mlp.experts.3.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.3.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.3.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.4.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.4.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.4.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.5.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.5.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.5.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.6.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.6.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.6.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.7.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.7.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.experts.7.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.mlp.gate.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00012-of-00019.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.0.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.0.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.0.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.1.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.1.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.1.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.2.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.2.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.2.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.3.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.3.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.3.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.4.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.4.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.4.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.5.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.5.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.5.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.6.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.6.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.6.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.7.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.7.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.experts.7.w3.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.mlp.gate.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.0.w1.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.22.mlp.experts.0.w2.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.22.mlp.experts.0.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.1.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.1.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.1.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.2.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.2.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.2.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.3.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.3.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.3.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.4.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.4.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.4.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.5.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.5.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.5.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.6.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.6.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.6.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.7.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.7.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.experts.7.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.mlp.gate.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00013-of-00019.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.mlp.experts.0.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.0.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.0.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.1.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.1.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.1.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.2.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.2.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.2.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.3.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.3.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.3.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.4.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.4.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.4.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.5.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.5.w2.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.5.w3.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.6.w1.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.mlp.experts.6.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.mlp.experts.6.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.mlp.experts.7.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.mlp.experts.7.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.mlp.experts.7.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.mlp.gate.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00014-of-00019.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.0.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.0.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.0.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.1.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.1.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.1.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.2.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.2.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.2.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.3.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.3.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.3.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.4.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.4.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.4.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.5.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.5.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.5.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.6.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.6.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.6.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.7.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.7.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.experts.7.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.mlp.gate.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.0.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.0.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.0.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.1.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.1.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.1.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.2.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.2.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.2.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.3.w1.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.3.w2.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.3.w3.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.mlp.experts.4.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.4.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.4.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.5.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.5.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.5.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.6.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.6.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.6.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.7.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.7.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.experts.7.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.mlp.gate.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00015-of-00019.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.0.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.0.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.0.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.1.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.1.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.1.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.2.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.2.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.2.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.3.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.3.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.3.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.4.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.4.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.4.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.5.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.5.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.5.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.6.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.6.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.6.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.7.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.7.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.experts.7.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.mlp.gate.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.0.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.mlp.experts.0.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.mlp.experts.0.w3.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.mlp.experts.1.w1.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.mlp.experts.1.w2.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.mlp.experts.1.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.2.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.2.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.2.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.3.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.3.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.3.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.4.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.4.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.4.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.5.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.5.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.5.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.6.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.6.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.6.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.7.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.7.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.experts.7.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.mlp.gate.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00016-of-00019.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.28.mlp.experts.0.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.0.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.0.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.1.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.1.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.1.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.2.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.2.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.2.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.3.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.3.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.3.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.4.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.4.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.4.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.5.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.5.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.5.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.6.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.6.w2.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.6.w3.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.7.w1.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.mlp.experts.7.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.28.mlp.experts.7.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.28.mlp.gate.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00017-of-00019.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.0.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.0.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.0.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.1.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.1.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.1.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.2.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.2.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.2.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.3.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.3.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.3.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.4.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.4.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.4.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.5.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.5.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.5.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.6.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.6.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.6.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.7.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.7.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.experts.7.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.mlp.gate.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.0.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.0.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.0.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.1.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.1.w2.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.1.w3.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.2.w1.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.mlp.experts.2.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.2.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.3.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.3.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.3.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.4.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.4.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.4.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.5.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.5.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.5.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.6.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.6.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.6.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.7.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.7.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.experts.7.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.mlp.gate.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00002-of-00019.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.0.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.0.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.0.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.1.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.1.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.1.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.2.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.2.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.2.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.3.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.3.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.3.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.4.w1.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.4.w2.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.4.w3.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.mlp.experts.5.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.5.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.5.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.6.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.6.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.6.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.7.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.7.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.experts.7.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.mlp.gate.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00018-of-00019.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.0.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.0.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.0.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.1.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.1.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.1.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.2.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.2.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.2.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.3.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.3.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.3.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.4.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.4.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.4.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.5.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.5.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.5.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.6.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.6.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.6.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.7.w1.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.7.w2.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.experts.7.w3.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.mlp.gate.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00019-of-00019.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.0.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.0.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.0.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.1.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.1.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.1.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.2.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.2.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.2.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.3.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.3.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.3.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.4.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.4.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.4.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.5.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.5.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.5.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.6.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.6.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.6.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.7.w1.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.7.w2.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.experts.7.w3.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.mlp.gate.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.0.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.0.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.0.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.1.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.1.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.1.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.2.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.2.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.2.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.3.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.3.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.3.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.4.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.4.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.4.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.5.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.5.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.5.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.6.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.6.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.6.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.7.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.7.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.experts.7.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.mlp.gate.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00003-of-00019.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.0.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.0.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.0.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.1.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.1.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.1.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.2.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.2.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.2.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.3.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.3.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.3.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.4.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.4.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.4.w3.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.5.w1.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.5.w2.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.mlp.experts.5.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.6.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.6.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.6.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.7.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.7.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.experts.7.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.mlp.gate.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00004-of-00019.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.0.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.0.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.0.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.1.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.1.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.1.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.2.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.2.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.2.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.3.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.3.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.3.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.4.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.4.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.4.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.5.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.5.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.5.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.6.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.6.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.6.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.7.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.7.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.experts.7.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.mlp.gate.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.0.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.0.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.0.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.1.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.1.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.1.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.2.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.2.w2.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.2.w3.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.3.w1.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.mlp.experts.3.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.3.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.4.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.4.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.4.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.5.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.5.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.5.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.6.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.6.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.6.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.7.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.7.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.experts.7.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.mlp.gate.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00005-of-00019.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.0.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.0.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.0.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.1.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.1.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.1.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.2.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.2.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.2.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.3.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.3.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.3.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.4.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.4.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.4.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.5.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.5.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.5.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.6.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.6.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.6.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.7.w1.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.7.w2.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.experts.7.w3.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.mlp.gate.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00006-of-00019.bin",
+    "model.norm.weight": "pytorch_model-00019-of-00019.bin"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
+size 493443

tokenizer_config.json

@@ -0,0 +1,41 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": null,
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}