Upload RWKV6ForCausalLM

config.json

@@ -0,0 +1,36 @@
+{
+  "_name_or_path": "fla-hub/rwkv6-7B-finch",
+  "architectures": [
+    "RWKV6ForCausalLM"
+  ],
+  "attn_mode": "chunk",
+  "bos_token_id": 1,
+  "clamp_min": null,
+  "conv_size": 4,
+  "eos_token_id": 2,
+  "expand_k": 1,
+  "expand_v": 1,
+  "fuse_cross_entropy": true,
+  "fuse_norm": true,
+  "gate_low_rank_dim": 128,
+  "hidden_act": "sqrelu",
+  "hidden_ratio": 3.5,
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": null,
+  "max_position_embeddings": 2048,
+  "model_type": "rwkv6",
+  "norm_bias": true,
+  "norm_eps": 1e-05,
+  "norm_first": true,
+  "num_heads": 64,
+  "num_hidden_layers": 32,
+  "proj_low_rank_dim": 64,
+  "share_conv_kernel": true,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.40.2",
+  "use_cache": true,
+  "use_glu": false,
+  "vocab_size": 65536
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "transformers_version": "4.40.2"
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80c154954adb3e8e831f3d33375b386ef98f539b43bd659d7645b7ec9403340b
+size 4979018912

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a519916b85f1c8282e71e17db064e00b53bee6fcd46e4897ebacb3aca200ea68
+size 4983264920

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ac4c96282a46bfcfe0361680ca90a2109a41d818bda4bf43898f38759c2d080
+size 4772425368

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41e58fe98aad6f868d1e2f8ecaa87b1bc2fc1b0b1cd54cb3dd2d2587c473c9a3
+size 536871040

model.safetensors.index.json

@@ -0,0 +1,781 @@
+{
+  "metadata": {
+    "total_size": 15271493632
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00004-of-00004.safetensors",
+    "model.embeddings.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.0.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.0.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.1.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.1.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.10.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.10.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.10.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.10.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.11.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.11.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.12.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.12.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.13.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.13.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.14.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.14.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.15.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.15.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.16.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.16.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.17.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.17.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.18.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.18.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.19.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.19.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.2.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.2.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.g_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.g_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.g_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.20.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn.key.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn.key.mu": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn.receptance.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn.receptance.mu": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn.value.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.20.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.bonus": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.21.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.attn.k_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.attn.r_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.v_proj.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.w_proj.linear.lora.0.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.w_proj.linear.lora.2.bias": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.w_proj.linear.lora.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.x_bias": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.x_proj.0.linear.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.x_proj.0.mu": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn.x_proj.2.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn_norm.bias": "model-00002-of-00004.safetensors",
+    "model.layers.21.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.21.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.21.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.21.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.22.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.22.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.23.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.23.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.24.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.24.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.25.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.25.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.26.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.26.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.27.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.27.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.28.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.28.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.29.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.29.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.3.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.3.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.30.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.30.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.bonus": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.g_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.g_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.g_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.k_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.r_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.v_proj.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.w_proj.linear.lora.0.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.w_proj.linear.lora.2.bias": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.w_proj.linear.lora.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.x_bias": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.x_proj.0.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.x_proj.0.mu": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn.x_proj.2.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.31.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn.key.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn.key.mu": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn.receptance.linear.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn.receptance.mu": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn.value.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn_norm.bias": "model-00003-of-00004.safetensors",
+    "model.layers.31.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.4.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.4.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.4.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.5.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.5.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.6.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.6.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.7.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.7.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.8.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.8.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.bonus": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.g_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.g_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.g_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.k_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.r_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.v_proj.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.w_proj.linear.lora.0.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.w_proj.linear.lora.2.bias": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.w_proj.linear.lora.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.x_bias": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.x_proj.0.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.x_proj.0.mu": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn.x_proj.2.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.9.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn.key.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn.key.mu": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn.receptance.linear.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn.receptance.mu": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn.value.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn_norm.bias": "model-00001-of-00004.safetensors",
+    "model.layers.9.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.norm.bias": "model-00003-of-00004.safetensors",
+    "model.norm.weight": "model-00003-of-00004.safetensors"
+  }
+}

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": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenization_rwkv5.py

@@ -0,0 +1,229 @@
+# coding=utf-8
+# Copyright 2024 The HuggingFace Inc. team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Tokenization classes for RWKV5."""
+
+import os
+import re
+from typing import TYPE_CHECKING, List, Optional, Tuple
+
+from transformers.tokenization_utils import AddedToken, PreTrainedTokenizer
+from transformers.utils import logging
+
+
+if TYPE_CHECKING:
+    pass
+
+logger = logging.get_logger(__name__)
+
+VOCAB_FILES_NAMES = {
+    "vocab_file": "vocab.txt",
+}
+PRETRAINED_VOCAB_FILES_MAP = {
+    "vocab_file": {
+        "ArthurZ/rwkv-5-utf": "https://huggingface.co/ArthurZ/rwkv-5-utf/blob/main/vocab.txt",
+    },
+}
+
+
+def whitespace_tokenize(text):
+    """Runs basic whitespace cleaning and splitting on a piece of text.
+    The separators are kept
+    """
+    text = text.strip()
+    if not text:
+        return []
+    tokens = re.split(b"(?= )", text)
+    return tokens
+
+
+class WordpieceTokenizer(object):
+    """Runs WordPiece tokenization."""
+
+    def __init__(self, vocab, unk_token):
+        self.vocab = vocab
+        self.unk_token = unk_token
+
+    def tokenize(self, text):
+        """
+        Tokenizes a piece of text into its word pieces. This uses a greedy longest-match-first algorithm to perform
+        tokenization using the given vocabulary.
+
+        For example, `input = "unaffable"` wil return as output `["un", "##aff", "##able"]`.
+
+        Args:
+            text: A single token or whitespace separated tokens. This should have
+                already been passed through *BasicTokenizer*.
+
+        Returns:
+            A list of wordpiece tokens.
+        """
+
+        output_tokens = []
+        for token in whitespace_tokenize(text):
+            chars = list(token)
+            is_bad = False
+            start = 0
+            sub_tokens = []
+            while start < len(chars):
+                end = len(chars)
+                cur_substr = None
+                while start < end:
+                    substr = bytes(chars[start:end])
+                    if substr in self.vocab:
+                        cur_substr = substr
+                        break
+                    end -= 1
+                if cur_substr is None:
+                    is_bad = True
+                    break
+                try:
+                    cur_substr = cur_substr.decode()
+                except UnicodeDecodeError:
+                    cur_substr = str(cur_substr)
+                sub_tokens.append(cur_substr)
+                start = end
+            if is_bad:
+                output_tokens.append(self.unk_token)
+            else:
+                output_tokens.extend(sub_tokens)
+        return output_tokens
+
+
+class Rwkv5Tokenizer(PreTrainedTokenizer):
+    vocab_files_names = VOCAB_FILES_NAMES
+    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
+    max_model_input_sizes = {"ArthurZ/rwkv-5-utf": 2048}
+
+    model_input_names = ["input_ids", "attention_mask"]
+
+    def __init__(self, vocab_file, bos_token="<s>", eos_token="<s>", unk_token="<s>", **kwargs):
+        if not os.path.isfile(vocab_file):
+            raise ValueError(
+                f"Can't find a vocabulary file at path '{vocab_file}'. To load the vocabulary from a Google pretrained"
+                " model use `tokenizer = BertTokenizer.from_pretrained(PRETRAINED_MODEL_NAME)`"
+            )
+
+        with open(vocab_file, "r") as reader:
+            tokens = reader.readlines()
+        vocab = {}
+        for index, token in enumerate(tokens):
+            token = eval(token.rstrip("\n"))
+            vocab[token] = index
+
+        self.add_bos_token = True
+        self.encoder = vocab
+        self.decoder = {v: k for k, v in vocab.items()}
+        self.wordpiece_tokenizer = WordpieceTokenizer(vocab=self.encoder, unk_token=str(unk_token))
+        self._added_tokens_decoder = {0: AddedToken(str(bos_token))}
+        super().__init__(bos_token=bos_token, eos_token=eos_token, unk_token=unk_token, **kwargs)
+
+    @property
+    def vocab_size(self):
+        return len(self.encoder)
+
+    def get_vocab(self):
+        vocab = {str(self.convert_ids_to_tokens(i)): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text, split_special_tokens=False):
+        return self.wordpiece_tokenizer.tokenize(text.encode("utf-8"))
+
+    def _convert_token_to_id(self, token):
+        """Converts a token (byte) to an id using the vocab."""
+        if token.startswith("b'\\"):
+            token = eval(token)
+        elif not isinstance(token, bytes):
+            token = token.encode("utf-8", errors="replace")
+        return self.encoder.get(token, self.unk_token_id)
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (byte) using the vocab."""
+        token = self.decoder.get(index, self.unk_token)
+        if isinstance(token, (bytes)):
+            token = token.decode("utf-8", errors="replace")
+        return token
+
+    def convert_tokens_to_string(self, tokens):
+        """Converts a sequence of tokens (bytes) in a single string. Additional tokens are encoded to bytes"""
+        out_string = b"".join([k.encode(errors="replace") if isinstance(k, str) else k for k in tokens]).decode(
+            "utf-8"
+        )
+        return out_string
+
+    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
+        index = 0
+        if os.path.isdir(save_directory):
+            vocab_file = os.path.join(
+                save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
+            )
+        else:
+            vocab_file = (filename_prefix + "-" if filename_prefix else "") + save_directory
+        with open(vocab_file, "w") as writer:
+            for token, token_index in sorted(self.encoder.items(), key=lambda kv: kv[1]):
+                if index != token_index:
+                    logger.warning(
+                        f"Saving vocabulary to {vocab_file}: vocabulary indices are not consecutive."
+                        " Please check that the vocabulary is not corrupted!"
+                    )
+                    index = token_index
+                writer.write(str(token) + "\n")
+                index += 1
+        return (vocab_file,)
+
+    def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None):
+        if self.add_bos_token:
+            bos_token_ids = [self.bos_token_id]
+        else:
+            bos_token_ids = []
+
+        output = bos_token_ids + token_ids_0
+
+        if token_ids_1 is None:
+            return output
+
+        return output + bos_token_ids + token_ids_1
+
+    def get_special_tokens_mask(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
+    ) -> List[int]:
+        """
+        Retrieves sequence ids from a token list that has no special tokens added. This method is called when adding
+        special tokens using the tokenizer `prepare_for_model` or `encode_plus` methods.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
+                Whether or not the token list is already formatted with special tokens for the model.
+
+        Returns:
+            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
+        """
+        if already_has_special_tokens:
+            return super().get_special_tokens_mask(
+                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
+            )
+
+        if not self.add_bos_token:
+            return super().get_special_tokens_mask(
+                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=False
+            )
+
+        if token_ids_1 is None:
+            return [1] + ([0] * len(token_ids_0))
+        return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1))

tokenizer_config.json

@@ -0,0 +1,26 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_rwkv5.Rwkv5Tokenizer",
+      null
+    ]
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<s>",
+  "model_max_length": 1000000000000000019884624838656,
+  "tokenizer_class": "Rwkv5Tokenizer",
+  "unk_token": "<s>",
+  "use_fast": false
+}