Upload GPTBigCodeForCausalLM

config.json

@@ -0,0 +1,40 @@
+{
+  "_name_or_path": "bigcode/starcoderplus",
+  "activation_function": "gelu",
+  "architectures": [
+    "GPTBigCodeForCausalLM"
+  ],
+  "attention_softmax_in_fp32": true,
+  "attn_pdrop": 0.1,
+  "bos_token_id": 0,
+  "embd_pdrop": 0.1,
+  "eos_token_id": 0,
+  "inference_runner": 0,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "max_batch_size": null,
+  "max_sequence_length": null,
+  "model_type": "gpt_bigcode",
+  "multi_query": true,
+  "n_embd": 6144,
+  "n_head": 48,
+  "n_inner": 24576,
+  "n_layer": 40,
+  "n_positions": 8192,
+  "pad_key_length": true,
+  "pre_allocate_kv_cache": false,
+  "pretraining_tp": 1,
+  "resid_pdrop": 0.1,
+  "scale_attention_softmax_in_fp32": true,
+  "scale_attn_weights": true,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.36.2",
+  "use_cache": false,
+  "validate_runner_input": true,
+  "vocab_size": 49152
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 0,
+  "eos_token_id": 0,
+  "transformers_version": "4.36.2"
+}

model-00001-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a97b765b2809b7a35e8063f9dc5a4e3bc70359531262b72ab8ad4fe1018ff811
+size 4751346952

model-00002-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:693a7d9ca87cac22524a3924fac79b8fd7076788bbb4c2ad1674d70337fce069
+size 4550143488

model-00003-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b8f957d2eb0ce8171e44b9d663c842c52b510d521b001e67d100cccf4b1a682
+size 4550143488

model-00004-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b2b9d66b645ae10bea97d664eab8c206d80ee758f20f947fa64e0e465a98c7d
+size 4550143488

model-00005-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20b339bfae9cc5489b62c1d1b981380d18a6552dd12b565495a2dd03dd501a9a
+size 4550143528

model-00006-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d32a127a4e97566be4fc1094bf9609024c5bb1131837f99b012f099e822111a9
+size 4550143528

model-00007-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec42265b5b4f82e32ce33025cb552894e333a74da7e6c81c8471f0eab0c68c4f
+size 4550143528

model-00008-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d230ce5b52186bf416023554aae6ab62e5bc7b86f1838770ed7f3949109e3a70
+size 4550143528

model-00009-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4d9990b90eeacfe70fe37fcf5241eebc18e09fdb99cf610be5ff64379149b7a
+size 4550143528

model-00010-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84111db4b5cc1af4372986aef7e84a5f28f5c0dd83ebef4bf2294d0e376129f0
+size 4550143528

model-00011-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3147b1c5892bfe1d1c813a345a34d4f5de41da87f7c940259048e410b45e918a
+size 4550143528

model-00012-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8facecbff478c820555fdbf14aed31077472aa8ad033600317efec80657bf7b7
+size 4550143528

model-00013-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4d1ea5cbd623dbe3d5b8435b67f407b9c711c2866bbf27a40dff353365967bb
+size 4550143528

model-00014-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:384350394bbdd153c09e360af9d83a618c35db525f31fb593d019f9d79ad2a61
+size 2724846728

model.safetensors.index.json

@@ -0,0 +1,571 @@
+{
+  "metadata": {
+    "total_size": 62077853696
+  },
+  "weight_map": {
+    "transformer.h.0.attn.c_attn.base_layer.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.0.attn.c_attn.base_layer.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.attn.c_attn.lora_A.default.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.attn.c_attn.lora_B.default.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.attn.c_proj.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.0.attn.c_proj.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.ln_1.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.0.ln_1.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.ln_2.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.0.ln_2.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.mlp.c_fc.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.0.mlp.c_fc.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.0.mlp.c_proj.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.0.mlp.c_proj.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.attn.c_attn.base_layer.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.1.attn.c_attn.base_layer.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.attn.c_attn.lora_A.default.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.attn.c_attn.lora_B.default.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.attn.c_proj.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.1.attn.c_proj.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.ln_1.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.1.ln_1.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.ln_2.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.1.ln_2.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.mlp.c_fc.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.1.mlp.c_fc.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.1.mlp.c_proj.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.1.mlp.c_proj.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.10.attn.c_attn.base_layer.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.10.attn.c_attn.base_layer.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.attn.c_attn.lora_A.default.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.attn.c_attn.lora_B.default.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.attn.c_proj.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.10.attn.c_proj.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.ln_1.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.10.ln_1.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.ln_2.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.10.ln_2.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.mlp.c_fc.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.10.mlp.c_fc.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.10.mlp.c_proj.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.10.mlp.c_proj.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.attn.c_attn.base_layer.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.11.attn.c_attn.base_layer.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.attn.c_attn.lora_A.default.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.attn.c_attn.lora_B.default.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.attn.c_proj.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.11.attn.c_proj.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.ln_1.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.11.ln_1.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.ln_2.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.11.ln_2.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.11.mlp.c_fc.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.11.mlp.c_fc.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.11.mlp.c_proj.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.11.mlp.c_proj.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.attn.c_attn.base_layer.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.12.attn.c_attn.base_layer.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.attn.c_attn.lora_A.default.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.attn.c_attn.lora_B.default.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.attn.c_proj.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.12.attn.c_proj.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.ln_1.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.12.ln_1.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.ln_2.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.12.ln_2.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.mlp.c_fc.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.12.mlp.c_fc.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.12.mlp.c_proj.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.12.mlp.c_proj.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.attn.c_attn.base_layer.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.13.attn.c_attn.base_layer.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.attn.c_attn.lora_A.default.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.attn.c_attn.lora_B.default.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.attn.c_proj.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.13.attn.c_proj.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.ln_1.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.13.ln_1.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.ln_2.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.13.ln_2.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.mlp.c_fc.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.13.mlp.c_fc.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.13.mlp.c_proj.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.13.mlp.c_proj.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.attn.c_attn.base_layer.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.14.attn.c_attn.base_layer.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.attn.c_attn.lora_A.default.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.attn.c_attn.lora_B.default.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.attn.c_proj.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.14.attn.c_proj.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.ln_1.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.14.ln_1.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.ln_2.bias": "model-00005-of-00014.safetensors",
+    "transformer.h.14.ln_2.weight": "model-00005-of-00014.safetensors",
+    "transformer.h.14.mlp.c_fc.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.14.mlp.c_fc.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.14.mlp.c_proj.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.14.mlp.c_proj.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.attn.c_attn.base_layer.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.15.attn.c_attn.base_layer.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.attn.c_attn.lora_A.default.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.attn.c_attn.lora_B.default.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.attn.c_proj.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.15.attn.c_proj.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.ln_1.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.15.ln_1.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.ln_2.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.15.ln_2.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.mlp.c_fc.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.15.mlp.c_fc.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.15.mlp.c_proj.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.15.mlp.c_proj.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.attn.c_attn.base_layer.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.16.attn.c_attn.base_layer.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.attn.c_attn.lora_A.default.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.attn.c_attn.lora_B.default.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.attn.c_proj.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.16.attn.c_proj.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.ln_1.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.16.ln_1.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.ln_2.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.16.ln_2.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.mlp.c_fc.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.16.mlp.c_fc.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.16.mlp.c_proj.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.16.mlp.c_proj.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.attn.c_attn.base_layer.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.17.attn.c_attn.base_layer.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.attn.c_attn.lora_A.default.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.attn.c_attn.lora_B.default.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.attn.c_proj.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.17.attn.c_proj.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.ln_1.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.17.ln_1.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.ln_2.bias": "model-00006-of-00014.safetensors",
+    "transformer.h.17.ln_2.weight": "model-00006-of-00014.safetensors",
+    "transformer.h.17.mlp.c_fc.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.17.mlp.c_fc.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.17.mlp.c_proj.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.17.mlp.c_proj.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.attn.c_attn.base_layer.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.18.attn.c_attn.base_layer.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.attn.c_attn.lora_A.default.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.attn.c_attn.lora_B.default.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.attn.c_proj.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.18.attn.c_proj.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.ln_1.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.18.ln_1.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.ln_2.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.18.ln_2.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.mlp.c_fc.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.18.mlp.c_fc.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.18.mlp.c_proj.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.18.mlp.c_proj.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.attn.c_attn.base_layer.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.19.attn.c_attn.base_layer.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.attn.c_attn.lora_A.default.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.attn.c_attn.lora_B.default.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.attn.c_proj.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.19.attn.c_proj.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.ln_1.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.19.ln_1.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.ln_2.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.19.ln_2.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.mlp.c_fc.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.19.mlp.c_fc.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.19.mlp.c_proj.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.19.mlp.c_proj.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.2.attn.c_attn.base_layer.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.2.attn.c_attn.base_layer.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.2.attn.c_attn.lora_A.default.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.2.attn.c_attn.lora_B.default.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.2.attn.c_proj.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.2.attn.c_proj.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.2.ln_1.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.2.ln_1.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.2.ln_2.bias": "model-00001-of-00014.safetensors",
+    "transformer.h.2.ln_2.weight": "model-00001-of-00014.safetensors",
+    "transformer.h.2.mlp.c_fc.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.2.mlp.c_fc.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.2.mlp.c_proj.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.2.mlp.c_proj.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.20.attn.c_attn.base_layer.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.20.attn.c_attn.base_layer.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.20.attn.c_attn.lora_A.default.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.20.attn.c_attn.lora_B.default.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.20.attn.c_proj.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.20.attn.c_proj.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.20.ln_1.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.20.ln_1.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.20.ln_2.bias": "model-00007-of-00014.safetensors",
+    "transformer.h.20.ln_2.weight": "model-00007-of-00014.safetensors",
+    "transformer.h.20.mlp.c_fc.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.20.mlp.c_fc.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.20.mlp.c_proj.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.20.mlp.c_proj.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.attn.c_attn.base_layer.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.21.attn.c_attn.base_layer.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.attn.c_attn.lora_A.default.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.attn.c_attn.lora_B.default.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.attn.c_proj.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.21.attn.c_proj.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.ln_1.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.21.ln_1.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.ln_2.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.21.ln_2.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.mlp.c_fc.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.21.mlp.c_fc.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.21.mlp.c_proj.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.21.mlp.c_proj.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.attn.c_attn.base_layer.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.22.attn.c_attn.base_layer.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.attn.c_attn.lora_A.default.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.attn.c_attn.lora_B.default.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.attn.c_proj.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.22.attn.c_proj.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.ln_1.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.22.ln_1.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.ln_2.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.22.ln_2.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.mlp.c_fc.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.22.mlp.c_fc.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.22.mlp.c_proj.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.22.mlp.c_proj.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.attn.c_attn.base_layer.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.23.attn.c_attn.base_layer.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.attn.c_attn.lora_A.default.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.attn.c_attn.lora_B.default.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.attn.c_proj.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.23.attn.c_proj.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.ln_1.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.23.ln_1.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.ln_2.bias": "model-00008-of-00014.safetensors",
+    "transformer.h.23.ln_2.weight": "model-00008-of-00014.safetensors",
+    "transformer.h.23.mlp.c_fc.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.23.mlp.c_fc.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.23.mlp.c_proj.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.23.mlp.c_proj.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.attn.c_attn.base_layer.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.24.attn.c_attn.base_layer.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.attn.c_attn.lora_A.default.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.attn.c_attn.lora_B.default.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.attn.c_proj.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.24.attn.c_proj.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.ln_1.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.24.ln_1.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.ln_2.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.24.ln_2.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.mlp.c_fc.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.24.mlp.c_fc.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.24.mlp.c_proj.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.24.mlp.c_proj.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.attn.c_attn.base_layer.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.25.attn.c_attn.base_layer.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.attn.c_attn.lora_A.default.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.attn.c_attn.lora_B.default.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.attn.c_proj.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.25.attn.c_proj.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.ln_1.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.25.ln_1.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.ln_2.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.25.ln_2.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.mlp.c_fc.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.25.mlp.c_fc.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.25.mlp.c_proj.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.25.mlp.c_proj.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.attn.c_attn.base_layer.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.26.attn.c_attn.base_layer.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.attn.c_attn.lora_A.default.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.attn.c_attn.lora_B.default.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.attn.c_proj.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.26.attn.c_proj.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.ln_1.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.26.ln_1.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.ln_2.bias": "model-00009-of-00014.safetensors",
+    "transformer.h.26.ln_2.weight": "model-00009-of-00014.safetensors",
+    "transformer.h.26.mlp.c_fc.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.26.mlp.c_fc.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.26.mlp.c_proj.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.26.mlp.c_proj.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.attn.c_attn.base_layer.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.27.attn.c_attn.base_layer.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.attn.c_attn.lora_A.default.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.attn.c_attn.lora_B.default.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.attn.c_proj.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.27.attn.c_proj.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.ln_1.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.27.ln_1.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.ln_2.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.27.ln_2.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.mlp.c_fc.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.27.mlp.c_fc.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.27.mlp.c_proj.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.27.mlp.c_proj.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.attn.c_attn.base_layer.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.28.attn.c_attn.base_layer.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.attn.c_attn.lora_A.default.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.attn.c_attn.lora_B.default.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.attn.c_proj.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.28.attn.c_proj.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.ln_1.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.28.ln_1.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.ln_2.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.28.ln_2.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.mlp.c_fc.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.28.mlp.c_fc.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.28.mlp.c_proj.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.28.mlp.c_proj.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.attn.c_attn.base_layer.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.29.attn.c_attn.base_layer.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.attn.c_attn.lora_A.default.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.attn.c_attn.lora_B.default.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.attn.c_proj.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.29.attn.c_proj.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.ln_1.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.29.ln_1.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.ln_2.bias": "model-00010-of-00014.safetensors",
+    "transformer.h.29.ln_2.weight": "model-00010-of-00014.safetensors",
+    "transformer.h.29.mlp.c_fc.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.29.mlp.c_fc.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.29.mlp.c_proj.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.29.mlp.c_proj.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.3.attn.c_attn.base_layer.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.3.attn.c_attn.base_layer.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.attn.c_attn.lora_A.default.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.attn.c_attn.lora_B.default.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.attn.c_proj.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.3.attn.c_proj.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.ln_1.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.3.ln_1.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.ln_2.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.3.ln_2.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.mlp.c_fc.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.3.mlp.c_fc.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.3.mlp.c_proj.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.3.mlp.c_proj.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.30.attn.c_attn.base_layer.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.30.attn.c_attn.base_layer.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.attn.c_attn.lora_A.default.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.attn.c_attn.lora_B.default.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.attn.c_proj.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.30.attn.c_proj.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.ln_1.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.30.ln_1.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.ln_2.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.30.ln_2.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.mlp.c_fc.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.30.mlp.c_fc.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.30.mlp.c_proj.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.30.mlp.c_proj.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.attn.c_attn.base_layer.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.31.attn.c_attn.base_layer.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.attn.c_attn.lora_A.default.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.attn.c_attn.lora_B.default.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.attn.c_proj.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.31.attn.c_proj.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.ln_1.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.31.ln_1.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.ln_2.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.31.ln_2.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.mlp.c_fc.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.31.mlp.c_fc.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.31.mlp.c_proj.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.31.mlp.c_proj.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.attn.c_attn.base_layer.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.32.attn.c_attn.base_layer.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.attn.c_attn.lora_A.default.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.attn.c_attn.lora_B.default.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.attn.c_proj.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.32.attn.c_proj.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.ln_1.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.32.ln_1.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.ln_2.bias": "model-00011-of-00014.safetensors",
+    "transformer.h.32.ln_2.weight": "model-00011-of-00014.safetensors",
+    "transformer.h.32.mlp.c_fc.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.32.mlp.c_fc.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.32.mlp.c_proj.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.32.mlp.c_proj.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.attn.c_attn.base_layer.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.33.attn.c_attn.base_layer.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.attn.c_attn.lora_A.default.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.attn.c_attn.lora_B.default.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.attn.c_proj.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.33.attn.c_proj.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.ln_1.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.33.ln_1.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.ln_2.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.33.ln_2.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.mlp.c_fc.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.33.mlp.c_fc.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.33.mlp.c_proj.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.33.mlp.c_proj.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.attn.c_attn.base_layer.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.34.attn.c_attn.base_layer.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.attn.c_attn.lora_A.default.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.attn.c_attn.lora_B.default.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.attn.c_proj.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.34.attn.c_proj.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.ln_1.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.34.ln_1.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.ln_2.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.34.ln_2.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.mlp.c_fc.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.34.mlp.c_fc.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.34.mlp.c_proj.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.34.mlp.c_proj.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.attn.c_attn.base_layer.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.35.attn.c_attn.base_layer.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.attn.c_attn.lora_A.default.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.attn.c_attn.lora_B.default.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.attn.c_proj.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.35.attn.c_proj.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.ln_1.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.35.ln_1.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.ln_2.bias": "model-00012-of-00014.safetensors",
+    "transformer.h.35.ln_2.weight": "model-00012-of-00014.safetensors",
+    "transformer.h.35.mlp.c_fc.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.35.mlp.c_fc.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.35.mlp.c_proj.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.35.mlp.c_proj.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.attn.c_attn.base_layer.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.36.attn.c_attn.base_layer.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.attn.c_attn.lora_A.default.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.attn.c_attn.lora_B.default.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.attn.c_proj.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.36.attn.c_proj.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.ln_1.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.36.ln_1.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.ln_2.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.36.ln_2.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.mlp.c_fc.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.36.mlp.c_fc.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.36.mlp.c_proj.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.36.mlp.c_proj.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.attn.c_attn.base_layer.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.37.attn.c_attn.base_layer.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.attn.c_attn.lora_A.default.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.attn.c_attn.lora_B.default.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.attn.c_proj.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.37.attn.c_proj.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.ln_1.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.37.ln_1.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.ln_2.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.37.ln_2.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.mlp.c_fc.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.37.mlp.c_fc.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.37.mlp.c_proj.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.37.mlp.c_proj.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.attn.c_attn.base_layer.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.38.attn.c_attn.base_layer.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.attn.c_attn.lora_A.default.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.attn.c_attn.lora_B.default.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.attn.c_proj.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.38.attn.c_proj.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.ln_1.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.38.ln_1.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.ln_2.bias": "model-00013-of-00014.safetensors",
+    "transformer.h.38.ln_2.weight": "model-00013-of-00014.safetensors",
+    "transformer.h.38.mlp.c_fc.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.38.mlp.c_fc.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.38.mlp.c_proj.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.38.mlp.c_proj.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.attn.c_attn.base_layer.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.39.attn.c_attn.base_layer.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.attn.c_attn.lora_A.default.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.attn.c_attn.lora_B.default.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.attn.c_proj.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.39.attn.c_proj.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.ln_1.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.39.ln_1.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.ln_2.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.39.ln_2.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.mlp.c_fc.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.39.mlp.c_fc.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.39.mlp.c_proj.bias": "model-00014-of-00014.safetensors",
+    "transformer.h.39.mlp.c_proj.weight": "model-00014-of-00014.safetensors",
+    "transformer.h.4.attn.c_attn.base_layer.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.4.attn.c_attn.base_layer.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.attn.c_attn.lora_A.default.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.attn.c_attn.lora_B.default.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.attn.c_proj.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.4.attn.c_proj.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.ln_1.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.4.ln_1.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.ln_2.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.4.ln_2.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.mlp.c_fc.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.4.mlp.c_fc.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.4.mlp.c_proj.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.4.mlp.c_proj.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.attn.c_attn.base_layer.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.5.attn.c_attn.base_layer.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.attn.c_attn.lora_A.default.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.attn.c_attn.lora_B.default.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.attn.c_proj.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.5.attn.c_proj.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.ln_1.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.5.ln_1.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.ln_2.bias": "model-00002-of-00014.safetensors",
+    "transformer.h.5.ln_2.weight": "model-00002-of-00014.safetensors",
+    "transformer.h.5.mlp.c_fc.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.5.mlp.c_fc.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.5.mlp.c_proj.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.5.mlp.c_proj.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.attn.c_attn.base_layer.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.6.attn.c_attn.base_layer.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.attn.c_attn.lora_A.default.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.attn.c_attn.lora_B.default.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.attn.c_proj.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.6.attn.c_proj.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.ln_1.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.6.ln_1.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.ln_2.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.6.ln_2.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.mlp.c_fc.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.6.mlp.c_fc.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.6.mlp.c_proj.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.6.mlp.c_proj.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.attn.c_attn.base_layer.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.7.attn.c_attn.base_layer.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.attn.c_attn.lora_A.default.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.attn.c_attn.lora_B.default.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.attn.c_proj.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.7.attn.c_proj.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.ln_1.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.7.ln_1.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.ln_2.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.7.ln_2.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.mlp.c_fc.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.7.mlp.c_fc.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.7.mlp.c_proj.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.7.mlp.c_proj.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.attn.c_attn.base_layer.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.8.attn.c_attn.base_layer.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.attn.c_attn.lora_A.default.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.attn.c_attn.lora_B.default.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.attn.c_proj.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.8.attn.c_proj.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.ln_1.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.8.ln_1.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.ln_2.bias": "model-00003-of-00014.safetensors",
+    "transformer.h.8.ln_2.weight": "model-00003-of-00014.safetensors",
+    "transformer.h.8.mlp.c_fc.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.8.mlp.c_fc.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.8.mlp.c_proj.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.8.mlp.c_proj.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.attn.c_attn.base_layer.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.9.attn.c_attn.base_layer.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.attn.c_attn.lora_A.default.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.attn.c_attn.lora_B.default.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.attn.c_proj.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.9.attn.c_proj.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.ln_1.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.9.ln_1.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.ln_2.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.9.ln_2.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.mlp.c_fc.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.9.mlp.c_fc.weight": "model-00004-of-00014.safetensors",
+    "transformer.h.9.mlp.c_proj.bias": "model-00004-of-00014.safetensors",
+    "transformer.h.9.mlp.c_proj.weight": "model-00004-of-00014.safetensors",
+    "transformer.ln_f.bias": "model-00014-of-00014.safetensors",
+    "transformer.ln_f.weight": "model-00014-of-00014.safetensors",
+    "transformer.wpe.weight": "model-00001-of-00014.safetensors",
+    "transformer.wte.weight": "model-00001-of-00014.safetensors"
+  }
+}