diff --git a/.gitattributes b/.gitattributes
index a6344aac8c09253b3b630fb776ae94478aa0275b..aa7aacd0134a92c3c1943fdecc75cd8b7420cce6 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+model.safetensors.index.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/chat_template.jinja b/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..2ab98ef068d62829d17c5ade1827b9f013fa2bbf
--- /dev/null
+++ b/chat_template.jinja
@@ -0,0 +1,86 @@
+[gMASK]<sop>
+{%- if tools -%}
+<|system|>
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{% for tool in tools %}
+{{ tool | tojson(ensure_ascii=False) }}
+{% endfor %}
+</tools>
+
+For each function call, output the function name and arguments within the following XML format:
+<tool_call>{function-name}<arg_key>{arg-key-1}</arg_key><arg_value>{arg-value-1}</arg_value><arg_key>{arg-key-2}</arg_key><arg_value>{arg-value-2}</arg_value>...</tool_call>{%- endif -%}
+{%- macro visible_text(content) -%}
+    {%- if content is string -%}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if item is mapping and item.type == 'text' -%}
+                {{- item.text }}
+            {%- elif item is string -%}
+                {{- item }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{- content }}
+    {%- endif -%}
+{%- endmacro -%}
+{%- set ns = namespace(last_user_index=-1) %}
+{%- for m in messages %}
+    {%- if m.role == 'user' %}
+        {% set ns.last_user_index = loop.index0 -%}
+    {%- endif %}
+{%- endfor %}
+{% for m in messages %}
+{%- if m.role == 'user' -%}<|user|>{{ visible_text(m.content) }}
+{%- elif m.role == 'assistant' -%}
+<|assistant|>
+{%- set reasoning_content = '' %}
+{%- set content = visible_text(m.content) %}
+{%- if m.reasoning_content is string %}
+    {%- set reasoning_content = m.reasoning_content %}
+{%- else %}
+    {%- if '</think>' in content %}
+        {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+        {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+    {%- endif %}
+{%- endif %}
+{%- if ((clear_thinking is defined and not clear_thinking) or loop.index0 > ns.last_user_index) and reasoning_content -%}
+{{ '<think>' + reasoning_content.strip() +  '</think>'}}
+{%- else -%}
+{{ '</think>' }}
+{%- endif -%}
+{%- if content.strip() -%}
+{{ content.strip() }}
+{%- endif -%}
+{% if m.tool_calls %}
+{% for tc in m.tool_calls %}
+{%- if tc.function %}
+    {%- set tc = tc.function %}
+{%- endif %}
+{{- '<tool_call>' + tc.name -}}
+{% set _args = tc.arguments %}{% for k, v in _args.items() %}<arg_key>{{ k }}</arg_key><arg_value>{{ v | tojson(ensure_ascii=False) if v is not string else v }}</arg_value>{% endfor %}</tool_call>{% endfor %}
+{% endif %}
+{%- elif m.role == 'tool' -%}
+{%- if m.content is string -%}
+{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+    {{- '<|observation|>' }}
+{%- endif %}
+{{- '<tool_response>' }}
+{{- m.content }}
+{{- '</tool_response>' }}
+{%- else -%}
+<|observation|>{% for tr in m.content %}
+<tool_response>{{ tr.output if tr.output is defined else tr }}</tool_response>{% endfor -%}
+{% endif -%}
+{%- elif m.role == 'system' -%}
+<|system|>{{ visible_text(m.content) }}
+{%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    <|assistant|>{{- '</think>' if (enable_thinking is defined and not enable_thinking) else '<think>' -}}
+{%- endif -%}
\ No newline at end of file
diff --git a/config.json b/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfa857c27555fd894cd2da4564f16d02a0808a3b
--- /dev/null
+++ b/config.json
@@ -0,0 +1,2726 @@
+{
+  "architectures": [
+    "GlmMoeDsaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "dtype": "bfloat16",
+  "eos_token_id": [
+    154820,
+    154827,
+    154829
+  ],
+  "ep_size": 1,
+  "first_k_dense_replace": 3,
+  "head_dim": 64,
+  "hidden_act": "silu",
+  "hidden_size": 6144,
+  "index_head_dim": 128,
+  "index_n_heads": 32,
+  "index_topk": 2048,
+  "indexer_rope_interleave": true,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "kv_lora_rank": 512,
+  "max_position_embeddings": 202752,
+  "mlp_layer_types": [
+    "dense",
+    "dense",
+    "dense",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse",
+    "sparse"
+  ],
+  "model_type": "glm_moe_dsa",
+  "moe_intermediate_size": 2048,
+  "moe_layer_freq": 1,
+  "n_group": 1,
+  "n_routed_experts": 256,
+  "n_shared_experts": 1,
+  "norm_topk_prob": true,
+  "num_attention_heads": 64,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 78,
+  "num_key_value_heads": 64,
+  "num_nextn_predict_layers": 1,
+  "pad_token_id": 154820,
+  "pretraining_tp": 1,
+  "q_lora_rank": 2048,
+  "qk_head_dim": 256,
+  "qk_nope_head_dim": 192,
+  "qk_rope_head_dim": 64,
+  "quantization_config": {
+    "autoround_version": "0.10.0",
+    "bits": 4,
+    "data_type": "int",
+    "extra_config": {
+      "model.layers.0.mlp.down_proj": {
+        "bits": 8
+      },
+      "model.layers.0.mlp.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.0.mlp.up_proj": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.0.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.1.mlp.down_proj": {
+        "bits": 8
+      },
+      "model.layers.1.mlp.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.1.mlp.up_proj": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.1.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.10.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.10.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.10.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.10.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.11.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.11.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.11.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.11.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.12.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.12.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.12.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.12.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.13.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.13.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.13.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.13.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.14.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.14.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.14.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.14.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.15.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.15.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.15.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.15.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.16.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.16.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.16.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.16.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.17.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.17.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.17.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.17.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.18.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.18.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.18.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.18.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.19.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.19.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.19.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.19.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.2.mlp.down_proj": {
+        "bits": 8
+      },
+      "model.layers.2.mlp.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.2.mlp.up_proj": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.2.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.20.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.20.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.20.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.20.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.21.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.21.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.21.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.21.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.22.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.22.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.22.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.22.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.23.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.23.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.23.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.23.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.24.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.24.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.24.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.24.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.25.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.25.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.25.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.25.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.26.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.26.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.26.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.26.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.27.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.27.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.27.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.27.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.28.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.28.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.28.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.28.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.29.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.29.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.29.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.29.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.3.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.3.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.3.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.3.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.30.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.30.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.30.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.30.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.31.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.31.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.31.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.31.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.32.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.32.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.32.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.32.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.33.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.33.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.33.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.33.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.34.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.34.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.34.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.34.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.35.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.35.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.35.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.35.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.36.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.36.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.36.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.36.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.37.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.37.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.37.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.37.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.38.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.38.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.38.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.38.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.39.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.39.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.39.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.39.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.4.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.4.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.4.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.4.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.40.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.40.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.40.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.40.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.41.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.41.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.41.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.41.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.42.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.42.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.42.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.42.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.43.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.43.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.43.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.43.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.44.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.44.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.44.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.44.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.45.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.45.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.45.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.45.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.46.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.46.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.46.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.46.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.47.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.47.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.47.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.47.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.48.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.48.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.48.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.48.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.49.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.49.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.49.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.49.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.5.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.5.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.5.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.5.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.50.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.50.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.50.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.50.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.51.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.51.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.51.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.51.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.52.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.52.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.52.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.52.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.53.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.53.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.53.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.53.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.54.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.54.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.54.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.54.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.55.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.55.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.55.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.55.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.56.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.56.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.56.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.56.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.57.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.57.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.57.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.57.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.58.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.58.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.58.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.58.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.59.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.59.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.59.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.59.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.6.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.6.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.6.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.6.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.60.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.60.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.60.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.60.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.61.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.61.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.61.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.61.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.62.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.62.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.62.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.62.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.63.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.63.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.63.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.63.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.64.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.64.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.64.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.64.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.65.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.65.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.65.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.65.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.66.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.66.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.66.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.66.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.67.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.67.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.67.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.67.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.68.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.68.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.68.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.68.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.69.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.69.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.69.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.69.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.7.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.7.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.7.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.7.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.70.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.70.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.70.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.70.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.71.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.71.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.71.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.71.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.72.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.72.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.72.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.72.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.73.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.73.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.73.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.73.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.74.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.74.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.74.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.74.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.75.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.75.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.75.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.75.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.76.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.76.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.76.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.76.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.77.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.77.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.77.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.77.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.8.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.8.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.8.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.8.self_attn.wq_b": {
+        "bits": 8
+      },
+      "model.layers.9.mlp.shared_experts.down_proj": {
+        "bits": 8
+      },
+      "model.layers.9.mlp.shared_experts.gate_proj": {
+        "bits": 8
+      },
+      "model.layers.9.mlp.shared_experts.up_proj": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.kv_a_proj_with_mqa": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.kv_b_proj": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.o_proj": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.q_a_proj": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.q_b_proj": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.weights_proj": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.wk": {
+        "bits": 8
+      },
+      "model.layers.9.self_attn.wq_b": {
+        "bits": 8
+      }
+    },
+    "group_size": 128,
+    "iters": 0,
+    "packing_format": "auto_round:auto_gptq",
+    "quant_method": "auto-round",
+    "sym": true
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_interleave": true,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "routed_scaling_factor": 2.5,
+  "scoring_func": "sigmoid",
+  "tie_word_embeddings": false,
+  "topk_group": 1,
+  "topk_method": "noaux_tc",
+  "transformers_version": "5.2.0.dev0",
+  "use_cache": true,
+  "v_head_dim": 256,
+  "vocab_size": 154880
+}
diff --git a/model-00001-of-00076.safetensors b/model-00001-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..801663274476005bc715c86753600458d94a867d
--- /dev/null
+++ b/model-00001-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b83e9d0ad0a56dac2770a8af57bf576dce3cb904773a5422be0c5f9bfed25ab
+size 5366923072
diff --git a/model-00002-of-00076.safetensors b/model-00002-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..46a548065a0e1004f2326554ad5071d418c46552
--- /dev/null
+++ b/model-00002-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0fdede0711cdc57bd4c34d56034b6680502164ae686febb5a6e7a3de5653534
+size 5362713760
diff --git a/model-00003-of-00076.safetensors b/model-00003-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cf32ee9b47e2059f46dbcb9d1c5cbd9a60a6ee5e
--- /dev/null
+++ b/model-00003-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1431458a88bb44a875b48f3d882e21835a9dba6da4440154fba9d59b2566c79d
+size 5362713784
diff --git a/model-00004-of-00076.safetensors b/model-00004-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e9f949507ad75b8935b78d8c43d53c4c7596f0db
--- /dev/null
+++ b/model-00004-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3847a69c2a3b79c1b84e6738d38e8d596cf80981aa838631ec4ff0b99128fe91
+size 5362713816
diff --git a/model-00005-of-00076.safetensors b/model-00005-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0a7807bd0963c718a7e8cf78fd235b2479769941
--- /dev/null
+++ b/model-00005-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99305ad6522849718bb58af19270edc62b9c1271823ca599723a7ee77b11d46f
+size 5362713816
diff --git a/model-00006-of-00076.safetensors b/model-00006-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e39d1cf831dbfb17ba5e10b6e15428c51f8357b4
--- /dev/null
+++ b/model-00006-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a49bcd95f32ed0db3d10ec7cc5968d2a2dd213da1e4a6d8fc27fbbbe4bc47e71
+size 5362713816
diff --git a/model-00007-of-00076.safetensors b/model-00007-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4897a825aa4418c1aafc72104d3b8691e372d15a
--- /dev/null
+++ b/model-00007-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f8a1faf2f6d31e8949884e62ae749d2e8c4b0578540c6375a5b2e10187fb569
+size 5362713816
diff --git a/model-00008-of-00076.safetensors b/model-00008-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..be6f7e6077edb63dfcdf327f57ebe52cda968200
--- /dev/null
+++ b/model-00008-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e157b39397847af0c220f68e6ecde1d99fd681db8a55887e98647f21d06a31f
+size 5362715144
diff --git a/model-00009-of-00076.safetensors b/model-00009-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..231a05a887a51df7ed5d0c3002fb5b1d41012738
--- /dev/null
+++ b/model-00009-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff1e1b5087e555851b2309883a059b40ba660b8e8fb6a418f0939df152c2befb
+size 5362715432
diff --git a/model-00010-of-00076.safetensors b/model-00010-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..56adbcb3f87e0d6bdc49939916824df5c6b1cc00
--- /dev/null
+++ b/model-00010-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1fdbf5319f5c8c0ea30a2efdcea4795857e977ac297ee32d6b5d3757adbbc98
+size 5362715432
diff --git a/model-00011-of-00076.safetensors b/model-00011-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..b7f1c4f5fd2b3f8adee454134f9a10b68125cbbe
--- /dev/null
+++ b/model-00011-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac7784feb4019ff13f6f60000365aca8b42216ffaf3fcb2bd15fa2221662e648
+size 5362715432
diff --git a/model-00012-of-00076.safetensors b/model-00012-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5b0783f5d45948ce7d18be6da343de2c49f94cdd
--- /dev/null
+++ b/model-00012-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c1194b561e892c9bc098440d0dff6606f2be0254a27dfde18a761f5a551cb57
+size 5362715432
diff --git a/model-00013-of-00076.safetensors b/model-00013-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ac02482b319a2811244045d8832b16d846633e75
--- /dev/null
+++ b/model-00013-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf9e0b38c45a7b9544e4375926eef734bca8b0cb7c489565ca2333c8493977bc
+size 5368859320
diff --git a/model-00014-of-00076.safetensors b/model-00014-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e5479d3b41a7f0bb80a37aeac71a2697ee863d37
--- /dev/null
+++ b/model-00014-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff7a47582aeb84db0329cd918c78047028c35b6460b3d8edbb2e1aa29cccdc80
+size 5356470896
diff --git a/model-00015-of-00076.safetensors b/model-00015-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..77112b847374f8880df063adbbb6501a917ba257
--- /dev/null
+++ b/model-00015-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:765ed1987eb01eb0e308c982c52985918b8e29f430cf3199cc4675ca1c9dc37f
+size 5367674528
diff --git a/model-00016-of-00076.safetensors b/model-00016-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c9d20829ca56be1b40bedd1f27f426cf8b4aff7c
--- /dev/null
+++ b/model-00016-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c14f17bcf6c62cda9c2acd7d2dcc453bfb06e8151888c21eff1a15d39a9faf6d
+size 5366379680
diff --git a/model-00017-of-00076.safetensors b/model-00017-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d1a3273f0f71af40064d6dffc3d949bbde95af6b
--- /dev/null
+++ b/model-00017-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ade7714806556ed1250f38bbdbc1101cdb3398b45c82778e0d18f7f92cb9b36a
+size 5368856712
diff --git a/model-00018-of-00076.safetensors b/model-00018-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0ea71d1e9a753de81a43b56485dd54a933abceac
--- /dev/null
+++ b/model-00018-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b75c8a14e676558434f89230178c0b40244ae9698eff7e847f922da8d6f13a80
+size 5356568600
diff --git a/model-00019-of-00076.safetensors b/model-00019-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0014e461bb4e6ce1ce8b5e7c4ddfb555f3f48fb7
--- /dev/null
+++ b/model-00019-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d3cf8e26ec03c8191ad641dfe6ae16812e434d7ed51c5751687cc2ad3080d0e
+size 5368856728
diff --git a/model-00020-of-00076.safetensors b/model-00020-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..652c9b8bf93d81f7a2c1a548f9d5d60d4a7bd84c
--- /dev/null
+++ b/model-00020-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a06dac3bf746d1b0ba890a68e72c613a3cb076068841cee8f4e8b9dac44a46e
+size 5356568600
diff --git a/model-00021-of-00076.safetensors b/model-00021-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5326d11d26e55013eb4ff44fafc67b4f47d872e2
--- /dev/null
+++ b/model-00021-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94df641d686eadab356110218e69571c8f22813b0f71118a049ebdefce303ac2
+size 5368856728
diff --git a/model-00022-of-00076.safetensors b/model-00022-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..36ebbe506a00e437f4a1de820eb2ae5ca297e239
--- /dev/null
+++ b/model-00022-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5b3fea9d60d692a2bd38426bf8fb90b769b836a1a846c46c97b369637e019c9
+size 5356568600
diff --git a/model-00023-of-00076.safetensors b/model-00023-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4957d5cb4a43d572f9818c2171f46c010a15f7eb
--- /dev/null
+++ b/model-00023-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b19b0b0ce7a233006cb6b896c4ce3a2f721113e143794eee2454fcd05477961
+size 5368856728
diff --git a/model-00024-of-00076.safetensors b/model-00024-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3fcb0e3728116f18c808cae050a63bb5a96e5529
--- /dev/null
+++ b/model-00024-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cf773480a1d4638f8098e8bc3f47c5eade4aef2c5f24c928f52ab67ad270290
+size 5356568600
diff --git a/model-00025-of-00076.safetensors b/model-00025-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..657365905fd3d826bd1399ae1711896c15a1c6ee
--- /dev/null
+++ b/model-00025-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0307b2efa9b2086555fee2c9355f1d37c09f04699c4f774a51a0345c371f82ae
+size 5368856728
diff --git a/model-00026-of-00076.safetensors b/model-00026-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e11eed4aa713c2c8faa25f5a99570226787cd2af
--- /dev/null
+++ b/model-00026-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e6231d0db0663c90be2b0b6693b5cb499900e34e8f829b08c05a4b666b41579
+size 5356568600
diff --git a/model-00027-of-00076.safetensors b/model-00027-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..09babaaf33c2374567dc67e52503f57433a4ce13
--- /dev/null
+++ b/model-00027-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ac0df36d2e307cce9d461c594722eba2f46a19f24fd3c8dd59523cac389c85f
+size 5368856728
diff --git a/model-00028-of-00076.safetensors b/model-00028-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a84ca13870a05aa9e6bc1f669a24a56e40427fe2
--- /dev/null
+++ b/model-00028-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:216acd17044e6d3da142ee2d560cf11c7e46b582b0c91baf402d0d6e277e909e
+size 5356568600
diff --git a/model-00029-of-00076.safetensors b/model-00029-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3c005cef64e8e0612da273008e2ddb017918436b
--- /dev/null
+++ b/model-00029-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2259be935a9e8e7732a90bcb53c2064ab05e42be611d246a5be30efbb7392725
+size 5368856736
diff --git a/model-00030-of-00076.safetensors b/model-00030-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3e4290566f32d14b6567b5843f0f760a5a376ea6
--- /dev/null
+++ b/model-00030-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2e5e8deac03ddb66e3fbc3feb2796da710d81703d9a2dcae1ebbbc6acb61e1f
+size 5356568624
diff --git a/model-00031-of-00076.safetensors b/model-00031-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..685994b5162a986f29feba1a71ecbc6f65ddc8bb
--- /dev/null
+++ b/model-00031-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:589b2668d458c8c9d2c07ac3235da199e560613a2218522db6169e2b2672af86
+size 5368856752
diff --git a/model-00032-of-00076.safetensors b/model-00032-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f2aaacfe1f562e62a1a10eea1cfcf306d553855b
--- /dev/null
+++ b/model-00032-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91568559fe6de526fff2613e340a48945c8c72225c286ba63ca399bcfd283765
+size 5356568624
diff --git a/model-00033-of-00076.safetensors b/model-00033-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7f2bab68e1d67860928ecefd802e92793d3407bf
--- /dev/null
+++ b/model-00033-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:233f7746641c895661d3d81d25eabaea200f69ed63b0dbd254418430096052ac
+size 5368856752
diff --git a/model-00034-of-00076.safetensors b/model-00034-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6266fdcfc1b477b737a372fbb9b1ac0a52be67c0
--- /dev/null
+++ b/model-00034-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3459522256bdf4f13acfc381a7d3f5a7f1787f06943c870c20e1ea723709708
+size 5356568624
diff --git a/model-00035-of-00076.safetensors b/model-00035-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..dfe426f6271ec73963c4c119a9ab3bf2665b8670
--- /dev/null
+++ b/model-00035-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:525a9afbc52bbf13fe23bb0d6ef5ed06d0dc44c324775b7b88aa3d29cd25c2a6
+size 5368856752
diff --git a/model-00036-of-00076.safetensors b/model-00036-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9906e70f0604baab1c32c86b6d006b2d3a95dcb4
--- /dev/null
+++ b/model-00036-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8593a90dcc457f392be616c184f89698c331967017a0c5c929cb9dcf14817211
+size 5356568624
diff --git a/model-00037-of-00076.safetensors b/model-00037-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..85ac145d9999554e3a4109d6828060e879e6949c
--- /dev/null
+++ b/model-00037-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49ce7aecc9f4b28c5f3e1cfcf585dae5f7c47a9ca3cd0e0e34640437a00b045f
+size 5368856752
diff --git a/model-00038-of-00076.safetensors b/model-00038-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..25f5310166f69c42c7a81104c91775dd211bcacb
--- /dev/null
+++ b/model-00038-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3951004bad0f36dc24ab7e54b38cf44b3869a5613bf61f6f9a3badabb8f9236
+size 5356568624
diff --git a/model-00039-of-00076.safetensors b/model-00039-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..482c198afc495777e9587d90b536e0045457a51a
--- /dev/null
+++ b/model-00039-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e2c9db6c9ff59673047bc8e7a20e4c18d2f4d926b10169b936154af91da6fe9
+size 5368856752
diff --git a/model-00040-of-00076.safetensors b/model-00040-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2711496eca91ca1bc4e13498d56e28a57edfbe87
--- /dev/null
+++ b/model-00040-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c37da3518ab4cbcfd5757bd815b3e908441c69e65f6cc177f5dcb9946bc393a
+size 5356568624
diff --git a/model-00041-of-00076.safetensors b/model-00041-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..bec9371af65a368b8850370765f1fca49c8b5cc3
--- /dev/null
+++ b/model-00041-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7397a584b016ffc46120481fb864fa3c882b7f0bef8711cbc3c0e9c7e7f921
+size 5368856752
diff --git a/model-00042-of-00076.safetensors b/model-00042-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8fcaaaf1c7ecfd4ff59d88aacb12e4d23839176e
--- /dev/null
+++ b/model-00042-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7848d34b3af5defceeb87a1c77ab21ddaccebd4db071148fe5cee4cf3b086c7b
+size 5356568624
diff --git a/model-00043-of-00076.safetensors b/model-00043-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d47d29e591d323b51ae8bd75d9ab248a6ea5161c
--- /dev/null
+++ b/model-00043-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ecf7c5f6d3c4f353508036455df4f2abf286505902dc67c33fe4affee5dd07a
+size 5368856752
diff --git a/model-00044-of-00076.safetensors b/model-00044-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9b3de825e5554a16a5f6e96b11c96aa787774974
--- /dev/null
+++ b/model-00044-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e18e5b3098a8380cbc43fc14d764d035eee5dc987a00ce23a0f6bf4cbf36936f
+size 5356568624
diff --git a/model-00045-of-00076.safetensors b/model-00045-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6c4bcf34b3ac3a9d2d02c9272a35f37ece94ef21
--- /dev/null
+++ b/model-00045-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b9cb254a1e1eb1b192d1b10b3410bde1e05187cd02cbf502caa358b0b7286f3
+size 5368856752
diff --git a/model-00046-of-00076.safetensors b/model-00046-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2bd7b7f4c78c35c40b5ceead0c8ce4a5b56cf6c8
--- /dev/null
+++ b/model-00046-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:445f7e4ee11539b998415705c161b16a65438ab4fa55edead111f3774a5c7895
+size 5356568624
diff --git a/model-00047-of-00076.safetensors b/model-00047-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ff110e801a5dfb5a57ec809d193413235a7a15eb
--- /dev/null
+++ b/model-00047-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1d44e3f80307fccbe11a058e04aeb02490e3bfe185effeabd39fb35bcd5df9f
+size 5368856752
diff --git a/model-00048-of-00076.safetensors b/model-00048-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..eb933edc4ba0cdb6e50779019c54a59a07e7179d
--- /dev/null
+++ b/model-00048-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37f9311282e31e6b788b871dfbcdc590c08289253a4b003a8fd8490cb5b58832
+size 5356568624
diff --git a/model-00049-of-00076.safetensors b/model-00049-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..58604943458e74380698fd06c8dc138ac1844ac9
--- /dev/null
+++ b/model-00049-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:479a0b0acc2914871c983410a5e8a82a5ab25925e963df5e59ecb941353ccdcd
+size 5368856752
diff --git a/model-00050-of-00076.safetensors b/model-00050-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c524e8a51cf9f09c62c46a741a6d3a19fe368664
--- /dev/null
+++ b/model-00050-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2be1e9b2fdde7d2f1e6838f679c07975bd5b2532cb483b046c36b9c8fac37922
+size 5363107616
diff --git a/model-00051-of-00076.safetensors b/model-00051-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..90f94ae69b2f6ff0f77c99df884a5851c29c3125
--- /dev/null
+++ b/model-00051-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:797a8844e2e54b8c67e43f5c95b630b6fb3abd08bf1ebfae825c8041e1192f04
+size 5362715368
diff --git a/model-00052-of-00076.safetensors b/model-00052-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..737251432437cee8f34ce80a4b5fda5b68aacfed
--- /dev/null
+++ b/model-00052-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:494239e85cfec11a0240f27320227a573f99654fbe88f7bd51381a91f205948c
+size 5362715376
diff --git a/model-00053-of-00076.safetensors b/model-00053-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7046cc15ef0d81977d7488c60b46a9fdad2a58f0
--- /dev/null
+++ b/model-00053-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7c0e2903f6975055e3470ef02b190acbc063c496377174ef6c3cf977e1b7159
+size 5362715376
diff --git a/model-00054-of-00076.safetensors b/model-00054-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..04776d976a8ebfe45916fd1d92ebdc89f76f3cb5
--- /dev/null
+++ b/model-00054-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd2121741fb0f3419a8a922e8c6404e72a34dba36601e99e1d63d8589e8db2b7
+size 5362715376
diff --git a/model-00055-of-00076.safetensors b/model-00055-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..873e78170cadc0178148d4c4f84872715a24f5fa
--- /dev/null
+++ b/model-00055-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd01b43534bce0bfd0b92ffe04ff0741a5cc143174a07522f36e87a2cae67e1b
+size 5362715376
diff --git a/model-00056-of-00076.safetensors b/model-00056-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4af0d4cadd9047bfaf22fff21be06204f8b77b5a
--- /dev/null
+++ b/model-00056-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df27f6626c256f0defadd02ad72bba46dfe46d6b8efc346c041ac87007b687cd
+size 5362715376
diff --git a/model-00057-of-00076.safetensors b/model-00057-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7fc0e600d3715e65b1ea00b084c1435f2a14a1fe
--- /dev/null
+++ b/model-00057-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66731faa5796f8bd5617d3eaf93e81f6d6ce627ef1d6eb0e7a723f724df40fc9
+size 5362715416
diff --git a/model-00058-of-00076.safetensors b/model-00058-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ee5f77eb5f75eb5b4b7bac085b932d9b081b6072
--- /dev/null
+++ b/model-00058-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dafbc7dca86076337d8a187e947994797df360842e2d75fdfc31dba490174c3
+size 5362715432
diff --git a/model-00059-of-00076.safetensors b/model-00059-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..59a4e4d10af3008cc336f2d0fd16bb3622a89cc6
--- /dev/null
+++ b/model-00059-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:028b3b46bdbd1faa7b2e82d9a8db315b993168fcd59423818e7808a19d5ebdcb
+size 5362715432
diff --git a/model-00060-of-00076.safetensors b/model-00060-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..da38d13a85e68c0bfe4811469c49f0a87ba988da
--- /dev/null
+++ b/model-00060-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82ecf1b240abae7ffedc59528dd211ef00c90e7782115cee1d7fb2358f5b6617
+size 5362715432
diff --git a/model-00061-of-00076.safetensors b/model-00061-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a31b18e0399c5727b08e492a461a6bbd6add89e2
--- /dev/null
+++ b/model-00061-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b45a49880200f37f0c7c43f1d39163c2192e13635716b7d9d8aa9d666e52959b
+size 5362715432
diff --git a/model-00062-of-00076.safetensors b/model-00062-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c9d43b0b1cf2b99e6107d4968d6a278f77caf966
--- /dev/null
+++ b/model-00062-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbb3bccaea26cf2249775e841debab40ff7af20c638bbf621a727664ab703f12
+size 5362715432
diff --git a/model-00063-of-00076.safetensors b/model-00063-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e36579191172104337bdf3b2de71f54b44c59870
--- /dev/null
+++ b/model-00063-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:151f0f6b38218e1276bd96500b0cc04cc9bfaed8603be88957bde5714ec35311
+size 5362715432
diff --git a/model-00064-of-00076.safetensors b/model-00064-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1a4e508ad4a653f3bbb1e549aee66a89ededad74
--- /dev/null
+++ b/model-00064-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da28aa6b75aae2779b0b97a8eeaa8ca81e0f27fae7d28a79f7ac281ba4a21877
+size 5362715432
diff --git a/model-00065-of-00076.safetensors b/model-00065-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..59c3b16baf66b1e18709022cdae68b00f98708e5
--- /dev/null
+++ b/model-00065-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dda12884166d86ce214bfcb5c0f24e1e092d25e0c2d4edcce70ee0ea7d41c74e
+size 5362715432
diff --git a/model-00066-of-00076.safetensors b/model-00066-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4126cd7dd0714dc9b28aaedbd9e7d51b10f8ced7
--- /dev/null
+++ b/model-00066-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edcab54a182b3ae1479d0ec42e7508a2f2edeb583c28208d7ce2d7bf6b465138
+size 5362715432
diff --git a/model-00067-of-00076.safetensors b/model-00067-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6de87a7cc78bc163c8b16c4b14776986ab415b51
--- /dev/null
+++ b/model-00067-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90b057e57d73698f74dee96c68e393be4782ee42a2e75e5ba1a110c19476273a
+size 5361927520
diff --git a/model-00068-of-00076.safetensors b/model-00068-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a244c6b54702df0713a73d3e6d8ff7ecb6f14b2e
--- /dev/null
+++ b/model-00068-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94e945e92bff10c8386c78ee35354ee009e9e81d9a9f58492ff20dff5337c8d1
+size 5317639560
diff --git a/model-00069-of-00076.safetensors b/model-00069-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7017d826d77e7df27e8abcfb5e6d1e83b9a0f379
--- /dev/null
+++ b/model-00069-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7169ef3d90d290f86e9492c6d1a4cb198dddcd619df58c17378a1662cfe23f71
+size 5367674528
diff --git a/model-00070-of-00076.safetensors b/model-00070-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1ea6d3f6db6f3f1477a1653538f0cd7545cb7ed1
--- /dev/null
+++ b/model-00070-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d808b068d733204f44ce843ce9343af4b65f6104ce1ac51e9145e65972810231
+size 5366379680
diff --git a/model-00071-of-00076.safetensors b/model-00071-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3bfb3a8ada297ca78f5860a84f57012bb52e3a3a
--- /dev/null
+++ b/model-00071-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f9f9055e320cbfa871126ab258ca42883a9f344eb3a41f17d45b31aeafc5252
+size 5368856712
diff --git a/model-00072-of-00076.safetensors b/model-00072-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..edfa7b68b186cd8b00dccf37e7ac0d0ec5807cb6
--- /dev/null
+++ b/model-00072-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e06b5e481acac9534e0ab5525df163649aad971b71ca1aaf3fdf67ca3510ff07
+size 5356568600
diff --git a/model-00073-of-00076.safetensors b/model-00073-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7a8f9edba5010ed61975955e814a9c75e21860af
--- /dev/null
+++ b/model-00073-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c4c9f5847e60234454ecc4bc28a6fde6802cd176341fe5595d951634253c2ff
+size 5368856728
diff --git a/model-00074-of-00076.safetensors b/model-00074-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..00442b657f588beb7407644cccf13c0458e45a16
--- /dev/null
+++ b/model-00074-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0008c2cfc50a4c1261e7f27c969994b9cdb339e81355e641f233c58cf057b2e8
+size 4693707096
diff --git a/model-00075-of-00076.safetensors b/model-00075-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..b57ee18abf3a7a139f6d167b5d6cde7d7c111053
--- /dev/null
+++ b/model-00075-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a70692fb5ff521b8c8b5c055f2939e87e32b82eadfc257b3eab9e776d5e3e0c8
+size 5368296872
diff --git a/model-00076-of-00076.safetensors b/model-00076-of-00076.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..b8c3340a12c1f2b47c4a4ec0ff17e3ef93bec193
--- /dev/null
+++ b/model-00076-of-00076.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3013507d68a119122f3e34f6a603a00ec09ae96e705c7bf761bdd1122504cd3e
+size 3370177832
diff --git a/model.safetensors.index.json b/model.safetensors.index.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8821fc4527e670c4b4893656beb399908c6d215
--- /dev/null
+++ b/model.safetensors.index.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee5f6172a9195007aaa7f1b453c8519dcec74b0189695883ae418bb9f58250d3
+size 11022417
diff --git a/tokenizer.json b/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7edb3c9544185418004c88e219aae5cf34eedfa
--- /dev/null
+++ b/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cfe2c099a7702a0921abc315ee039deb51e4a34b4818fc509bd27fa3dc4acc1
+size 20217541
diff --git a/tokenizer_config.json b/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1723f7d90e3fb497303ec7b18f88cf5d05928f37
--- /dev/null
+++ b/tokenizer_config.json
@@ -0,0 +1,33 @@
+{
+  "backend": "tokenizers",
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": [
+    "<|endoftext|>",
+    "[MASK]",
+    "[gMASK]",
+    "[sMASK]",
+    "<sop>",
+    "<eop>",
+    "<|system|>",
+    "<|user|>",
+    "<|assistant|>",
+    "<|observation|>",
+    "<|begin_of_image|>",
+    "<|end_of_image|>",
+    "<|begin_of_video|>",
+    "<|end_of_video|>",
+    "<|begin_of_audio|>",
+    "<|end_of_audio|>",
+    "<|begin_of_transcription|>",
+    "<|end_of_transcription|>"
+  ],
+  "is_local": true,
+  "model_max_length": 202752,
+  "model_specific_special_tokens": {},
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "remove_space": false,
+  "tokenizer_class": "TokenizersBackend"
+}