diff --git a/README.md b/README.md
index 7b95401dc46245ac339fc25059d4a56d90b4cde5..1b16aa4b9e272a8021c12c0d2917d7365e18019e 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,3 @@
----
-license: apache-2.0
----
+# README
+
+This is the hex data of old reference model, where the softmax are all ones, and rope constants are only collected for one `position_id`
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/formatted_tensors.json b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/formatted_tensors.json
new file mode 100644
index 0000000000000000000000000000000000000000..16293915175ab0e1f6673e2efab21a85d3244514
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/formatted_tensors.json
@@ -0,0 +1,19984 @@
+{
+    "meta-llama_Llama-2-7b-hf.model.embed_tokens.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32000,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.embed_tokens.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.embed_tokens.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.embed_tokens.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.0.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.1.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.2.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.3.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.4.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.5.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.6.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.7.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.8.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.9.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.10.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.11.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.12.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.13.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.14.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.15.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.16.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.17.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.18.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.19.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.20.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.21.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.22.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.23.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.24.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.24.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.25.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.26.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.27.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.28.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.29.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.30.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.input_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.input_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.input_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.input_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.k_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.k_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.k_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.k_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.k_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.k_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.v_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.v_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.v_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.v_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.v_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.v_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.cos": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.cos.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.sin": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.sin.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.query_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.query_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.key_states": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.key_states.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                128,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights_masked": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights_masked.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights_softmaxed": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                32,
+                1,
+                1
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_weights_softmaxed.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_output.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32,
+                1,
+                1
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_output.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_output.other": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -2,
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_output.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_output.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                32,
+                1,
+                128
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.attn_output.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.o_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.o_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.o_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.o_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.o_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.o_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.add1.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.add1.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.add1.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.add1.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.add1.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.add1.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.up_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.up_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.up_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                11008,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.up_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.up_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.up_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.act.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.act.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.act.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.act.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.mult.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.mult.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.mult.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.mult.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.mult.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.mult.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.down_proj.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.down_proj.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.down_proj.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                4096,
+                11008
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.down_proj.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.mlp.down_proj.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.down_proj.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.add2.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.add2.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.add2.other": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.add2.other.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.layers.31.add2.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.add2.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.norm.input": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.norm.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.model.norm.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.norm.output.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.lm_head.input": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                1,
+                1,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.lm_head.input.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.lm_head.weight": {
+        "tensor_meta": {
+            "is_emulated": true,
+            "dtype": "emulated_mxint8",
+            "block_size": 32,
+            "block_axis": -1,
+            "shape": [
+                32000,
+                4096
+            ]
+        },
+        "exp_mantissa": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.lm_head.weight.npy"
+    },
+    "meta-llama_Llama-2-7b-hf.lm_head.output": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.bfloat16",
+            "shape": [
+                1,
+                1,
+                32000
+            ]
+        },
+        "hex": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/meta-llama_Llama-2-7b-hf.lm_head.output.npy"
+    },
+    "input_ids": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.int64",
+            "shape": [
+                1,
+                1
+            ]
+        },
+        "int": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/input_ids.npy"
+    },
+    "position_ids": {
+        "tensor_meta": {
+            "is_emulated": false,
+            "dtype": "torch.int64",
+            "shape": [
+                1,
+                1
+            ]
+        },
+        "int": "saved_tensors/meta-llama-Llama-2-7b-hf-mxint8/token-1024-pos-1/position_ids.npy"
+    }
+}
\ No newline at end of file
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/input_ids.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/input_ids.npy
new file mode 100644
index 0000000000000000000000000000000000000000..6593c1ffa3b9085aace5bfe3b258f2915b04c94d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/input_ids.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1eb47bb5e8322411424c1d3e4312e472aeb5cf21135a0650a14111ad8ac8b287
+size 136
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.lm_head.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.lm_head.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..df9d29579609ee8b5f65f7f5a7e674b72d437b3f
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.lm_head.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fef09aca4d64e1cbf4dc77219595c05a79b449c4764469f46947c12e9c09cfc
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.embed_tokens.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.embed_tokens.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..74e23fac51f5dba00e33129abf4b2c7b6b3e0116
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.embed_tokens.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98c01f35cf299b5c3bad268f643f2dee32d1126712b435d4a2e38febdb63c841
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add1.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add1.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..74e23fac51f5dba00e33129abf4b2c7b6b3e0116
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.add1.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98c01f35cf299b5c3bad268f643f2dee32d1126712b435d4a2e38febdb63c841
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..3b584b6a757eabbec19b08e8d7ef6e7409b1b9a4
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.attn_weights.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61eb9ada5ac316af7282ebdb5baa7e0f6743dfb5cf23bc5cf07fd4a4568ff96f
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..07cac02aed478a1adc29a26c9fcff1ddcb68d4f4
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.0.self_attn.o_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50168efd64364b7850abbfa5e76dd0f8fd33f80b441e1e67761d7249b875640c
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add2.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add2.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..e550e99186fbf993b4668789f2effdb760a13bef
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.add2.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b8a5f6e8c7639360ef2f66f0443ff4f2f55581cfa57296acfa7e6b7096f86a2
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..89d51d730493f2b6d515342c3e4e3cd4f44f49cd
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.1.self_attn.k_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a83a238fcc2fdecc40a8082c62c96abcfbbe3df5972c304b86a3ff498235e782
+size 268435584
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..547e7929c67bd7e4afc25cbc67a140a26a021e05
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bafff44f40cf2f36dade98b079c4c0530fef17f035d5ecb64d5312642a0dab74
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..f0db41c89afbbc031aa1b2475271eaf903a733d5
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b2755d8a2b1441a51b2dec2cd8df3a198ac558f60101c17d0409a739bd526ce
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..1d1dd1d31a067c7a270c00c475f1f930d59047f1
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.gate_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9463cfd66b2e4b1f1d6f6b6c8f9e236884677cdc3c118d562f2cae6a7d1ccb4
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..56e9252c6e71a17645100f76c8177a9226af3dd1
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.mlp.up_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:30d4b4f788bf5bbdd89a694d6bbcb8624933b7f41d65683b5a334eb1f8078d1c
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..34257cc8300aaaa067792d0ec4b1140fc405c845
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.post_attention_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ed9eab217ec58e4a686e6a8d7f15b223db17cd57ea1d5a5bc6165626d341438
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..40845e4655e6eda9b00e4deeb36171a2f4c4fce3
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_output.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdb348d73413f154f082c3f0b805fd576463c4f5198ea81471916c30eb72ea44
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..69032c4b1108785c1f575c8325e23c9ae8d1c742
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.attn_weights.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d7ecea0b0668e7e60ced58dd07fd902987cd67355b965a645b0b8dcd4bc92db
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d4c12d6122050d813356cbbe80f1e83721f20acd
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.10.self_attn.k_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42dda51383e81913f469fa0618581c6b7d885a5c45e85de808d613a0ab8c4681
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..ae1f901dbf9d44118e5b55d9226516d71e3c8976
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6c0c89f1eb3c586ddc589bd21d7ad369b090ca1f530600e4af7038b004c65bb
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bb670e3dc1afc1cc05ec84a4a9563505024550a7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.mlp.gate_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbd28bfdf5d88cb963e70acf8cba5f6cd50f52e227f45529c75e635df3ca9d59
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b1a0080b978c894c70acf84f10eae6014a80f7dc
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.post_attention_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b024e7037e480b7d5653911ba984aa1cfce6bc8f9676099356b73c9ef1f8257f
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.cos.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.cos.npy
new file mode 100644
index 0000000000000000000000000000000000000000..c58e9ca3966c385461c89a37bf4457e224b4e8c5
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.rope.cos.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ebbd8f3476ca6bc4fdb19a4f9666c225e1a81e13bf0817e834fdcf052b8deac
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7ee34fb881ba3222effc2e1adb657f6bb2ed40a5
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.11.self_attn.v_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db0e54ead296034456cb96d7195f750b62efd699ef2ed1e28824473713026df7
+size 268435584
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..93b4c42a75bf0387455e05e61c68bc35704dcdf1
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.post_attention_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7dbc1094b1d3a2d5f0ff55d6ef2791cfa799ac0bbdab9d03ec1216330739bfc
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d2813ebf0ee242cbd169cf1c392fd87670e4561d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_output.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:820df60c5718222ab9f45d46fd2e3b97acb2d8e0ac2a400473d6f57b1a803109
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_masked.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_masked.npy
new file mode 100644
index 0000000000000000000000000000000000000000..1d5e77d4f81d7a11c5c79f5c71ee09c3c3556cdd
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.attn_weights_masked.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a545e9fd2f65453a3d6cbf12efb7d05f6627121d541ba924cf12a4729ae27d0
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9a82e08ef50ce6219be1fc5d0702569ec3c51ff1
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.12.self_attn.o_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2ea664198ebbdffb55fd81e0ac1365637968546a20f8d844573d6965d4d2665
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..42cf0b2b705aa23a775cdb7295d2b8e795bb182d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9fc568cb57303846160bc4a253191b6f32a7473035d86b5f0d093cf8c3d7be3
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..56f9fb4f3b584b63ee7af8b9bcb187e32bfdf068
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a926f792ef1c418ccd325ed6836da5a0a86332ec938268387621972f377ce9a
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..274c250316229bc76d434ced974ab2e909be1c72
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.mult.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6329ef6a7228e34ac2d576c86c0c8dd26fec33b3708a448823dda798af08974f
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..c482ec71a44ed0260765cfae3d379120f5e728da
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.mlp.up_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4df7ad6d2623e0762467c6a653cc6704ca9ae4227f417121e8c85ac83746bd9
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..fc798c58405ec0ca9cd0674200dd036d35ca7357
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.attn_weights.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa5b5fed2e2ae101285c301d5419346c02f3874f27567661e17a34154dba2e98
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..31dadc8512992592de0472e099c2828139d560ef
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.k_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e80ada026fc5b563ddd886f35aa96e87fb08839b6c4c2109a2216b38527691a
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.sin.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.sin.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cd3ee152df8d523dff2b90f07b4ab1161ba415f7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.rope.sin.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e564b613bab09285dcfc1f321f81072871ec2c377bb6b0e1fdb7aa800c7922
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..31dadc8512992592de0472e099c2828139d560ef
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.13.self_attn.v_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e80ada026fc5b563ddd886f35aa96e87fb08839b6c4c2109a2216b38527691a
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cecbdefd6ac92b8e0b1c69234762d75b35e68150
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec0e9cdac8cef885af453f289d0d9e158bc79bfa177253aeb224bf4199b7b2d2
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9f89c895066a5c9703176e38cd78bd79b08781d2
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.post_attention_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7cf2bc506c36e3c465444a113ad6def65a009c3a805dbbea074d18f661299f6
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.key_states.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.key_states.npy
new file mode 100644
index 0000000000000000000000000000000000000000..a69f08ce7d9cb751cdcb08e99aee7f2a495959c3
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.rope.key_states.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa939287e8ffef277120885717c10c71be7bd06dcdaecf64f01b3f6eb5a3a746
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9aeff8e359478c6c03b60f791ab3059953c4d1de
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.14.self_attn.v_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b538d5a28019068b41362e5220c61377f3c30a715ca4869dafb4b0bee83b04e
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..e24cf32bf7cc5c98e46c2d56b6aa8188f12881c6
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.mlp.down_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:297e64290d5d188c109d0c96bdd010f8734ec9094733a0f803e0374a18a3ffb7
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..48050a3674615ddac1eef39f85c479b196da4340
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_output.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a99322d7b560c81b4b34c794b7937bbdba9254efc3cef3b3a4bad03f75f1d89e
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..c00c350d0ac9e63b49c64ab80007d95e82318811
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.attn_weights.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e3a78cba4f675d0788ba28e412aa042b5c109df41a0c719786032705c294a52
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..888539793d5b72d2df27cc7ca7efac5866299566
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.15.self_attn.q_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb64cd56adcee725477e6a729aca5159568531f649e42acfe42a0a6fe4b8abe7
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7e3f077c8bba21139af1fcad963251656509aa67
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5f07de098c161ef165cb4c51337b65bd0b812d7604b088f8911bda6f3025657
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..075d0d489714854fc765c4c1284a342c31d2c280
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed5832fd59c776f9c137f5419a7701b184048ebea5da5bcf80cbe83c2d9a17af
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..ebd032ba8ea457bea2abad137d5d952310bf8a93
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.16.mlp.mult.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8999e634e38ac081b6a98bcb88ee18e3ba8203583b6dfdca6ebfd84d4f8ffd14
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..45ea86af512cdf09b74cb414a4fb37902f385ab9
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.act.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd383d766bbd38f3e3294aebd69603241c2d56a25260a600d804d95585daccd0
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..3c217d0fca7933b5e0467236c1528f485d6ee901
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.mlp.mult.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e8757639a1a6a03d0b4e01cb191e85b97dcb8d4a9764cdeb9f7192c771430ce
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_masked.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_masked.npy
new file mode 100644
index 0000000000000000000000000000000000000000..62a583439485f65d698898d070e8cdd35c655126
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.attn_weights_masked.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:456a111705ca2cc90231560dcf347d07ca60b06e69d1f9f691b4a59e349c8abd
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b77f93aec69dfebe5d7d5d1a1417f8298698cdfa
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.o_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d99c67f7c780060c152ffb68dce9c7d45fde84bbcf0617c4285ed37716c97228
+size 268435584
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..496e67b5015af14256782a214e1127e60ce9e6ac
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.17.self_attn.v_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2bc3aa3e368f6817cff5ac33b2f63e24e0c6882f846225523dcbdf24d0ce44fd
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..0d981d572430020f2ff4506c39cb28eddc3acdda
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9199b797f421765607e20cdd0aa549dae6dcc2b0b1d7aff3fac28be58b2bb2dd
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d34676943f34f8d2370752db42021bd3c969990a
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.18.post_attention_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:101d3ad7fba73a7390451a985a31ab4559a9f3760a5ba713bbc59542098d071c
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d4f6c2ec9989e5ae64241ec32f06d672ec7ed6ef
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74c2bc832dfe6895cea948f58498661b6a56507659ab6241fb2e7221eac47fb7
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..3a521aa15805865417b6e22f4e300b9c3feb4e9a
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.mlp.down_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:051498e6cec73f18123ad2641a9e3775de3dfc29dd7067a9b6723d6abbe83d8a
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..acd3c8c44705ec1617d1ea2a4ff906b3e5b14596
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.post_attention_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f8aa162fc6556cc4948aa711695084b902f167de285bb1289f48ffbe410cb66
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..5359369542a7e941f04fe26884086d302118500b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d316dd1e431c403ae3d0cc891c16c13abd70fef404d9a41346842d967f5e4845
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_softmaxed.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_softmaxed.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bae379e173c543699aee73afd50597bcb3faf549
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.19.self_attn.attn_weights_softmaxed.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17cc502b3f0d8fa3f52d7c3292055a318f473fc2116c299449a3d8175e08b212
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add1.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add1.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..e550e99186fbf993b4668789f2effdb760a13bef
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.add1.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b8a5f6e8c7639360ef2f66f0443ff4f2f55581cfa57296acfa7e6b7096f86a2
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..03e6338c00b57e67b71a39d6faf5ac6249376c1b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.mlp.down_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe5ab7174d4d2c295976271c15db35df0a3dea232a3ee3ef00dc339f7f939a43
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.sin.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.sin.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cd3ee152df8d523dff2b90f07b4ab1161ba415f7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.2.self_attn.rope.sin.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e564b613bab09285dcfc1f321f81072871ec2c377bb6b0e1fdb7aa800c7922
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add1.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add1.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b4631f9ed3da2724e1914ebcfa68f4001b7f9cec
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.add1.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54282bb299d782fb590a5ea1d5d1019f9cd2c67ac51899c37831d223b15671a5
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..128f3de59edf818737cf8eb4d03f61589c7c8c42
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.down_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd41df22f421a2a68ff785706146530aa958a4a1f3400864bfe8e00feb399d25
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..633db1588bf60348d819ac5aab0020a201bfc07c
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.mlp.up_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef0cae5b12de9a28a190af5f924c62d16aae70da187f79617146a8e48abb8c80
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..f61006421df270fa7e042ea4d94d9c42a5e6d094
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.post_attention_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63953f0bc26f8192b7df3c458f0142fdab98fa4abfd023c6cf80db1325e86aef
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..e0163a95ec1cfbb2d21ffac40a99c5cbf895b2fc
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.20.self_attn.k_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce0f9b94e278a30c5cfdd7b13f2af20966c0af1483e31621beed8532039475e2
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add2.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add2.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..a34e2cd27ebce00b943150b27e3685d26b431e52
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.add2.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ce6a1307434f9faeb439f5afeb264f6610fd09f066392ff2ae92868a228f058
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..209d5c2e176f6ff713ad7aca9a1f06adfe0cd197
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e6278edc2fdc45d684d87f286970031299276fdd0ccc01583bedeb1d4cd7a6e
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..371dccf41711dffd4a4e4a5aa95319d321b23fbb
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.mlp.down_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:030aa68bd0dd572e0277e825ccfba13e04a24e039dce6aa2028e41bcc4a675b5
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.sin.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.sin.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cd3ee152df8d523dff2b90f07b4ab1161ba415f7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.21.self_attn.rope.sin.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e564b613bab09285dcfc1f321f81072871ec2c377bb6b0e1fdb7aa800c7922
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bd1c9e9ca55556b85f059a7a3b079dcbd4f72c7a
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c11e5eea91f5244a9dfa2ce4c8d3e3dcc1fc319194a1d6f3f9ed3ec7144642aa
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..0aea7ed2ec6a1efab96b51829691247ee6c558b1
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.add2.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47b1e7058f4b9f950ca921c4c46dab9f98bee25641d3fbc69997cbae91667e51
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cee84d56b709a41dd0c6c80221c1f2c4f6a5a2f3
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.attn_output.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef3a14090c430dbf5b8a09de82c09b254a8abba1ae82e1d3fe9c1eee806a2c15
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.key_states.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.key_states.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b0e5288bbc4f46a20926e518370fe73dc841d397
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.key_states.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da9a9abb4f827d12804cc0a47a00a3f78942f937686f8be6527a1ef6c97f16fa
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.query_states.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.query_states.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d37aa7aac7aecf652b23580c055401c295b7542d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.22.self_attn.rope.query_states.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41010578b9b6a5734c27c07b82f4e3f8be689c55b4df8cca82b5773691222b36
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add2.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add2.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..ba9d7c1755c9c4fcb208fed9249d72cb3dd67f57
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.add2.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8027ee05549d0785bfff28f688b864eff6c54f3e129c967dc4f46972069820d
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7180a9a92fb9fc76da555c88e2933f799eab0227
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8841370bed7f88d514449bf7b35260eb1d07fdabc770ac449d9715e008c8aeae
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..4c0ead654d4bc6e9fee4f712cef76130890c51c5
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.act.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:064e68437d0a2aacdd0f7eba473c9d0c01eeba7ae3751a277985448979db2ffd
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7b62cecc675d9fb1dffa4b441a3f204f2fe58a79
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.gate_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66330fc9b86a44e3e93f8ee671ca7dc22f90507126c3c96e0261eda763e6959e
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d8691639b68a5e132e132aeb05c227b2d66e42e8
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.mlp.mult.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b685b501453d1f4b0c2ced122bd4ab19fd547e7318cb787e7e5d9da79ddf65fc
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..587119efe111d185df273658a359f954ce015fc6
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.23.self_attn.v_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a003c458846f053b231cb8dd9f89335eb53bace603c1da4151668f4c9e28a621
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..f72e486d75bddf071e3db837fb16c52ccfc93932
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0de8d465b66ab270901692f7f7b4208f00b530a9021a548f62933b71bf8fe88
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..8dd28776385e2bbc9cd37dc3e4c695323f26d810
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.attn_output.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d362680c60b393b3845a44904fa64144d98f4ec46e29f68408206b1fcdaca44c
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.query_states.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.query_states.npy
new file mode 100644
index 0000000000000000000000000000000000000000..8f2d99041a3012aece0bdd8a97d98aaa011a03eb
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.25.self_attn.rope.query_states.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aaefbeb9f110e2882bbfc3278a955e34baab38161ab4933ec539445225d3d689
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add1.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add1.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..50c0fcfc0ed7d6002bd57d816f449ba4667212c3
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.add1.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f238b55f0c2b0786fe077bfb7e9780c95a92482ac26ff81c5ee41221772bc6ab
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9c6218eaca5bde40fcac54e0e5a54d8a40d123bc
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd98ac4dd45fcf2e6302e2bae7c18d935c1acc2c772e12ba30fcbdd2ca2605e6
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..92e6fae16095b56d439ef718f1078889efdd67dd
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.down_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15baf5b71c4c0c0f8ef7ada63a98901b8415a1d9933db91d0e6343c9108bd610
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..4df87abc48533b61d8bd563211238c83629ab05e
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.mlp.up_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edcc29a776be885b9ad33d3884393b437764481c83a1481671c51dfcd64f111b
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..90c522f6f2c6ef35d47a26cea5b18eb38c5c08eb
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.26.self_attn.attn_output.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f91f84a998faf60febf5cc370068cc25087dd28529454af6895effc80f749cdf
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..153aedeb4d00b2e10e10eb59e53ed2fb482f84bb
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a349cf50cc847d976c42bec9208b2d19077ea11cf4636a040ac65180160d53f
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..930f2ec2d149a6c585c243d0acf0ad2b7bc50a94
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.mlp.up_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75be35abbc66d452b7111a9045a5884666d7a9a3813f61b00fe5f83d66db6d4b
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_softmaxed.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_softmaxed.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bae379e173c543699aee73afd50597bcb3faf549
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.attn_weights_softmaxed.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17cc502b3f0d8fa3f52d7c3292055a318f473fc2116c299449a3d8175e08b212
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..4328656313f0c5426ba47a5824c7f993f037597f
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.q_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79a233fabb103a7a4dad236fa8761e0b36b194f0bc14725d0fb2907a2b1952bd
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.cos.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.cos.npy
new file mode 100644
index 0000000000000000000000000000000000000000..c58e9ca3966c385461c89a37bf4457e224b4e8c5
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.cos.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ebbd8f3476ca6bc4fdb19a4f9666c225e1a81e13bf0817e834fdcf052b8deac
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.sin.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.sin.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cd3ee152df8d523dff2b90f07b4ab1161ba415f7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.27.self_attn.rope.sin.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e564b613bab09285dcfc1f321f81072871ec2c377bb6b0e1fdb7aa800c7922
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add2.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add2.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..a70b1f34be1ce0d9786c7276c6b9601103d2cbfd
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.add2.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:705112a7935a98d93424c843d5a83907fc19004ae068edfdece4d5cfb343e5a9
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..c2b1852b695b851959dc212f81b24dbe75d8796d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa38131671e90cfcfa4165e3e94b869954da4f473b0b8ac203093daee6ec1bc1
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_softmaxed.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_softmaxed.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bae379e173c543699aee73afd50597bcb3faf549
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.attn_weights_softmaxed.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17cc502b3f0d8fa3f52d7c3292055a318f473fc2116c299449a3d8175e08b212
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7717d08eae08f8cafc5d7863e72640849897b63d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.o_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6920ad87478a5eed7b8fca402ec8002bd1ffb7709a97c71ae59f66452cd4b1f
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.sin.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.sin.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cd3ee152df8d523dff2b90f07b4ab1161ba415f7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.28.self_attn.rope.sin.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e564b613bab09285dcfc1f321f81072871ec2c377bb6b0e1fdb7aa800c7922
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add2.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add2.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9d383aa408404c316df1f266927f682b394a31f0
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.add2.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b30cda3fb1fa127f0783e168950ee039ec31595b87b65d1bc8b78fcdd18e6613
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..18f80ffd23ee3a8429865edce9fd6691bfbe8089
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.act.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0328e3932695f8f3bcaeebdce33215200109cdadc105d5db82c7affe36466a1
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..21f1e9ee7fe084543b7d743bba683f594b79bcae
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.mlp.down_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef15c0a944ea380e0eb9de87fbee468bbbba0482f2353e59602ac8e71771bfd4
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b4824cb9d96eb485ccd5336cca4256000737e124
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27f9200c938ae803be258d38e85b19f963b96aa93c600061c14abb6e446d3fac
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..05537b2489a39901c0981e4f427325e0608294e7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.q_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf92cdc98b250c6df760251bf5f7ed3a3f418accea52b3bf91e940c8b29d42eb
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b4824cb9d96eb485ccd5336cca4256000737e124
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.29.self_attn.v_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27f9200c938ae803be258d38e85b19f963b96aa93c600061c14abb6e446d3fac
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add1.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add1.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7bb6fae2dc09b3e84bb474fe3d08a18942f5ea5d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.add1.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62c9deb422c71f7ef0ccfc4d1dab1d808e2fe425e2ac1f15f9f439ffc53c8944
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7c7519f853adb258324e372284c1b1ae4a8f95d2
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0691c719ccd40bd3f1e1a9484988df5ebfff600ba5a1a17e3a7225249fa34ae4
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..a4174a4b3c5b186194510f073d260ea7e2fd1e9a
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.gate_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d39f464de75cb7cd3305ae339455c65dcb95d3fd1e8b7084b724d67dd094fe52
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.other.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.other.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7d137ba0b44ce51e21fd8ead9d86565c1f2cf1c7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.mlp.mult.other.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d02b9d6f8c2a03484b07a5c5f5287e6e8cf0a6af574a10a64d1e3738a5ee95a3
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.sin.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.sin.npy
new file mode 100644
index 0000000000000000000000000000000000000000..cd3ee152df8d523dff2b90f07b4ab1161ba415f7
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.3.self_attn.rope.sin.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e564b613bab09285dcfc1f321f81072871ec2c377bb6b0e1fdb7aa800c7922
+size 2176
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..575a8c1e2cc13d6a247cabc84519bda85e36d82d
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.down_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c528112072f868385fe0097e41da529f67247165e3ccacf6354f19832f37ac7
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..4d4669ca93740792b25204d1d1e174db3a6ebfc4
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.mlp.up_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:539dc4733bd63805ad4f42a05d213c3ae44425bff5b4db9e8b09e26ca99b89ad
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..f501626702e5aada39f01fba19b655ff546990d0
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.post_attention_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c8e54787df4958c7444e5185a2a94c9b522913e70f451e4a2bd087bfb714887
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b97b5bc34e16bede7933cf3d50122db6955b1117
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.30.self_attn.o_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:356e8bcb63b5e67b41e1d383e21294ea47fe369812fa307bd7c28476746feaeb
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..816cfd32ef641a709ebbe5721308f2ffd0488c6a
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.mlp.gate_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5633f56b137d507322e9aff500b60271c4a506da5b351315032001872090d0c1
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..774538cceea1d3dc96541f94d7e77ba6c3aa0f2f
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.post_attention_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2259d51bf4f4524a4dae737936970143ce99d9066bce43947664bbbfb17f875c
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..8453d09434d2eb30e1c9ec132a7e0f6341bf64af
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22a4c9c3935fc85a4d6d02102987ffde8ceeb0dc61ff3d19ff0ba7114751d48b
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..8cd0e20759ad64124911f34e654ea229c6599206
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.q_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d60417cbfe25c14a0251cad781fe99653316e4dab3a01c724f94642471520f6
+size 268435584
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.key_states.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.key_states.npy
new file mode 100644
index 0000000000000000000000000000000000000000..b2f006b67e1238c739609fa134fb130edf4078c8
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.31.self_attn.rope.key_states.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15da9bc4a94a4bfa26ef71ce4f29302f86753c6746d2bf6cf160cd1441b1f2e7
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..37704f037579eeaf97b3246e9a7b0195e3395208
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.down_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce134c7505ab99672bd9f50b76f06c228797b943fd4303650611a45087528683
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..fb5cc24e3a09d6c13484fa760cd87b00ea96a0a2
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef2d8f5db03d0c00f44c512f8e9aad57ab37bf1a8f366bd49e0537e7baed9926
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..d27953dd2236a89847ec9c83cf3b40abbf2fbfad
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.k_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:425264104cb615035513750b75b78284b1524f6faca34a383b48a3cd47aadf54
+size 268435584
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..c697856c8fcf88bbf5be5d7bd7c1ae1873d21eba
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.4.self_attn.v_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f563473a251633661ea34b881c24511b16566fc05abd4eda0875cab1c5514c55
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7ba94545d5260bcfcb48a3d084d467641118271a
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb74aad8283691dfa2e8f76aa6bc715bb9bbc81b287485150daa78c41bfd44fe
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9485bca4544fff74f9a1971be78b293cc0992842
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.down_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74388f83e7aef17e7670f6fe0bde906eb8a2cbda837b2ed3077ad658a74e8b50
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..1bca169beeea6150129f6acc021fe77aa5145bee
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.mlp.mult.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c21c305667bf2c2f02ff082180e3ccb5233da514e1a1f667df00329b3d71fed
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_softmaxed.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_softmaxed.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bae379e173c543699aee73afd50597bcb3faf549
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.5.self_attn.attn_weights_softmaxed.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17cc502b3f0d8fa3f52d7c3292055a318f473fc2116c299449a3d8175e08b212
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..a500e0fa5840c4a4766d014e26ad69fb3b758c47
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.act.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18273e97af5d16ea2e6d730d85c90cf61d049eb4b87837243d22e7b760284e9d
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..a500e0fa5840c4a4766d014e26ad69fb3b758c47
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18273e97af5d16ea2e6d730d85c90cf61d049eb4b87837243d22e7b760284e9d
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..f5a6424570f02436d45f2c5cb1991c5b4a833c52
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.post_attention_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b6b6d2689d9b5e6dff7c14c5c4c5e50df0847610f1041ace475063dc4c5b18b
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9cf5fcb6c2b0209691e907d4aaac367030dc5d2b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.o_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28741477603cb33bf0cfce037b15982df4ae676f0171f727c0f6c62a2e3ab37b
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9cf5fcb6c2b0209691e907d4aaac367030dc5d2b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.6.self_attn.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28741477603cb33bf0cfce037b15982df4ae676f0171f727c0f6c62a2e3ab37b
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add2.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add2.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..2eb2af5e828b522aedda655701145a6e2c507f35
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.add2.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6413b76abb4735151f3f4eecf6fab7b1d942130364340cb223134cc63ee5f6f9
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..8454b81b146a7c2b08864877eb45e684e90aa5d1
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.input_layernorm.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b80b28545278ef0e448d22ee6199b44ec8b09d2dfd5d6cfa9940a9c8ddc83a13
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..204b761d27e6263deb1591ea3bbc359180d6e126
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.mlp.up_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1106df5bca39424becd7794e3f62acb1a10b642bfc757b6974dbba98f22b9d89
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..90c522f6f2c6ef35d47a26cea5b18eb38c5c08eb
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.7.self_attn.attn_output.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f91f84a998faf60febf5cc370068cc25087dd28529454af6895effc80f749cdf
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..6a93553e551078afee295f36e6121ae168231efe
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.act.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6183bd73acbee4491ed04a550ff614466937761c74a9b5fb19f4fb3069ffa697
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..dee5061df5d367a2d61b4788d6498fda70d5fc7b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.down_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a80fb07cdbeadec4e1fe9bbd1e31fe7231b965dbc92f28032c540f0ccf96389e
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..ef4fd3021a7b42a4bff04f21a5e1a50c39d3d8f6
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.mlp.gate_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72e6ed421503cbe70fcce0c9eaaddee76777b8fc22d9cc4d3039c3cef184393a
+size 721420416
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..09cc7426d2714438c1aa7c971e52ee227bc65621
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.post_attention_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8f8becac4ccfcfde1c9c9cb77a6ca904f9fd48b9831d02e1f429e7632f075a4
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..9d16299f4eb0ce12e186fb1f1b10d99e280d5141
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f521ccd60564ea0d7a5160168d2c1c96ff0abacca64247478971a5d28fc2f782
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_masked.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_masked.npy
new file mode 100644
index 0000000000000000000000000000000000000000..7a5fd13eee6a1c81beffc18cbe25642174b8b06b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_masked.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c058154ac2fcf65e0856e6107d7009487dd744be6b51ccac1900fab4f96706d8
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_softmaxed.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_softmaxed.npy
new file mode 100644
index 0000000000000000000000000000000000000000..bae379e173c543699aee73afd50597bcb3faf549
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.attn_weights_softmaxed.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17cc502b3f0d8fa3f52d7c3292055a318f473fc2116c299449a3d8175e08b212
+size 640
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.weight.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.weight.npy
new file mode 100644
index 0000000000000000000000000000000000000000..185b03b7d5c3c3ebd0a2059db563c12d7e037e12
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.8.self_attn.o_proj.weight.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b98329d3a3120a636f1838cc571c51908a667009ec54a80256e8008a9a00a38a
+size 268435584
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..34e0c3b4fcc2b6d39f6c0efedd520a488066039b
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.input_layernorm.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b4008fec8b801a70c4d979f22b74019010ac439a12acd8fb473dbe0641b876c
+size 65664
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..e4dac2061c096ea84a24bf4b46be40d3ca916d25
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.act.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfab11a4733db8256c8a1e712180af8fa2863c3732e21b1eb2d1917708f53165
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.input.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.input.npy
new file mode 100644
index 0000000000000000000000000000000000000000..62126a6378014019d99d66b9c59e6bb78d189a37
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.down_proj.input.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d663936646810d5e9d4cfea2f36c5a193d855042ef07e86c8a17835d61768efa
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..e4dac2061c096ea84a24bf4b46be40d3ca916d25
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.gate_proj.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfab11a4733db8256c8a1e712180af8fa2863c3732e21b1eb2d1917708f53165
+size 176256
diff --git a/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.output.npy b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.output.npy
new file mode 100644
index 0000000000000000000000000000000000000000..18add7af2f7fb868bcef8985385caeb587394f76
--- /dev/null
+++ b/meta-llama-Llama-2-7b-hf/token-1024-pos-1/meta-llama_Llama-2-7b-hf.model.layers.9.mlp.mult.output.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20dc3c1959d76df9d7e86b33ec36d5df95f496ed5e51007372430ff11bb729c3
+size 176256