Synchronizing local compiler cache.

.gitattributes

@@ -7711,3 +7711,5 @@ neuronxcc-2.17.194.0+d312836f/MODULE_b22e2f5845207f7f190e+613edded/model.neff fi
 neuronxcc-2.17.194.0+d312836f/MODULE_b73b1eaea0f6f5dfe7f3+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.17.194.0+d312836f/MODULE_c299f9936437dd67f341+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.17.194.0+d312836f/MODULE_ff8fad012e7766ca23ef+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.17.194.0+d312836f/MODULE_4caf7deb700973d9f449+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.17.194.0+d312836f/MODULE_a6b635cb5d8d2a5539a7+613edded/model.neff filter=lfs diff=lfs merge=lfs -text

neuronxcc-2.17.194.0+d312836f/0_REGISTRY/0.2.0.dev2/inference/llama/meta-llama/Meta-Llama-3.1-8B/6fd804b37216317e4f8e.json

@@ -0,0 +1,45 @@
+{
+  "_entry_class": "SingleModelCacheEntry",
+  "_model_id": "meta-llama/Meta-Llama-3.1-8B",
+  "_task": "text-generation",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "neuron": {
+    "auto_cast_type": "bf16",
+    "batch_size": 8,
+    "checkpoint_id": "meta-llama/Meta-Llama-3.1-8B",
+    "checkpoint_revision": "d04e592bb4f6aa9cfee91e2e20afa771667e1d4b",
+    "compiler_type": "neuronx-cc",
+    "compiler_version": "2.17.194.0+d312836f",
+    "num_cores": 8,
+    "sequence_length": 4096,
+    "task": "text-generation"
+  },
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "use_cache": true,
+  "vocab_size": 128256
+}

neuronxcc-2.17.194.0+d312836f/MODULE_4caf7deb700973d9f449+613edded/compile_flags.json

@@ -0,0 +1 @@
+["--target=trn1", "--model-type=transformer", "--auto-cast=none", "--execute-repetition=1"]

neuronxcc-2.17.194.0+d312836f/MODULE_4caf7deb700973d9f449+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0aecaf44b94e47028d46b0cc8fbe1e995c4b589422c8abacdc232e0a69787e69
+size 466825

neuronxcc-2.17.194.0+d312836f/MODULE_4caf7deb700973d9f449+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e17c3317e06d2eb0c20d2821434698230925dcdc940dff9d4220beed0f53d2b
+size 30618624

neuronxcc-2.17.194.0+d312836f/MODULE_a6b635cb5d8d2a5539a7+613edded/compile_flags.json

@@ -0,0 +1 @@
+["--target=trn1", "--model-type=transformer", "--auto-cast=none", "--execute-repetition=1"]

neuronxcc-2.17.194.0+d312836f/MODULE_a6b635cb5d8d2a5539a7+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02143068ab84889c2a2bb5aed5a9d68a89105597b10429124f73c1bb74eac529
+size 363567

neuronxcc-2.17.194.0+d312836f/MODULE_a6b635cb5d8d2a5539a7+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05b4a67ecce4ee4517807c4cd69bb08b8b7033f70da22064ecef6bc1bc37b3ca
+size 3646464