Synchronizing local compiler cache.

.gitattributes

@@ -6954,3 +6954,5 @@ neuronxcc-2.15.143.0+e39249ad/MODULE_8ce80c0a7e475086c914+39f12043/model.neff fi
 neuronxcc-2.15.143.0+e39249ad/MODULE_f992331f929adb7094fb+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.143.0+e39249ad/MODULE_2694a26b0ce70a599f06+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.143.0+e39249ad/MODULE_915e1c451537f102263b+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.143.0+e39249ad/MODULE_5e4561088b45bbe15d94+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.143.0+e39249ad/MODULE_6049c0a59c7006c2e8bf+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text

neuronxcc-2.15.143.0+e39249ad/0_REGISTRY/0.0.28.dev2/inference/llama/meta-llama/Meta-Llama-3-8B/8e885ee6ec8cc4ab8c3b.json

@@ -0,0 +1 @@
+{"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "eos_token_id": 128001, "head_dim": 128, "hidden_act": "silu", "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 14336, "max_position_embeddings": 8192, "mlp_bias": false, "model_type": "llama", "neuron": {"auto_cast_type": "bf16", "batch_size": 1, "checkpoint_id": "meta-llama/Meta-Llama-3-8B", "checkpoint_revision": "8cde5ca8380496c9a6cc7ef3a8b46a0372a1d920", "compiler_type": "neuronx-cc", "compiler_version": "2.15.143.0+e39249ad", "num_cores": 2, "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": null, "rope_theta": 500000.0, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 128256}

neuronxcc-2.15.143.0+e39249ad/MODULE_5e4561088b45bbe15d94+39f12043/compile_flags.json

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

neuronxcc-2.15.143.0+e39249ad/MODULE_5e4561088b45bbe15d94+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40599ce76c4a96baa779e2360200f76b062a2b7a63f6a5de8d5cec3398feae49
+size 331813

neuronxcc-2.15.143.0+e39249ad/MODULE_5e4561088b45bbe15d94+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f4ce5f7ed97321c65d74ed35c07389f9e8efc4e4ed810b529dcb2079c67c6bb
+size 7783424

neuronxcc-2.15.143.0+e39249ad/MODULE_6049c0a59c7006c2e8bf+39f12043/compile_flags.json

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

neuronxcc-2.15.143.0+e39249ad/MODULE_6049c0a59c7006c2e8bf+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91050013d1b057968fd267677a1f32ee078bf94ff04e7245f6294d2004b733e1
+size 399094

neuronxcc-2.15.143.0+e39249ad/MODULE_6049c0a59c7006c2e8bf+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28c758ad91e41a86c6fa03e50edc6254b3d30f185cbc8bf90cd52f598a2919fd
+size 6022144