Synchronizing local compiler cache.

.gitattributes

@@ -7180,3 +7180,5 @@ neuronxcc-2.16.372.0+4a9b2326/MODULE_8ce80c0a7e475086c914+613edded/model.neff fi
 neuronxcc-2.16.372.0+4a9b2326/MODULE_f992331f929adb7094fb+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.372.0+4a9b2326/MODULE_8b8991d92b27943066fc+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.372.0+4a9b2326/MODULE_e90f5722b4a5608d3821+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_5e4561088b45bbe15d94+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_a822fdf490eb818df0ad+613edded/model.neff filter=lfs diff=lfs merge=lfs -text

neuronxcc-2.16.372.0+4a9b2326/0_REGISTRY/0.1.0.dev1/inference/llama/meta-llama/Meta-Llama-3-8B/c779aa50e0bee568e4f9.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": "2.16.372.0+4a9b2326", "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-2.16.372.0+4a9b2326/MODULE_5e4561088b45bbe15d94+613edded/compile_flags.json

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

neuronxcc-2.16.372.0+4a9b2326/MODULE_5e4561088b45bbe15d94+613edded/model.hlo_module.pb

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

neuronxcc-2.16.372.0+4a9b2326/MODULE_5e4561088b45bbe15d94+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67681a783f74972b96c4d3aab72798a8e72b83c9043a1b7efe8f687c7353ae84
+size 7803904

neuronxcc-2.16.372.0+4a9b2326/MODULE_a822fdf490eb818df0ad+613edded/compile_flags.json

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

neuronxcc-2.16.372.0+4a9b2326/MODULE_a822fdf490eb818df0ad+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39c16b17b8ae15c12b75e9d4555e205e49562e8540ef09f8bab2400f56313c35
+size 401014

neuronxcc-2.16.372.0+4a9b2326/MODULE_a822fdf490eb818df0ad+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1de97be9c08604a231334565ffdcd33dc73d1f94b2f255369cd20a2501f5b3f9
+size 5970944