Synchronizing local compiler cache.

.gitattributes

@@ -1861,3 +1861,5 @@ neuronxcc-2.16.372.0+4a9b2326/MODULE_8a11747c1cf643ac58ca+613edded/model.neff fi
 neuronxcc-2.16.372.0+4a9b2326/MODULE_90381dde99583f54a60b+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.372.0+4a9b2326/MODULE_9ad3b35220e91074c83c+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.372.0+4a9b2326/MODULE_a7513d3ab2f1b0306086+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.143.0+e39249ad/MODULE_4ed353419620544e02a0+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.143.0+e39249ad/MODULE_e8c36bfc1fde3e269180+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text

neuronxcc-2.15.143.0+e39249ad/0_REGISTRY/0.0.28/inference/llama/NousResearch/Hermes-2-Theta-Llama-3-8B/af163c75a04bbd5785ad.json

@@ -0,0 +1 @@
+{"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "eos_token_id": 128003, "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": "fp16", "batch_size": 4, "checkpoint_id": "NousResearch/Hermes-2-Theta-Llama-3-8B", "checkpoint_revision": "57a73110702e7b05ba3f39fef36297454c680725", "compiler_type": "neuronx-cc", "compiler_version": "2.15.143.0+e39249ad", "num_cores": 2, "sequence_length": 2048, "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_4ed353419620544e02a0+39f12043/compile_flags.json

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

neuronxcc-2.15.143.0+e39249ad/MODULE_4ed353419620544e02a0+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:596c6c5105874127df0e513fa6d53236e50aa6abe9484e1c5f8d76283b834d44
+size 363576

neuronxcc-2.15.143.0+e39249ad/MODULE_4ed353419620544e02a0+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11fefbc2a189016d4ac1aab75f1db906cdf6bfdb940d45721711d1337e3a241d
+size 9780224

neuronxcc-2.15.143.0+e39249ad/MODULE_e8c36bfc1fde3e269180+39f12043/compile_flags.json

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

neuronxcc-2.15.143.0+e39249ad/MODULE_e8c36bfc1fde3e269180+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77511de6eac2a81bc4d70a449700e527919a6fbc8c7d8bde026927437262c5ea
+size 301970

neuronxcc-2.15.143.0+e39249ad/MODULE_e8c36bfc1fde3e269180+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f25746fd12d86f5bb45c662b5727ab9ab85a81ae8016a4863aed262e8fd2165
+size 2868224