Synchronizing local compiler cache.

.gitattributes

@@ -6958,3 +6958,5 @@ neuronxcc-2.15.143.0+e39249ad/MODULE_5e4561088b45bbe15d94+39f12043/model.neff fi
 neuronxcc-2.15.143.0+e39249ad/MODULE_6049c0a59c7006c2e8bf+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.143.0+e39249ad/MODULE_3089fe5c0ae178af3a6e+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.143.0+e39249ad/MODULE_4a4553dc6cf05a499016+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.143.0+e39249ad/MODULE_cebb33a758d78c3cffe9+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.143.0+e39249ad/MODULE_f9d7b269e731f2895949+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/Llama-3.1-70B-Instruct/7e003f9cde3ad2990b98.json

@@ -0,0 +1 @@
+{"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "eos_token_id": [128001, 128008, 128009], "head_dim": 128, "hidden_act": "silu", "hidden_size": 8192, "initializer_range": 0.02, "intermediate_size": 28672, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "llama", "neuron": {"auto_cast_type": "bf16", "batch_size": 4, "checkpoint_id": "meta-llama/Llama-3.1-70B-Instruct", "checkpoint_revision": "1605565b47bb9346c5515c34102e054115b4f98b", "compiler_type": "neuronx-cc", "compiler_version": "2.15.143.0+e39249ad", "num_cores": 24, "sequence_length": 4096, "task": "text-generation"}, "num_attention_heads": 64, "num_hidden_layers": 80, "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, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 128256}

neuronxcc-2.15.143.0+e39249ad/MODULE_cebb33a758d78c3cffe9+39f12043/compile_flags.json

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

neuronxcc-2.15.143.0+e39249ad/MODULE_cebb33a758d78c3cffe9+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b837fc2378fffecd859ba44115b41a9762865941d70d3065d0e414c758eaa73
+size 915725

neuronxcc-2.15.143.0+e39249ad/MODULE_cebb33a758d78c3cffe9+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d16def962638f05e0818dcfffd24396f004e5213485e3362df8fb93c01218c18
+size 10681344

neuronxcc-2.15.143.0+e39249ad/MODULE_f9d7b269e731f2895949+39f12043/compile_flags.json

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

neuronxcc-2.15.143.0+e39249ad/MODULE_f9d7b269e731f2895949+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6171269061b8aebfb07c36020904633f1e5397055e705c4479d7df51bfd55fc3
+size 1112352

neuronxcc-2.15.143.0+e39249ad/MODULE_f9d7b269e731f2895949+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de24fb5e5a50d1de834bff00f4267eeced3e884d7cbce16f02aa55517f060398
+size 2612224