Synchronizing local compiler cache.

.gitattributes

@@ -7165,3 +7165,5 @@ neuronxcc-2.16.372.0+4a9b2326/MODULE_324968d40a0692640f69/vae_decoder/model.neur
 neuronxcc-2.16.372.0+4a9b2326/MODULE_324968d40a0692640f69/vae_encoder/model.neuron filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.372.0+4a9b2326/MODULE_1b6d7576802d84bea017+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.372.0+4a9b2326/MODULE_c578025927b993537ac9+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_addf5b05055902bae4fd+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_cebb33a758d78c3cffe9+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/Llama-3.1-70B-Instruct/20b51ff77994bb4c1cc6.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": "2.16.372.0+4a9b2326", "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-2.16.372.0+4a9b2326/MODULE_addf5b05055902bae4fd+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_addf5b05055902bae4fd+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52628fbb5f9e37c1dde67921bb189a6ade9e7cc885637461bfddabaf53e43175
+size 1117152

neuronxcc-2.16.372.0+4a9b2326/MODULE_addf5b05055902bae4fd+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9374dcaad1bff389cb86d2ca568a9dd3a5230e125f459ce8a583d871f5b1a7ed
+size 2581504

neuronxcc-2.16.372.0+4a9b2326/MODULE_cebb33a758d78c3cffe9+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_cebb33a758d78c3cffe9+613edded/model.hlo_module.pb

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

neuronxcc-2.16.372.0+4a9b2326/MODULE_cebb33a758d78c3cffe9+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07b7a060b8292005e32e03793e99673f379b3f2e7082aae1885d39e2b35943e3
+size 10589184