Synchronizing local compiler cache.

.gitattributes

@@ -6486,3 +6486,5 @@ neuronxcc-2.16.303.0+d9f03cda/MODULE_b988d06a565871cc1943+7ac4dbae/model.neff fi
 neuronxcc-2.16.303.0+d9f03cda/MODULE_d0a7a042e0008322e298+7ac4dbae/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.303.0+d9f03cda/MODULE_efc674ceb57a1f7010a5+7ac4dbae/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.16.303.0+d9f03cda/MODULE_ff0a9145bffbede97cde+7ac4dbae/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.303.0+d9f03cda/MODULE_5e72d9266a7fb87f15cb+7ac4dbae/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.303.0+d9f03cda/MODULE_97b24281cd74002e4729+7ac4dbae/model.neff filter=lfs diff=lfs merge=lfs -text

neuronxcc-2.16.303.0+d9f03cda/0_REGISTRY/0.0.28.dev0/inference/llama/meta-llama/Llama-3.1-70B-Instruct/8382ce744316c5e96168.json

@@ -0,0 +1 @@
+{"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "eos_token_id": [128001, 128008, 128009], "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": 16, "checkpoint_id": "meta-llama/Llama-3.1-70B-Instruct", "checkpoint_revision": "1605565b47bb9346c5515c34102e054115b4f98b", "compiler_type": "neuronx-cc", "compiler_version": "2.16.303.0+d9f03cda", "num_cores": 64, "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-2.16.303.0+d9f03cda/MODULE_5e72d9266a7fb87f15cb+7ac4dbae/compile_flags.json

@@ -0,0 +1 @@
+["--target=trn2", "--model-type=transformer", "--auto-cast=none", "--execute-repetition=1"]

neuronxcc-2.16.303.0+d9f03cda/MODULE_5e72d9266a7fb87f15cb+7ac4dbae/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4b3f0035d2496dcba327bc0e0d6f5028e4dcda899afe4f8bb578eeefc021cb3
+size 861177

neuronxcc-2.16.303.0+d9f03cda/MODULE_5e72d9266a7fb87f15cb+7ac4dbae/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:30a60485ef7d317a8ee8cd72cdb9f1a2a786d7b335f2030e4b337d59ce8ce891
+size 11930624

neuronxcc-2.16.303.0+d9f03cda/MODULE_97b24281cd74002e4729+7ac4dbae/compile_flags.json

@@ -0,0 +1 @@
+["--target=trn2", "--model-type=transformer", "--auto-cast=none", "--execute-repetition=1"]

neuronxcc-2.16.303.0+d9f03cda/MODULE_97b24281cd74002e4729+7ac4dbae/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d240e09e566045d57bcfcd7e7fdc26de68106e2bc08bec016f3272743873010a
+size 1093802

neuronxcc-2.16.303.0+d9f03cda/MODULE_97b24281cd74002e4729+7ac4dbae/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a64aa8818e3a5ad1eb4f6e0dfb576b1741031a26583b92026ce2011846adf5bb
+size 1864704