Synchronizing local compiler cache.

.gitattributes

@@ -4466,3 +4466,5 @@ neuronxcc-2.15.128.0+56dc5a86/MODULE_5921fe7851e4324b0368+39f12043/model.neff fi
 neuronxcc-2.15.128.0+56dc5a86/MODULE_a908ab12c136ae19ce8b+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.128.0+56dc5a86/MODULE_314799b42bb1f45cec02+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.128.0+56dc5a86/MODULE_3e25032356233fc84aee+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.128.0+56dc5a86/MODULE_0050ca426ceec1f9aac8+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.128.0+56dc5a86/MODULE_c9fba0f83c6bc7c8d897+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text

neuronxcc-2.15.128.0+56dc5a86/0_REGISTRY/0.0.25.dev0/inference/llama/meta-llama/Meta-Llama-3-8B/b10b93b32a09e72c6bdc.json

@@ -0,0 +1 @@
+{"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "eos_token_id": 128001, "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": "meta-llama/Meta-Llama-3-8B", "checkpoint_revision": "62bd457b6fe961a42a631306577e622c83876cb6", "compiler_type": "neuronx-cc", "compiler_version": "2.15.128.0+56dc5a86", "num_cores": 8, "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-2.15.128.0+56dc5a86/MODULE_0050ca426ceec1f9aac8+39f12043/compile_flags.json

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

neuronxcc-2.15.128.0+56dc5a86/MODULE_0050ca426ceec1f9aac8+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f75e6264d9c48b0591cfc7ac719215a7da0a4c3d611c7b2aab52476207a8e8a
+size 460157

neuronxcc-2.15.128.0+56dc5a86/MODULE_0050ca426ceec1f9aac8+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:423401255bc1834fb335d6341ac113424cac7bfb3917665d86776fc1a43849ae
+size 30915584

neuronxcc-2.15.128.0+56dc5a86/MODULE_c9fba0f83c6bc7c8d897+39f12043/compile_flags.json

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

neuronxcc-2.15.128.0+56dc5a86/MODULE_c9fba0f83c6bc7c8d897+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63c42a4837553d7790bedb0d502cf744bff0ee95085c2b717227d5527dbbaa64
+size 373541

neuronxcc-2.15.128.0+56dc5a86/MODULE_c9fba0f83c6bc7c8d897+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d342d48bc0e5730284c18a5a3c474dd8985b2e3bc687b768b0ee8d2c56503846
+size 3165184