Synchronizing local compiler cache.

.gitattributes

@@ -4490,3 +4490,5 @@ neuronxcc-2.15.128.0+56dc5a86/MODULE_3af38c884595905e2da7+39f12043/model.neff fi
 neuronxcc-2.15.128.0+56dc5a86/MODULE_a5aa835f854ee92301d0+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.128.0+56dc5a86/MODULE_22cb7b0e8a1f9fc5ab88+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.128.0+56dc5a86/MODULE_9850d48f308811a3bf7c+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.128.0+56dc5a86/MODULE_7f99a181290326a67234+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.15.128.0+56dc5a86/MODULE_d35838e9962cd0fda4dd+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/Llama-2-7b-chat-hf/6e49b149885a41bdd543.json

@@ -0,0 +1 @@
+{"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 11008, "max_position_embeddings": 4096, "mlp_bias": false, "model_type": "llama", "neuron": {"auto_cast_type": "fp16", "batch_size": 16, "checkpoint_id": "meta-llama/Llama-2-7b-chat-hf", "checkpoint_revision": "f5db02db724555f92da89c216ac04704f23d4590", "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": 32, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 10000.0, "tie_word_embeddings": false, "torch_dtype": "float16", "use_cache": true, "vocab_size": 32000}

neuronxcc-2.15.128.0+56dc5a86/MODULE_7f99a181290326a67234+39f12043/compile_flags.json

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

neuronxcc-2.15.128.0+56dc5a86/MODULE_7f99a181290326a67234+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c012d63630517a942e8ffcb3ffa6e85ffc35e92f42e0259bc20320d6279b0bc0
+size 448317

neuronxcc-2.15.128.0+56dc5a86/MODULE_7f99a181290326a67234+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:303f139c968a7b67e93abdc36c64b532d2e0ff89eb8841e5023329608212a89a
+size 30198784

neuronxcc-2.15.128.0+56dc5a86/MODULE_d35838e9962cd0fda4dd+39f12043/compile_flags.json

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

neuronxcc-2.15.128.0+56dc5a86/MODULE_d35838e9962cd0fda4dd+39f12043/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bce6badb3599436b6ccf1d5521e571e89dcf15129891c1a7b92afeed0dc0111f
+size 349797

neuronxcc-2.15.128.0+56dc5a86/MODULE_d35838e9962cd0fda4dd+39f12043/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:766dde95c4819bc1d4dfff45f6b005b7baf95a83f7db943fea6ab2e535d29863
+size 7588864