Synchronizing local compiler cache.

.gitattributes

@@ -7464,3 +7464,11 @@ neuronxcc-2.15.143.0+e39249ad/MODULE_1695330137f296fdba60+39f12043/model.neff fi
 neuronxcc-2.15.143.0+e39249ad/MODULE_807378dcfffa44901b59+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.143.0+e39249ad/MODULE_9d7ecdc9231b4ba0c0d8+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
 neuronxcc-2.15.143.0+e39249ad/MODULE_b15e1b7eb2bfb8f6dc3e+39f12043/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_38847bfec695e36ba44b+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_48c6444adcac1b53fda6+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_52108672c0a9ba2de711+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_5b45c246024901dcd434+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_9e3cdf817e494ba86fd4+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_b2aaa317ae645625c9d1+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_d07d2aeb772cbb9c0ccd+613edded/model.neff filter=lfs diff=lfs merge=lfs -text
+neuronxcc-2.16.372.0+4a9b2326/MODULE_ef26a928b1a6003d9b67+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/Meta-Llama-3.1-8B-Instruct/3b6b0065e24ca9d2eeb0.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": 4096, "initializer_range": 0.02, "intermediate_size": 14336, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "llama", "neuron": {"auto_cast_type": "bf16", "batch_size": 1, "checkpoint_id": "meta-llama/Meta-Llama-3.1-8B-Instruct", "checkpoint_revision": "0e9e39f249a16976918f6564b8830bc894c89659", "compiler_type": "neuronx-cc", "compiler_version": "2.16.372.0+4a9b2326", "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": {"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+4a9b2326/0_REGISTRY/0.1.0.dev1/inference/phi3/microsoft/Phi-3-medium-4k-instruct/ae83aa0fea9a9d0315d5.json

@@ -0,0 +1 @@
+{"architectures": ["Phi3ForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "auto_map": {"AutoConfig": "microsoft/Phi-3-medium-4k-instruct--configuration_phi3.Phi3Config", "AutoModelForCausalLM": "microsoft/Phi-3-medium-4k-instruct--modeling_phi3.Phi3ForCausalLM"}, "bos_token_id": 1, "embd_pdrop": 0.0, "eos_token_id": 32000, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 4096, "model_type": "phi3", "neuron": {"auto_cast_type": "bf16", "batch_size": 4, "checkpoint_id": "microsoft/Phi-3-medium-4k-instruct", "checkpoint_revision": "ae004ae82eb6eddc32906dfacb1d6dfea8f91996", "compiler_type": "neuronx-cc", "compiler_version": "2.16.372.0+4a9b2326", "num_cores": 8, "sequence_length": 4096, "task": "text-generation"}, "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 4096, "pad_token_id": 32000, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 10000.0, "sliding_window": 2047, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 32064}

neuronxcc-2.16.372.0+4a9b2326/0_REGISTRY/0.1.0.dev1/inference/phi3/microsoft/Phi-3-mini-4k-instruct/9a957125ff99f484a66e.json

@@ -0,0 +1 @@
+{"architectures": ["Phi3ForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "auto_map": {"AutoConfig": "microsoft/Phi-3-mini-4k-instruct--configuration_phi3.Phi3Config", "AutoModelForCausalLM": "microsoft/Phi-3-mini-4k-instruct--modeling_phi3.Phi3ForCausalLM"}, "bos_token_id": 1, "embd_pdrop": 0.0, "eos_token_id": 32000, "hidden_act": "silu", "hidden_size": 3072, "initializer_range": 0.02, "intermediate_size": 8192, "max_position_embeddings": 4096, "model_type": "phi3", "neuron": {"auto_cast_type": "bf16", "batch_size": 4, "checkpoint_id": "microsoft/Phi-3-mini-4k-instruct", "checkpoint_revision": "0a67737cc96d2554230f90338b163bc6380a2a85", "compiler_type": "neuronx-cc", "compiler_version": "2.16.372.0+4a9b2326", "num_cores": 8, "sequence_length": 4096, "task": "text-generation"}, "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 32, "original_max_position_embeddings": 4096, "pad_token_id": 32000, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 10000.0, "sliding_window": 2047, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 32064}

neuronxcc-2.16.372.0+4a9b2326/0_REGISTRY/0.1.0.dev1/inference/phi3/microsoft/Phi-4-mini-instruct/6ff6f9ae27117fd24bd7.json

@@ -0,0 +1 @@
+{"architectures": ["Phi3ForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "auto_map": {"AutoConfig": "microsoft/Phi-4-mini-instruct--configuration_phi3.Phi3Config", "AutoModelForCausalLM": "microsoft/Phi-4-mini-instruct--modeling_phi3.Phi3ForCausalLM", "AutoTokenizer": "microsoft/Phi-4-mini-instruct--Xenova/gpt-4o"}, "bos_token_id": 199999, "embd_pdrop": 0.0, "eos_token_id": 199999, "full_attn_mod": 1, "hidden_act": "silu", "hidden_size": 3072, "initializer_range": 0.02, "intermediate_size": 8192, "interpolate_factor": 1, "lm_head_bias": false, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "phi3", "neuron": {"auto_cast_type": "bf16", "batch_size": 4, "checkpoint_id": "microsoft/Phi-4-mini-instruct", "checkpoint_revision": "4b00ec8714b0cb224e4fb33380cbf0919f177f3e", "compiler_type": "neuronx-cc", "compiler_version": "2.16.372.0+4a9b2326", "num_cores": 8, "sequence_length": 4096, "task": "text-generation"}, "num_attention_heads": 24, "num_hidden_layers": 32, "num_key_value_heads": 8, "original_max_position_embeddings": 4096, "pad_token_id": 199999, "partial_rotary_factor": 0.75, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": {"long_factor": [1, 1.118320672, 1.250641126, 1.398617824, 1.564103225, 1.74916897, 1.956131817, 2.187582649, 2.446418898, 2.735880826, 3.059592084, 3.421605075, 3.826451687, 4.279200023, 4.785517845, 5.351743533, 5.984965424, 6.693110555, 7.485043894, 8.370679318, 9.36110372, 10.4687158, 11.70738129, 13.09260651, 14.64173252, 16.37415215, 18.31155283, 20.47818807, 22.90118105, 25.61086418, 28.64115884, 32.03, 32.1, 32.13, 32.23, 32.6, 32.61, 32.64, 32.66, 32.7, 32.71, 32.93, 32.97, 33.28, 33.49, 33.5, 44.16, 47.77], "short_factor": [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0], "type": "longrope"}, "rope_theta": 10000.0, "sliding_window": 262144, "tie_word_embeddings": true, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 200064}

neuronxcc-2.16.372.0+4a9b2326/0_REGISTRY/0.1.0.dev1/inference/phi3/microsoft/phi-4/42644c76f78cfd9108b4.json

@@ -0,0 +1 @@
+{"architectures": ["Phi3ForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "neuron": {"auto_cast_type": "bf16", "batch_size": 1, "checkpoint_id": "microsoft/phi-4", "checkpoint_revision": "187ef0342fff0eb3333be9f00389385e95ef0b61", "compiler_type": "neuronx-cc", "compiler_version": "2.16.372.0+4a9b2326", "num_cores": 2, "sequence_length": 4096, "task": "text-generation"}, "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 100352}

neuronxcc-2.16.372.0+4a9b2326/0_REGISTRY/0.1.0.dev1/inference/phi3/microsoft/phi-4/c488c1faf1a22992a2ea.json

@@ -0,0 +1 @@
+{"architectures": ["Phi3ForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "neuron": {"auto_cast_type": "bf16", "batch_size": 16, "checkpoint_id": "microsoft/phi-4", "checkpoint_revision": "187ef0342fff0eb3333be9f00389385e95ef0b61", "compiler_type": "neuronx-cc", "compiler_version": "2.16.372.0+4a9b2326", "num_cores": 8, "sequence_length": 4096, "task": "text-generation"}, "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "use_cache": true, "vocab_size": 100352}

neuronxcc-2.16.372.0+4a9b2326/MODULE_38847bfec695e36ba44b+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_38847bfec695e36ba44b+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71e610135421e7ea92c7bbe52ade46edd0eceeaecc7860e8b4389c63f135be8e
+size 446036

neuronxcc-2.16.372.0+4a9b2326/MODULE_38847bfec695e36ba44b+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83478fbc4b65be0f3f4d4b017c0bdd9516c24667d5e349907d97aa1bcf0b3832
+size 18361344

neuronxcc-2.16.372.0+4a9b2326/MODULE_48c6444adcac1b53fda6+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_48c6444adcac1b53fda6+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da33932abafecde481f769f8c365b4389cf32ef68dd270cfeda7fa2ebe556543
+size 331676

neuronxcc-2.16.372.0+4a9b2326/MODULE_48c6444adcac1b53fda6+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:590e828f6406bf328681eed814557293021f4720943d4817c81b9bb27fff7ec6
+size 2417664

neuronxcc-2.16.372.0+4a9b2326/MODULE_52108672c0a9ba2de711+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_52108672c0a9ba2de711+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cee3003becf81e54b8e394e4620975900d88d65f791a6e08a23126ad3e6403e
+size 424232

neuronxcc-2.16.372.0+4a9b2326/MODULE_52108672c0a9ba2de711+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4348492ffc10e831699a14c9e8ff79b263922714421b3bced3dd19a2cc81afd9
+size 6861824

neuronxcc-2.16.372.0+4a9b2326/MODULE_5b45c246024901dcd434+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_5b45c246024901dcd434+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2906b8f8b2a663a093a64a55311c17b4ab5fe54e35680905c95cad882a69fb47
+size 432521

neuronxcc-2.16.372.0+4a9b2326/MODULE_5b45c246024901dcd434+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f373da3484942dfe188fe5fad810d7cc2f263829b3e881ac3b5d209f09cbe71b
+size 21914624

neuronxcc-2.16.372.0+4a9b2326/MODULE_9e3cdf817e494ba86fd4+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_9e3cdf817e494ba86fd4+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2a4208b0c41de4690c470e5450910fb06c7f9d839f21d65d0d18aae9b60acd
+size 542459

neuronxcc-2.16.372.0+4a9b2326/MODULE_9e3cdf817e494ba86fd4+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce1ae47347f1ae17be5e4b1574462f207c5d07243b221b6be19a3a63c7601fb3
+size 3257344

neuronxcc-2.16.372.0+4a9b2326/MODULE_b2aaa317ae645625c9d1+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_b2aaa317ae645625c9d1+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9288d7b117745da497a6fa121d6fa620fb1461a4de05d0835fb9afb7b719512c
+size 338446

neuronxcc-2.16.372.0+4a9b2326/MODULE_b2aaa317ae645625c9d1+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:945244c03a7a599ca4bb8a47d157523a41578c6646a53e76de388c2971b7dfc7
+size 2786304

neuronxcc-2.16.372.0+4a9b2326/MODULE_d07d2aeb772cbb9c0ccd+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_d07d2aeb772cbb9c0ccd+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1227f49f0cb61a4bb3a55a87e763f4fc8fce66a3c84aec9053b834d3c9036116
+size 400621

neuronxcc-2.16.372.0+4a9b2326/MODULE_d07d2aeb772cbb9c0ccd+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:217146d508819886b2dac35fa8078e80982b60326feb65dff13f6d9db64045dc
+size 28959744

neuronxcc-2.16.372.0+4a9b2326/MODULE_ef26a928b1a6003d9b67+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_ef26a928b1a6003d9b67+613edded/model.hlo_module.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:897db48b7e2fc00a32f66b41f0dea0789f5f65c7dda5a8a50c3266a9df00fead
+size 363255

neuronxcc-2.16.372.0+4a9b2326/MODULE_ef26a928b1a6003d9b67+613edded/model.neff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0eae8a8e6f93fafba664dfc04235a982dd59f8052a5267e7fb1166f0607fc77d
+size 1936384