Synchronizing local compiler cache.

neuronxcc-2.16.372.0+4a9b2326/0_REGISTRY/0.2.0.dev2/inference/granite/ibm-granite/granite-3.1-2b-instruct/921d5e10a55e2f6031f2.json

@@ -0,0 +1 @@
+{"_entry_class": "SingleModelCacheEntry", "_model_id": "ibm-granite/granite-3.1-2b-instruct", "_task": "text-generation", "architectures": ["GraniteForCausalLM"], "attention_bias": false, "attention_dropout": 0.1, "attention_multiplier": 0.015625, "embedding_multiplier": 12.0, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "intermediate_size": 8192, "logits_scaling": 8.0, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "granite", "neuron": {"all_reduce_dtype": null, "allow_flash_attention": true, "attention_layout": "HSB", "attn_output_transposed": false, "auto_cast_type": "bf16", "backend": "hlo", "batch_size": 4, "checkpoint_id": "ibm-granite/granite-3.1-2b-instruct", "checkpoint_revision": "374ef54e020a3ce208c65e96d6213922a87d8952", "collectives_layout": "HSB", "continuous_batching": true, "fuse_qkv": true, "group_query_attention": "shard-over-heads", "log_softmax_scores": false, "neuronxcc_version": "2.16.372.0+4a9b2326", "optimum_neuron_version": "0.2.0.dev2", "output_all_logits": false, "sequence_length": 4096, "tp_degree": 2}, "num_attention_heads": 32, "num_hidden_layers": 40, "num_key_value_heads": 8, "residual_multiplier": 0.22, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 5000000.0, "tie_word_embeddings": true, "use_cache": true, "vocab_size": 49155}