optimum-neuron-cache
/
neuronxcc-2.12.68.0+4480452af
/0_REGISTRY
/0.0.20
/inference
/llama
/abacusai
/Smaug-72B-v0.1
/3ebaece7bc9fad14e30e.json
{"vocab_size": 152064, "max_position_embeddings": 32768, "hidden_size": 8192, "intermediate_size": 24576, "num_hidden_layers": 80, "num_attention_heads": 64, "num_key_value_heads": 64, "hidden_act": "silu", "initializer_range": 0.02, "rms_norm_eps": 1e-06, "pretraining_tp": 1, "use_cache": true, "rope_theta": 1000000, "rope_scaling": null, "attention_bias": true, "attention_dropout": 0.0, "torch_dtype": "bfloat16", "tie_word_embeddings": false, "architectures": ["LlamaForCausalLM"], "bos_token_id": 1, "pad_token_id": 151643, "eos_token_id": 151643, "model_type": "llama", "seq_length": 32768, "neuron": {"task": "text-generation", "batch_size": 4, "num_cores": 24, "auto_cast_type": "fp16", "sequence_length": 4096, "compiler_type": "neuronx-cc", "compiler_version": "2.12.68.0+4480452af", "checkpoint_id": "abacusai/Smaug-72B-v0.1", "checkpoint_revision": "a1d657156f82c24b670158406378648233487011"}} |