{ "producer": { "name": "modelopt", "version": "0.19.0" }, "architecture": "LlamaForCausalLM", "dtype": "bfloat16", "logits_dtype": "float16", "num_hidden_layers": 88, "num_attention_heads": 96, "num_key_value_heads": 8, "hidden_size": 12288, "norm_epsilon": 1e-05, "vocab_size": 32768, "max_position_embeddings": 131072, "hidden_act": "silu", "use_parallel_embedding": true, "embedding_sharding_dim": 0, "quantization": { "quant_algo": "W4A16_AWQ", "kv_cache_quant_algo": null, "group_size": 128, "has_zero_point": false, "pre_quant_scale": true, "exclude_modules": [ "transformer.layers.64.input_layernorm", "transformer.layers.62.post_layernorm", "transformer.layers.0.input_layernorm", "transformer.layers.73.post_layernorm", "transformer.ln_f", "transformer.layers.42.input_layernorm", "transformer.layers.37.input_layernorm", "transformer.layers.32.input_layernorm", "transformer.layers.10.post_layernorm", "transformer.layers.60.post_layernorm", "transformer.layers.52.input_layernorm", "transformer.layers.62.input_layernorm", "transformer.layers.70.post_layernorm", "transformer.layers.36.post_layernorm", "transformer.layers.53.post_layernorm", "transformer.layers.10.input_layernorm", "transformer.layers.26.post_layernorm", "transformer.layers.7.post_layernorm", "transformer.layers.15.post_layernorm", "transformer.layers.85.input_layernorm", "transformer.layers.9.post_layernorm", "transformer.layers.87.input_layernorm", "transformer.layers.19.post_layernorm", "transformer.layers.29.input_layernorm", "transformer.layers.65.input_layernorm", "transformer.layers.79.post_layernorm", "transformer.layers.5.input_layernorm", "transformer.layers.61.input_layernorm", "transformer.layers.4.input_layernorm", "transformer.layers.16.post_layernorm", "transformer.layers.32.post_layernorm", "transformer.layers.41.post_layernorm", "transformer.layers.55.post_layernorm", "transformer.layers.81.post_layernorm", "transformer.layers.81.input_layernorm", "transformer.layers.43.input_layernorm", "transformer.layers.31.input_layernorm", "transformer.layers.41.input_layernorm", "transformer.layers.7.input_layernorm", "transformer.layers.0.post_layernorm", "transformer.layers.67.post_layernorm", "transformer.layers.6.input_layernorm", "transformer.layers.38.post_layernorm", "transformer.layers.9.input_layernorm", "transformer.layers.23.input_layernorm", "transformer.layers.25.post_layernorm", "transformer.layers.66.post_layernorm", "transformer.layers.75.post_layernorm", "transformer.layers.84.input_layernorm", "transformer.layers.42.post_layernorm", "transformer.layers.11.input_layernorm", "transformer.layers.63.post_layernorm", "transformer.layers.65.post_layernorm", "transformer.layers.19.input_layernorm", "transformer.layers.44.post_layernorm", "transformer.layers.77.post_layernorm", "transformer.layers.18.post_layernorm", "transformer.layers.77.input_layernorm", "transformer.layers.82.input_layernorm", "transformer.layers.30.post_layernorm", "transformer.layers.45.input_layernorm", "transformer.layers.51.input_layernorm", "transformer.layers.47.input_layernorm", "transformer.layers.57.post_layernorm", "transformer.layers.20.post_layernorm", "transformer.layers.33.post_layernorm", "transformer.layers.33.input_layernorm", "transformer.layers.25.input_layernorm", "transformer.layers.12.input_layernorm", "transformer.layers.56.post_layernorm", "transformer.layers.20.input_layernorm", "transformer.layers.69.post_layernorm", "transformer.layers.35.input_layernorm", "transformer.layers.86.input_layernorm", "transformer.layers.15.input_layernorm", "transformer.layers.2.post_layernorm", "transformer.layers.14.input_layernorm", "transformer.layers.87.post_layernorm", "transformer.layers.17.input_layernorm", "transformer.layers.8.post_layernorm", "transformer.layers.27.input_layernorm", "transformer.layers.74.post_layernorm", "transformer.layers.1.input_layernorm", "transformer.layers.51.post_layernorm", "transformer.layers.23.post_layernorm", "transformer.layers.79.input_layernorm", "transformer.layers.63.input_layernorm", "transformer.layers.3.post_layernorm", "transformer.layers.18.input_layernorm", "transformer.layers.46.post_layernorm", "transformer.layers.40.post_layernorm", "transformer.layers.24.post_layernorm", "transformer.layers.54.input_layernorm", "transformer.layers.22.post_layernorm", "transformer.layers.26.input_layernorm", "transformer.layers.40.input_layernorm", "transformer.layers.76.input_layernorm", "transformer.layers.72.post_layernorm", "transformer.layers.83.input_layernorm", "transformer.layers.58.post_layernorm", "transformer.layers.28.post_layernorm", "transformer.layers.73.input_layernorm", "transformer.layers.45.post_layernorm", "transformer.layers.24.input_layernorm", "transformer.layers.56.input_layernorm", "transformer.layers.11.post_layernorm", "transformer.layers.30.input_layernorm", "transformer.layers.71.input_layernorm", "transformer.layers.22.input_layernorm", "transformer.layers.13.post_layernorm", "transformer.layers.74.input_layernorm", "transformer.layers.14.post_layernorm", "transformer.layers.86.post_layernorm", "transformer.layers.6.post_layernorm", "transformer.layers.68.input_layernorm", "transformer.layers.16.input_layernorm", "transformer.layers.37.post_layernorm", "transformer.layers.36.input_layernorm", "transformer.vocab_embedding", "transformer.layers.50.input_layernorm", "transformer.layers.49.input_layernorm", "transformer.layers.34.post_layernorm", "transformer.layers.66.input_layernorm", "transformer.layers.61.post_layernorm", "transformer.layers.2.input_layernorm", "transformer.layers.85.post_layernorm", "transformer.layers.67.input_layernorm", "transformer.layers.83.post_layernorm", "transformer.layers.52.post_layernorm", "transformer.layers.17.post_layernorm", "transformer.layers.76.post_layernorm", "lm_head", "transformer.layers.54.post_layernorm", "transformer.layers.48.input_layernorm", "transformer.layers.38.input_layernorm", "transformer.layers.50.post_layernorm", "transformer.layers.53.input_layernorm", "transformer.layers.60.input_layernorm", "transformer.layers.72.input_layernorm", "transformer.layers.39.post_layernorm", "transformer.layers.28.input_layernorm", "transformer.layers.48.post_layernorm", "transformer.layers.5.post_layernorm", "transformer.layers.43.post_layernorm", "transformer.layers.46.input_layernorm", "transformer.layers.69.input_layernorm", "transformer.layers.75.input_layernorm", "transformer.layers.34.input_layernorm", "transformer.layers.57.input_layernorm", "transformer.layers.12.post_layernorm", "transformer.layers.44.input_layernorm", "transformer.layers.82.post_layernorm", "transformer.layers.71.post_layernorm", "transformer.layers.4.post_layernorm", "transformer.layers.84.post_layernorm", "transformer.layers.78.input_layernorm", "transformer.layers.21.input_layernorm", "transformer.layers.80.input_layernorm", "transformer.layers.3.input_layernorm", "transformer.layers.31.post_layernorm", "transformer.layers.49.post_layernorm", "transformer.layers.59.input_layernorm", "transformer.layers.29.post_layernorm", "transformer.layers.58.input_layernorm", "transformer.layers.27.post_layernorm", "transformer.layers.21.post_layernorm", "transformer.layers.1.post_layernorm", "transformer.layers.55.input_layernorm", "transformer.layers.64.post_layernorm", "transformer.layers.80.post_layernorm", "transformer.layers.47.post_layernorm", "transformer.layers.70.input_layernorm", "transformer.layers.68.post_layernorm", "transformer.layers.8.input_layernorm", "transformer.layers.39.input_layernorm", "transformer.layers.59.post_layernorm", "transformer.layers.78.post_layernorm", "transformer.layers.35.post_layernorm", "transformer.layers.13.input_layernorm" ] }, "mapping": { "world_size": 4, "tp_size": 4, "pp_size": 1 }, "head_size": 128, "intermediate_size": 28672, "position_embedding_type": "rope_gpt_neox", "share_embedding_table": false, "residual_mlp": false, "bias": false, "rotary_pct": 1.0, "rank": 3, "decoder": "llama", "rmsnorm": true, "lm_head_bias": false, "rotary_base": 1000000.0, "model_type": "llama" }