| { |
| "builder_config": { |
| "apply_query_key_layer_scaling": false, |
| "cross_attention": false, |
| "gather_context_logits": false, |
| "gather_generation_logits": false, |
| "gpus_per_node": 8, |
| "has_position_embedding": false, |
| "has_token_type_embedding": false, |
| "head_size": 64, |
| "hidden_act": "gelu_new", |
| "hidden_size": 2048, |
| "int8": false, |
| "max_batch_size": 8, |
| "max_beam_width": 1, |
| "max_decoder_input_len": 1, |
| "max_encoder_input_len": 1024, |
| "max_output_len": 200, |
| "max_position_embeddings": 512, |
| "max_prompt_embedding_table_size": 0, |
| "name": "flan-t5-xl", |
| "num_heads": 32, |
| "num_layers": 24, |
| "parallel_build": false, |
| "pipeline_parallel": 1, |
| "precision": "float32", |
| "strongly_typed": false, |
| "tensor_parallel": 1, |
| "use_refit": false, |
| "vocab_size": 32128 |
| }, |
| "plugin_config": { |
| "attention_qk_half_accumulation": false, |
| "bert_attention_plugin": "float32", |
| "context_fmha": false, |
| "context_fmha_fp32_acc": false, |
| "enable_xqa": false, |
| "gemm_plugin": "float32", |
| "gpt_attention_plugin": "float32", |
| "identity_plugin": null, |
| "layernorm_quantization_plugin": null, |
| "lookup_plugin": null, |
| "lora_plugin": null, |
| "multi_block_mode": false, |
| "nccl_plugin": null, |
| "paged_kv_cache": false, |
| "quantize_per_token_plugin": false, |
| "quantize_tensor_plugin": false, |
| "remove_input_padding": true, |
| "rmsnorm_quantization_plugin": null, |
| "smooth_quant_gemm_plugin": null, |
| "tokens_per_block": 128, |
| "use_context_fmha_for_generation": false, |
| "use_custom_all_reduce": false, |
| "use_paged_context_fmha": false, |
| "weight_only_groupwise_quant_matmul_plugin": null, |
| "weight_only_quant_matmul_plugin": null |
| } |
| } |
|
|
