tensorRT-LLM test

config.json

@@ -0,0 +1,57 @@
+{
+  "builder_config": {
+    "gather_context_logits": false,
+    "gather_generation_logits": false,
+    "hidden_act": "silu",
+    "hidden_size": 4096,
+    "int8": false,
+    "max_batch_size": 2,
+    "max_beam_width": 1,
+    "max_input_len": 2048,
+    "max_num_tokens": 4096,
+    "max_output_len": 2048,
+    "max_position_embeddings": 32768,
+    "max_prompt_embedding_table_size": 0,
+    "mlp_hidden_size": 22016,
+    "name": "qwen",
+    "num_heads": 32,
+    "num_kv_heads": 32,
+    "num_layers": 32,
+    "parallel_build": false,
+    "pipeline_parallel": 1,
+    "precision": "float16",
+    "quant_mode": 33,
+    "strongly_typed": false,
+    "tensor_parallel": 1,
+    "use_refit": false,
+    "vocab_size": 151936
+  },
+  "plugin_config": {
+    "attention_qk_half_accumulation": false,
+    "bert_attention_plugin": null,
+    "context_fmha": true,
+    "context_fmha_fp32_acc": false,
+    "enable_xqa": true,
+    "gemm_plugin": "float16",
+    "gpt_attention_plugin": "float16",
+    "identity_plugin": null,
+    "layernorm_quantization_plugin": null,
+    "lookup_plugin": null,
+    "lora_plugin": null,
+    "moe_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": "float16",
+    "weight_only_quant_matmul_plugin": null
+  }
+}