manifest nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 @ B300 (3f8d640e4155f214)

nvidia__NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4/B300/tp4-seq262144-lora64x4/3f8d640e4155f214/manifest.json

@@ -0,0 +1,106 @@
+{
+  "model": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4",
+  "gpu_type": "B300",
+  "tensor_parallel_size": 4,
+  "max_seq_length": 262144,
+  "enable_lora": true,
+  "max_lora_rank": 64,
+  "max_loras": 4,
+  "lora_target_modules": [],
+  "lora_target_module_preset": null,
+  "weights_source": null,
+  "cudagraph_capture_sizes": [
+    1,
+    2,
+    4,
+    8,
+    16,
+    32,
+    64,
+    128,
+    192,
+    256,
+    384,
+    512,
+    640,
+    768,
+    896,
+    1000
+  ],
+  "use_mega_aot_artifact": true,
+  "deep_gemm_warmup": "skip",
+  "enable_prefix_caching": true,
+  "vllm_version": "0.25.1",
+  "torch_version": "2.11.0+cu130",
+  "torch": "2.11.0+cu130",
+  "torch_cuda": "13.0",
+  "vllm": "0.25.1",
+  "image_tag": "baseten/baseten-weight-sync-inference:will-cu130-sampler-image-3fb7b93-cu130",
+  "caller": "github-actions:William-Gao1",
+  "model_revision": "183968f87ae4cedce3039313cac1fd43d112c578",
+  "build_id": "29977831034-1",
+  "opaque_sampler_payload": {
+    "tensor_parallel_size": 4,
+    "max_seq_length": 262144,
+    "enable_lora": true,
+    "max_lora_rank": 64,
+    "max_loras": 4,
+    "lora_target_modules": [],
+    "lora_target_module_preset": null,
+    "cudagraph_capture_sizes": [
+      1,
+      2,
+      4,
+      8,
+      16,
+      32,
+      64,
+      128,
+      192,
+      256,
+      384,
+      512,
+      640,
+      768,
+      896,
+      1000
+    ],
+    "use_mega_aot_artifact": true,
+    "deep_gemm_warmup": "skip",
+    "enable_prefix_caching": true,
+    "load_format": "fastsafetensors",
+    "disable_custom_all_reduce": true,
+    "gpu_memory_utilization": 0.88
+  },
+  "build_profile": {
+    "cudagraph_capture_sizes": [
+      1,
+      2,
+      4,
+      8,
+      16,
+      32,
+      64,
+      128,
+      192,
+      256,
+      384,
+      512,
+      640,
+      768,
+      896,
+      1000
+    ],
+    "use_mega_aot_artifact": true,
+    "deep_gemm_warmup": "skip"
+  },
+  "ready_in_seconds_no_cache": 1482.33,
+  "kv_cache_max_tokens": 10144972,
+  "kv_cache_max_concurrency": 38.7,
+  "kv_cache_gpu_memory_utilization": 0.88,
+  "cache_size_uncompressed_bytes": 502067200,
+  "cache_size_compressed_bytes": 31470325,
+  "compression": "zstd -9 -T0 (multithreaded)",
+  "compress_time_seconds": 0.83,
+  "upload_time_seconds": 8.97
+}