manifest nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 @ B200 (07c59cc4d89b61bb)
Browse files
nvidia__NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/B200/tp8-seq131072-lora64x1/07c59cc4d89b61bb/manifest.json
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16",
|
| 3 |
+
"gpu_type": "B200",
|
| 4 |
+
"tensor_parallel_size": 8,
|
| 5 |
+
"max_seq_length": 131072,
|
| 6 |
+
"enable_lora": true,
|
| 7 |
+
"max_lora_rank": 64,
|
| 8 |
+
"max_loras": 1,
|
| 9 |
+
"cudagraph_capture_sizes": [
|
| 10 |
+
1,
|
| 11 |
+
2,
|
| 12 |
+
4,
|
| 13 |
+
8,
|
| 14 |
+
16,
|
| 15 |
+
32,
|
| 16 |
+
64,
|
| 17 |
+
128,
|
| 18 |
+
192,
|
| 19 |
+
256,
|
| 20 |
+
384,
|
| 21 |
+
512
|
| 22 |
+
],
|
| 23 |
+
"use_mega_aot_artifact": true,
|
| 24 |
+
"deep_gemm_warmup": "skip",
|
| 25 |
+
"enable_prefix_caching": true,
|
| 26 |
+
"vllm_version": "0.22.0",
|
| 27 |
+
"torch_version": "2.11.0+cu129",
|
| 28 |
+
"torch": "2.11.0+cu129",
|
| 29 |
+
"torch_cuda": "12.9",
|
| 30 |
+
"vllm": "0.22.0",
|
| 31 |
+
"image_tag": "baseten/baseten-weight-sync-inference:main-15e6be27",
|
| 32 |
+
"caller": "github-actions:github-actions[bot]",
|
| 33 |
+
"model_revision": "624ba927cfbef0427354998700de3d51173c8c04",
|
| 34 |
+
"build_id": "28193124938-1",
|
| 35 |
+
"opaque_sampler_payload": {
|
| 36 |
+
"tensor_parallel_size": 8,
|
| 37 |
+
"max_seq_length": 131072,
|
| 38 |
+
"enable_lora": true,
|
| 39 |
+
"max_lora_rank": 64,
|
| 40 |
+
"max_loras": 1,
|
| 41 |
+
"cudagraph_capture_sizes": [
|
| 42 |
+
1,
|
| 43 |
+
2,
|
| 44 |
+
4,
|
| 45 |
+
8,
|
| 46 |
+
16,
|
| 47 |
+
32,
|
| 48 |
+
64,
|
| 49 |
+
128,
|
| 50 |
+
192,
|
| 51 |
+
256,
|
| 52 |
+
384,
|
| 53 |
+
512
|
| 54 |
+
],
|
| 55 |
+
"use_mega_aot_artifact": true,
|
| 56 |
+
"deep_gemm_warmup": "skip",
|
| 57 |
+
"enable_prefix_caching": true,
|
| 58 |
+
"load_format": "fastsafetensors",
|
| 59 |
+
"disable_custom_all_reduce": true
|
| 60 |
+
},
|
| 61 |
+
"build_profile": {
|
| 62 |
+
"cudagraph_capture_sizes": [
|
| 63 |
+
1,
|
| 64 |
+
2,
|
| 65 |
+
4,
|
| 66 |
+
8,
|
| 67 |
+
16,
|
| 68 |
+
32,
|
| 69 |
+
64,
|
| 70 |
+
128,
|
| 71 |
+
192,
|
| 72 |
+
256,
|
| 73 |
+
384,
|
| 74 |
+
512
|
| 75 |
+
],
|
| 76 |
+
"use_mega_aot_artifact": true,
|
| 77 |
+
"deep_gemm_warmup": "skip"
|
| 78 |
+
},
|
| 79 |
+
"ready_in_seconds_no_cache": 522.1,
|
| 80 |
+
"kv_cache_max_tokens": 366142,
|
| 81 |
+
"kv_cache_max_concurrency": 2.7934426229508196,
|
| 82 |
+
"kv_cache_gpu_memory_utilization": 0.92,
|
| 83 |
+
"cache_size_uncompressed_bytes": 1087150080,
|
| 84 |
+
"cache_size_compressed_bytes": 74665461,
|
| 85 |
+
"compression": "zstd -9 -T0 (multithreaded)",
|
| 86 |
+
"compress_time_seconds": 2.15,
|
| 87 |
+
"upload_time_seconds": 2.19
|
| 88 |
+
}
|