manifest Qwen/Qwen3.5-4B @ H100 (45504f3eb15dac72)

Qwen__Qwen3.5-4B/H100/tp1-seq131072-lora64x4/45504f3eb15dac72/manifest.json

@@ -0,0 +1,95 @@
+{
+  "model": "Qwen/Qwen3.5-4B",
+  "gpu_type": "H100",
+  "tensor_parallel_size": 1,
+  "max_seq_length": 131072,
+  "enable_lora": true,
+  "max_lora_rank": 64,
+  "max_loras": 4,
+  "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": "0.22.0",
+  "torch_version": "2.11.0+cu129",
+  "torch": "2.11.0+cu129",
+  "torch_cuda": "12.9",
+  "vllm": "0.22.0",
+  "image_tag": "baseten/baseten-weight-sync-inference:main-09f8dc2a",
+  "caller": "github-actions:William-Gao1",
+  "model_revision": "851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a",
+  "build_id": "27358404720-1",
+  "opaque_sampler_payload": {
+    "tensor_parallel_size": 1,
+    "max_seq_length": 131072,
+    "enable_lora": true,
+    "max_lora_rank": 64,
+    "max_loras": 4,
+    "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
+  },
+  "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": 304.09,
+  "cache_size_uncompressed_bytes": 292802560,
+  "cache_size_compressed_bytes": 18480365,
+  "compression": "zstd -9 -T0 (multithreaded)",
+  "compress_time_seconds": 0.7,
+  "upload_time_seconds": 4.24
+}