Skip to content

Actions: kacper-daftcode/vLLM-Moet

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
86 workflow runs
86 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
DeepSeek-V4.1 sm120: the mHC sublayer boundary off the decode critica… bench-lint #99: Commit dd29d12 pushed by kacper-daftcode
15s main
main
15s
DeepSeek-V4.1 sm120: the next three fusions measured - one restores a… bench-lint #98: Commit 205930e pushed by kacper-daftcode
22s main
main
22s
DeepSeek-V4.1 sm120: the MoE input quantization fused with the DeepGE… bench-lint #97: Commit 12b21d9 pushed by kacper-daftcode
17s main
main
17s
DeepSeek-V4.1 sm120: the CED prefill is served - since 2026-09-23 21:… bench-lint #96: Commit 12fda7b pushed by kacper-daftcode
13s main
main
13s
DeepSeek-V4.1 sm120: the report's CED prefill measured on the served … bench-lint #95: Commit c15454c pushed by kacper-daftcode
18s main
main
18s
DeepSeek-V4.1 sm120: the KV offload keeps one host copy of the TP-rep… bench-lint #94: Commit c4b9bee pushed by kacper-daftcode
16s main
main
16s
docker/sm120/kvcache-ttl.sh: executable, like the launchers next to i… bench-lint #93: Commit 37a47cb pushed by kacper-daftcode
16s main
main
16s
DeepSeek-V4.1 sm120: KV offload with a disk tier behind host RAM, gen… bench-lint #92: Commit c3b183c pushed by kacper-daftcode
16s main
main
16s
DeepSeek-V4.1 sm120: KV outside HBM, measured - vLLM main's native Of… bench-lint #91: Commit ca23dce pushed by kacper-daftcode
12s main
main
12s
DeepSeek-V4.1 sm120: FlashInfer's PCIe IPC all-reduce measured agains… bench-lint #90: Commit 34ae08c pushed by kacper-daftcode
13s main
main
13s
DeepSeek-V4.1 sm120: the vLLM main line is the served image (dsv41-ni… bench-lint #89: Commit 4f2b6de pushed by kacper-daftcode
16s main
main
16s
DeepSeek-V4.1 sm120: the decode gap of the vLLM-main image explained … bench-lint #88: Commit f6011e4 pushed by kacper-daftcode
16s main
main
16s
DeepSeek-V4.1 sm120: the vLLM main line as a candidate image (Dockerf… bench-lint #87: Commit 7afd6c4 pushed by kacper-daftcode
17s main
main
17s
DeepSeek-V4.1 sm120: DSpark adaptive verification re-measured on the … bench-lint #86: Commit 16d0d06 pushed by kacper-daftcode
14s main
main
14s
DeepSeek-V4.1 sm120: FlashInfer main reads the FP4 compressed record … bench-lint #85: Commit 40ab5c5 pushed by kacper-daftcode
15s main
main
15s
DeepSeek-V4.1 sm120: the compressed KV in its training format - FP4 r… bench-lint #84: Commit 0d375b1 pushed by kacper-daftcode
18s main
main
18s
DeepSeek-V4.1 sm120: the indexer in its training format - MXFP4 K cac… bench-lint #83: Commit 025c0a7 pushed by kacper-daftcode
21s main
main
21s
DeepSeek-V4.1 sm120: the prompt encoder renders the checkpoint's reas… bench-lint #82: Commit a6a8bcd pushed by kacper-daftcode
12s main
main
12s
DeepSeek-V4.1 sm120: hardware-counter anatomy of the decode step (MoE… bench-lint #81: Commit 139c398 pushed by kacper-daftcode
17s main
main
17s
DeepSeek-V4.1 sm120: what the other stack has that we could take - me… bench-lint #80: Commit 2609a43 pushed by kacper-daftcode
16s main
main
16s
DeepSeek-V4.1 on 4x RTX PRO 6000: head-to-head with 0xSero's SGLang r… bench-lint #79: Commit 743a5eb pushed by kacper-daftcode
13s main
main
13s
README: DeepSeek-V4.1 and Qwen3.8 sections describe the served state … bench-lint #78: Commit 07eaa71 pushed by kacper-daftcode
14s main
main
14s
DeepSeek-V4.1: README/Dockerfile/port doc say 4x RTX PRO 6000 (TP4, 5… bench-lint #77: Commit 87f2eca pushed by kacper-daftcode
18s main
main
18s
deploy guide: build times and the from-scratch (--no-cache) reproduci… bench-lint #76: Commit cfe5902 pushed by kacper-daftcode
16s main
main
16s
Confidence gate gains an fp32 signal path (vllm 2eb267534) bench-lint #75: Commit 36a4708 pushed by kacper-daftcode
21s main
main
21s