Skip to page content
SAVRN
Search Plan a project

Open-weight model

sddqwen35a3b

by Max Sun HelloSun/sddqwen35a3b

sddqwen35a3b is an open-weight model from Max Sun. Its published files total 664.0 KB.

在 8 GB RAM 的 Intel 機器上真的跑得動 Qwen3.6-35B-A3B(22.1 GB / Q4KM): 冷專家留在 SSD,RAM 只放有限容量的 expert cache,不用 swap 冒充 RAM。 - 引擎基礎:llama.cpp(pinned commit,見 sdq/config.py) - 核心實作:cpp/sdqpager.cpp(RAM arena + 槽位 + pin + 淘汰 + 指標) + cpp/sdqmoe.cpp(自訂 MoE op)+…

Parameters—
Context—
Weights664.0 KB
License—
AccessOpen weights
Monthly Downloads—

Model Card

在 8 GB RAM 的 Intel 機器上真的跑得動 Qwen3.6-35B-A3B(22.1 GB / Q4KM): 冷專家留在 SSD,RAM 只放有限容量的 expert cache,不用 swap 冒充 RAM。 - 引擎基礎:llama.cpp(pinned commit,見 sdq/config.py) - 核心實作:cpp/sdqpager.cpp(RAM arena + 槽位 + pin + 淘汰 + 指標) + cpp/sdqmoe.cpp(自訂 MoE op)+ cpp/sdqbuft.cpp(讓 18.2 GiB 專家權重不進 RAM) - 進度真相來源:stage.txt(每階段更新,並上傳 HF) - 文件:docs/01 執行路徑分析、docs/02 設計、docs/03 實測結果與瓶頸、 docs/04 mode C 與負結果、docs/05 I/O 段平行與 8 GB 預算修正 - 一鍵從零到聊天:python demo.py 條件:--ram-budget-mb 8192 + RLIMITDATA 8 GB、8 執行緒、ctx 4096、temp 0, 短提示詞、48 個生成 token。換機器數字一定不同(見 docs/05 §1:本機的 SSD 上下文設計上支援 1M,實測只驗 4K。細節見 docs/03results.md(量測方法)、 docs/04experiments.md(mode C 與兩個負結果)、docs/05optimization.md(本輪的 一句話:一個 expert 的三段權重(gate/up/down)在…

Excerpt from the card by Max Sun.

Identity and Version

Repository
HelloSun/sddqwen35a3b
Publisher
Max Sun
Task
Not stated by the source
Modality
Other
Library
Not stated by the source
Parameters
Not stated by the source
Languages
Not stated by the source
Revision
464252e88a63e2a1e7a94f7fcb23d2d5880bdceb
First published
2026-10-03
Last updated
2026-10-04

Files and Weights

49 files, 664.0 KB in total.

Configuration23 files · 160.9 KB
Documentation11 files · 49.7 KB
Other13 files · 451.8 KB
Repository2 files · 1.6 KB
Every file
FileTypeSizeSHA-256
bench/expert_index.jsonConfiguration33.2 KB —
bench/results/bench_first.jsonConfiguration6.8 KB —
bench/results/bench_pipeline.jsonConfiguration6.8 KB —
bench/results/bench_s2.jsonConfiguration6.5 KB —
bench/results/bench_s4_modec.jsonConfiguration6.4 KB —
bench/results/exp_final_ab.jsonConfiguration2.7 KB —
bench/results/exp_rest_modec.jsonConfiguration4.1 KB —
bench/results/exp_s5_io.jsonConfiguration3.1 KB —
bench/results/exp_s5_io2.jsonConfiguration4.0 KB —
bench/results/exp_smoke.jsonConfiguration884 B —
demo.pyConfiguration15.0 KB —
sdq/__init__.pyConfiguration —
sdq/bench.pyConfiguration6.6 KB —
sdq/config.pyConfiguration2.3 KB —
sdq/experiments.pyConfiguration12.8 KB —
sdq/hf_sync.pyConfiguration7.2 KB —
sdq/verify.pyConfiguration9.2 KB —
tools/drop_model_cache.pyConfiguration1.6 KB —
tools/gguf_index.pyConfiguration10.1 KB —
tools/io_probe.pyConfiguration4.3 KB —
tools/moe_layer_check.pyConfiguration3.0 KB —
tools/moe_ref.pyConfiguration9.2 KB —
tools/route_overlap.pyConfiguration5.2 KB —
README.mdDocumentation3.2 KB —
bench/results/exp_final_ab.mdDocumentation715 B —
bench/results/exp_rest_modec.mdDocumentation939 B —
bench/results/exp_s5_io.mdDocumentation875 B —
bench/results/exp_s5_io2.mdDocumentation1.1 KB —
bench/results/exp_smoke.mdDocumentation382 B —
docs/01_moe_path_analysis.mdDocumentation5.9 KB —
docs/02_design.mdDocumentation6.3 KB —
docs/03_results.mdDocumentation5.5 KB —
docs/04_experiments.mdDocumentation13.9 KB —
docs/05_optimization.mdDocumentation11.0 KB —
bench/results/exp_s5_io_final.txtOther463 B —
bench/results/exp_s5_io_run1.logOther1.8 KB —
bench/results/route_decode48.logOther89.5 KB —
bench/results/stats_cold.jsonlOther1.0 KB —
bench/results/stats_hot0.jsonlOther1.0 KB —
bench/results/stats_hot1.jsonlOther1.0 KB —
cpp/sdq_buft.cppOther8.0 KB —
cpp/sdq_cli.cppOther31.8 KB —
cpp/sdq_moe.cppOther31.0 KB —
cpp/sdq_pager.cppOther65.6 KB —
cpp/sdq_pager.hOther17.0 KB —
patches/0001-sdq-expert-pager.patchOther163.4 KB —
stage.txtOther40.2 KB —
.gitattributesRepository1.5 KB —
.gitignoreRepository72 B —

License and Download

License
Not stated by the source
Access
Open weights, no gate
Download from Max Sun

Released by Max Sun through its official repository on Hugging Face.