Sakura DeepSeek-V4.1-Flash-1017 · Maya-XS-IQ1_S · Engram Q3_K
The original Maya-XS-IQ1_S backbone is unchanged. Only the separate Engram lookup-table files are replaced with Q3_K. All experts and all lookup rows are retained.
No full inference test has been performed. The checks cover integrity, format, addressing and isolated decoder paths. No generation-quality, perplexity, speed or hardware-fit score is claimed for this exact package.
Original versus Sakura
| Component | Bytes | Decimal GB |
|---|---|---|
| Matching upstream main GGUF set | 276,626,308,260 | 276.626 |
| Unchanged upstream backbone | 115,341,509,376 | 115.342 |
| Upstream Engram files | 161,284,798,884 | 161.285 |
| Replacement Q3_K Engram files | 84,482,513,884 | 84.483 |
| Sakura main GGUF set | 199,824,023,260 | 199.824 |
| Saved versus this upstream variant | 76,802,285,000 | 76.802 (27.76%) |
Main files: 186.101 GiB. Optional DSpark/drafter, vision, imatrix, runtime and documentation are outside this main-model total. File size is not RAM/VRAM consumption or proof that the package fits a particular machine.
What changed and what stayed identical
| Part | Treatment |
|---|---|
| 3 backbone shards, including tokenizer/model metadata | Copied server-side; source sizes and full HF LFS SHA-256 identities retained |
| Routed experts | No pruning, retraining or requantization by Sakura |
| Engram lookup payload | Existing Sakura Q3_K data reused without another quantization pass |
| Table layout | two separate layer tables; correct split index/count and source tensor-count contract |
| Actual format | GGML type 11 / Q3_K, 110 bytes per 256-value row |
The Maya-XS-IQ1_S label describes the upstream backbone recipe. Q3_K describes the Engram tables, not the entire model.
Provenance
- Original checkpoint: deepseek-ai/DeepSeek-V4.1-Flash-1017, revision
1bae0e1088993b9b82de08f59de5be5a34e42170. - Direct backbone: peasantsmith/DeepSeek-V4.1-Flash-1017-Maya-GGUF, revision
c273fcbdda76a1db074c2f414cb70edbfca2f2ef. - Q3 payload: Sakura Maya-S Engram Q3_K, frozen input revision
4a701f59f623b08a859368926cd9a210ab446941. It was quantized from peasantsmith's Maya-S Q6_K tables atb9e25f34b53806b24e5b18085de6ed5967ad9ed5, using GGMLc7fc5094179bf8a9c73c18593db714f8da41a402. - This repack does not requantize the upstream IQ4_NL, Q5_K, Q6_K or MXFP8 table. It reuses the existing Q3 payload and adjusts only the GGUF table header where required.
The official safetensors checkpoint is about 510.297 GB as a broader package; original Maya-S is 339.478 GB. These are context, not scope-matched quality or memory baselines. The matching-upstream row above is this release's reduction baseline.
Table geometry and addressing
| Property | Value |
|---|---|
| Engram layers | 1 and 14 |
| Layer row counts | 384,006,168 / 384,016,682 |
| Total rows | 768,022,850 |
| Values per row | 256 |
| Total Q3 payload | 84,482,513,500 bytes |
| Storage | two separate layer tables |
| Combined second-layer starting row, where applicable | 384,006,168 |
Tokenizer-derived token mapping, hash constants, bucket sizes and addressing metadata were compared against the relevant source layout. A 192-row Q6 sample matched Maya and Unsloth before the original table reuse. Maya-XS additionally matched 96 Q6 sample rows and the complete token-map hash. Samples alone do not prove complete original Q6 payload identity.
Download and load
hf download webmp3/Sakura-DeepSeek-V4.1-Flash-1017-Maya-XS-Engram-Q3_K-GGUF --local-dir Sakura-Maya-XS-IQ1_S-Engram-Q3_K
Keep all 5 numbered GGUF files together. Load Maya-XS/DeepSeek-V4.1-Flash-1017-Maya-XS-IQ1_S-00001-of-00005.gguf with the matching runtime.
Use Project Maya at ffe2e5e53fd5254e7f3492a20e8ad2b51b26a197 with the included Q3_K source patch. Maya-XS needs this revision's IQ1_S/IQ1_M backbone support. Apply the patch to that exact revision and build using Maya's CUDA or HIP instructions. The scalar Q3 decoder was checked against GGML. CUDA translation units and the isolated GPU test were compiled and linked without GPU execution. HIP compilation, Windows operation and full-model loading remain untested.
What was verified
| Check | Result and scope |
|---|---|
| Backbone identity | Every copied file's size and LFS SHA-256 matched the pinned source |
| Table stream | Header-only repacks verify complete existing Q3 input hashes and independent Python/native output hashes; see the build record |
| Published table layout | Exact headers, shapes, real types, offsets and split metadata checked |
| Published real-row samples | 192 Q3 rows across both layers, including middle/final rows, compared with frozen fixtures |
| Decoder and addressing | Isolated CPU checks; see compatibility record |
| Full inference, generation quality, perplexity and speed | Not tested for this exact package |
| GPU execution / Windows / AMD qualification | Not performed |
Upstream benchmarks and another Sakura variant's results are not measurements of this package. Row reconstruction error is not model quality. Build evidence and exact file identities: build-manifest.json.
Compare the Sakura DeepSeek releases
Open the complete family overview.
| Sakura package | Matching upstream GB | Unchanged backbone GB | Sakura main files GB | Saving GB |
|---|---|---|---|---|
| Maya-XS-IQ1_S | 276.626 | 115.342 | 199.824 | 76.802 |
| UD-IQ1_S | 232.980 | 122.384 | 206.867 | 26.113 |
| UD-IQ1_M | 250.404 | 139.809 | 224.292 | 26.113 |
| UD-IQ2_XXS | 300.475 | 165.303 | 249.786 | 50.690 |
| Maya-S-IQ2_XXS | 339.478 | 178.193 | 262.676 | 76.802 |
| UD-Q2_K_XL | 334.794 | 199.622 | 284.104 | 50.690 |
| UD-IQ3_XXS | 394.737 | 233.452 | 317.934 | 76.802 |
| UD-IQ3_S | 423.631 | 262.346 | 346.829 | 76.802 |
| UD-Q3_K_XL | 485.069 | 282.311 | 366.793 | 118.276 |
| UD-Q4_K_XL | 500.708 | 297.950 | 382.432 | 118.276 |
| UD-Q8_K_XL | 508.377 | 305.619 | 390.101 | 118.276 |
These compare disk sizes and source recipes, not measured answer quality.
Exact main files
| File | Bytes | SHA-256 |
|---|---|---|
Maya-XS/DeepSeek-V4.1-Flash-1017-Maya-XS-IQ1_S-00001-of-00005.gguf |
47,781,530,400 | a0cb7966afd68bc35c46f78018d1fa1d2e4bcbe3c3baf39f60cd1733ed9ad7e5 |
Maya-XS/DeepSeek-V4.1-Flash-1017-Maya-XS-IQ1_S-00002-of-00005.gguf |
47,775,748,128 | ba5afb43adfc667436ebcfbd82d25411fa2843ee7e1d877290611cd60f93cca1 |
Maya-XS/DeepSeek-V4.1-Flash-1017-Maya-XS-IQ1_S-00003-of-00005.gguf |
19,784,230,848 | 1482443bb4957c8d2f005ded365445f98558ff05c50674f9ee9a96ddabc122a4 |
Maya-XS/DeepSeek-V4.1-Flash-1017-Maya-XS-IQ1_S-00004-of-00005.gguf |
42,240,678,672 | 2a4b2a5313fa32df91a8ac7730b224b6986c17a2ecfff686229e8dae466407d5 |
Maya-XS/DeepSeek-V4.1-Flash-1017-Maya-XS-IQ1_S-00005-of-00005.gguf |
42,241,835,212 | 2b6b0ae645890f9e59455fc6a1c9e8dd10ce96a71406955e4f9ff21b6bbb19cd |
Credits and license
DeepSeek-AI: original model and Engram. Unsloth / peasantsmith / Project Maya: the credited upstream recipes and runtimes. GGML / llama.cpp: Q3_K format, reference quantizer and decoder. Sakura: table compression, format adaptation, integrity checks and release assembly. No upstream endorsement or authorship of its backbone is claimed.
MIT model/source notices are retained in the included license files. Builds use free public standard CPU runners, no paid runner/GPU or inference API. Public HF storage is best-effort; no quota-neutral deduplication is promised.
中文说明（点击展开）模型、来源与大小比较
本版本保留 Maya-XS-IQ1_S 原始模型主体，所有专家和查找行均保留；仅将独立的 Engram 表文件替换为 Q3_K。Q3_K 只描述查找表，不表示整个模型都是 3 位。
| 内容 | 十进制 GB |
|---|---|
| 对应上游主模型分片 | 276.626 |
| 完全不变的模型主体 | 115.342 |
| 上游 Engram 表 | 161.285 |
| 本版本 Q3_K 表 | 84.483 |
| 本版本主分片合计 | 199.824 |
| 减少 | 76.802 |
直接主体来源为 peasantsmith/DeepSeek-V4.1-Flash-1017-Maya-GGUF，固定版本 c273fcbdda76a1db074c2f414cb70edbfca2f2ef。原始模型为 deepseek-ai/DeepSeek-V4.1-Flash-1017，版本 1bae0e1088993b9b82de08f59de5be5a34e42170。Q3 数据来自已有 Sakura Maya-S 表，其输入为 Maya-S Q6_K；本次只适配表头，不再次量化模型主体或 Q3 数据。
完整保留 768,022,850 行，每行 256 值、110 字节。层 1/14 分别有 384,006,168 / 384,016,682 行。主体分片大小及 SHA-256 与上游一致；发布后的表头和 192 行真实 Q3 数据已核对。查找地址、哈希参数及完整 token map 按相应布局检查。
下载时将 5 个编号 GGUF 放在一起，从第一个分片加载。需要指定版本的 Project Maya 和本仓库 Q3 补丁，Maya-XS 依赖新版 IQ1 支持。 固定运行时版本：ffe2e5e53fd5254e7f3492a20e8ad2b51b26a197。详见上方加载说明。
未进行完整模型推理测试。 没有本版本的回答质量、困惑度、速度、GPU 执行、Windows 或 AMD 合格结论。源模型分数不能直接移用；文件大小不等于内存需求。文件校验和、许可、检查边界和其他 Sakura 版本比较见上方英文说明及清单。