Add Qwen Community License for chat_template.jinja; add the bf16 source of the int8 maps; data release 1.2

chat_template.jinja is Qwen's template with a RidgeFill header, so the repo now carries Qwen's licence (LICENSE-Qwen) and a NOTICE entry. source/kva-big-s24.safetensors is the bf16 fit the int8 maps were quantized from (the manifest's fit.proj_sha256); the Flash-Next run kit copies its final map into the engine files.

CITATION.cff

@@ -11,7 +11,7 @@ authors:
   - given-names: Dylan
     family-names: Johnston
   - alias: tcclaviger
-version: "1.1"
+version: "1.2"
 license: Apache-2.0
 repository-code: "https://github.com/Approximate-Science/ridgefill-data"
 url: "https://huggingface.co/TheCacheIsALie/RidgeFill-Projector-Qwen3.8-Flash-Next-i8"

LICENSE-Qwen

@@ -0,0 +1,16 @@
+Qwen Community License 1.0
+
+Copyright (c) 2026 Qwen
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software, including the model weights, parameters, configuration files, inference code and associated documentation files (collectively, the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, sell, deploy, host, fine-tune, and create derivative works from (collectively, "Use" or "Using") copies of the Software; and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+1. The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. If the Software (or any derivative works thereof) is Used for any of the licensee's commercial products or services that have more than 100,000,000 monthly active users or US$ 20,000,000 (or equivalent in other currencies) monthly revenue, respective model name must be prominently displayed on the user interface of such product or service; and,
+
+2. If the licensee or any of its affiliates conducts a Model as a Service or AI Work Assistant business, the licensee shall obtain a separate license from Qwen before Using the Software or its derivative works for any commercial purpose. The foregoing requirement shall not apply to the licensee's internal Use of the Software, provided that such Use does not make the Software, its outputs, or its underlying model capabilities available to any third party.
+
+"Model as a Service" means giving a third party access to language model inference or fine-tuning (e.g., via API or a hosted endpoint) in a manner that allows such third parties to exercise meaningful control over the inputs, parameters, or training data. This does not include the mere relaying of requests to models hosted by other third parties. 
+“AI Work Assistant” means an independent AI-powered product primarily designed for AI-assisted coding or office productivity (e.g., Qoder and QwenWork). It does not include: (a) a single-purpose AI tool (such as an AI translation tool); (b) an AI assistant primarily designed for a domain other than coding or office productivity (such as Taobao AI Shopping Assistant or AMap AI Chat); or (c) an AI assistant that is a feature of a product whose primary purpose is not AI-assisted coding or office productivity.
+
+THE SOFTWARE AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL QWEN, ITS AFFILIATES OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. THE USE OF THE SOFTWARE MUST COMPLY WITH APPLICABLE LAWS AND REGULATIONS, AND MUST NOT INFRINGE THE INTELLECTUAL PROPERTY RIGHTS OF ANY THIRD PARTY.
+
+For any questions regarding this license, please contact model-business@notice.qwencloud.com.

NOTICE

@@ -3,7 +3,7 @@ RidgeFill: Copyright 2026 Dylan Johnston and tcclaviger
 Licensed under the Apache License, Version 2.0 (the "License"); see the LICENSE file that
 accompanies this NOTICE, or https://www.apache.org/licenses/LICENSE-2.0.
 
-Cite as: Dylan Johnston and tcclaviger, RidgeFill data and code, version 1.1,
+Cite as: Dylan Johnston and tcclaviger, RidgeFill data and code, version 1.2,
 https://github.com/Approximate-Science/ridgefill-data. An earlier technical note is at
 doi:10.5281/zenodo.23179168 (https://doi.org/10.5281/zenodo.23179168).
 
@@ -30,3 +30,10 @@ What RidgeFill adds: a closed-form ridge fit with no gradient training; maps to
 late layers' INPUTS, so each late layer computes its own K/V and recurrent state; an
 exact tail of prompt tokens; exact-row selection; and a correction of the recurrent
 (gated delta net) state.
+
+Third-party file:
+  chat_template.jinja is Qwen's chat template for Qwen3.8-Flash-Next, unchanged, with a
+  RidgeFill header added at the top. It is Copyright (c) 2026 Qwen and is distributed under
+  the Qwen Community License 1.0, whose full text, with Qwen's copyright notice and
+  permission notice, is in LICENSE-Qwen. The Apache License above covers RidgeFill's own
+  files. Using these maps with Qwen3.8-Flash-Next is subject to that model's licence.

README.md

@@ -39,8 +39,12 @@ inside the serving engine below.
 | `proj8.L24.safetensors` … `proj8.L47.safetensors` (24 files) | Maps for layers 24 to 47. Each: `proj.L.codes` int8 `[2560, 10240]`, `proj.L.scale` bf16 `[2560, 80]`, `proj.L.bias` bf16 `[2560]` | 26,629,560 bytes each |
 | `correction.safetensors` | Gated DeltaNet terminal-state corrections `st.L`, f32 `[48, 128, 128]`, for 18 layers | 56,624,736 bytes |
 | `rowsel.safetensors` | Token-frequency tables `score`, `score_none`, `score_all`, f32 `[248320]` | 2,980,272 bytes |
-| `chat_template.jinja` | Chat template listed in the manifest (not used by the paper's runs) | 12,159 bytes |
-| `LICENSE`, `NOTICE`, `CITATION.cff`, `SHA256SUMS` | Licence, attribution, citation and checksums | |
+| `chat_template.jinja` | Qwen's chat template with a RidgeFill header, listed in the manifest (not used by the paper's runs); Qwen Community License 1.0 | 12,159 bytes |
+| `source/kva-big-s24.safetensors` | The bf16 fit these int8 maps were quantized from (the manifest's `fit.proj_sha256`). The Flash-Next run kit copies its `final` map into the engine's files; no reported number reads that map. Not needed to serve the maps | 1,468,152,288 bytes |
+| `LICENSE`, `LICENSE-Qwen`, `NOTICE`, `CITATION.cff`, `SHA256SUMS` | Licences, attribution, citation and checksums | |
+
+The maps and the manifest are Apache-2.0. `chat_template.jinja` is Qwen's and stays under the Qwen Community License
+1.0 (`LICENSE-Qwen`). Using the maps with Qwen3.8-Flash-Next is subject to that model's licence.
 
 ## Use
 
@@ -51,9 +55,11 @@ and the determinism fix in [`code/engine-fix/`](https://github.com/Approximate-S
 the image is.
 
 ```sh
-hf download TheCacheIsALie/RidgeFill-Projector-Qwen3.8-Flash-Next-i8 --local-dir projector
+hf download TheCacheIsALie/RidgeFill-Projector-Qwen3.8-Flash-Next-i8 --local-dir projector --exclude "source/*"
 ```
 
+Add `source/kva-big-s24.safetensors` only to rebuild the run kit's engine files byte for byte.
+
 ## Results
 
 Two AMD Radeon AI PRO R9700. Exact is the same server with the fill off. Time to first token is measured at the client,
@@ -100,7 +106,7 @@ preparation. Until it is out, cite the data release:
   author  = {Johnston, Dylan and {tcclaviger}},
   title   = {{RidgeFill}: data and code for ``Training-Free Late-Layer Cache Fill for Faster Prefill''},
   year    = {2026},
-  version = {1.1},
+  version = {1.2},
   url     = {https://github.com/Approximate-Science/ridgefill-data}
 }
 ```

SHA256SUMS

@@ -1,6 +1,7 @@
 cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30  LICENSE
-703df017ec66616417fe04ce4f3e04b62569ae7d25537bc9a376e219339fb82f  NOTICE
-3e0d87f5e53b06c154d44a112d821282ca4c97c859aec63fc8352a6f8eff8220  README.md
+a0dc422560841fd68e06d974907f8b4c709bca44a67daad2b528437bdf676c08  LICENSE-Qwen
+2e1afc3da60ab081f579b69072607b078c8dcbb95491806269e607eb862f5107  NOTICE
+984fd23ac106f1f7e7e88b5fd390677152b17d4b225d76c79220bbc6383146d5  README.md
 3c1c94830884ec96c61733698f7af4af088900e8852a2c2c385a779bb5d4c9fe  chat_template.jinja
 d63ecba03570cccd84a2f0e966c98f5c2f2a2053bddcb37c85449808830f52bc  correction.safetensors
 523f6aa12b839fefac1ed532d8810fa284ab0bfe6da988e8b8e2bd87cbbaac96  proj8.L24.safetensors
@@ -29,3 +30,4 @@ a2975e6edeb5165fc4ed4d095dd20f519a7f409163b08fbe3799202bbef29da3  proj8.L44.safe
 e04d881074f100bc271c61d93867677110041bfca8e7f0257835690cb515306a  proj8.L47.safetensors
 795363fa153f0e7e16454af6acabd644709e703d84de0dd9535565226bb51faa  ridgefill.json
 272414e5e19912ba8a93199558dbe7874ace73a382e7ed500ed16f6d566e29f3  rowsel.safetensors
+c9db6064eb275d3166df8139239959dfdeb64f2778707a2f42215164999cb9cd  source/kva-big-s24.safetensors

source/kva-big-s24.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9db6064eb275d3166df8139239959dfdeb64f2778707a2f42215164999cb9cd
+size 1468152288