Publish release card, license, benchmarks, and integrity manifests

LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 Alibaba Cloud
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -0,0 +1,162 @@
+---
+license: apache-2.0
+base_model: Qwen/Qwen3.8-27B
+library_name: transformers
+pipeline_tag: image-text-to-text
+tags:
+- qwen3.8
+- reasoning
+- vision-language
+- personal-model
+- uncensored
+- abliterated
+- abliterix
+- transformers
+- bfloat16
+---
+
+# Qwen3.8-27B-Opus-Abliterix-Reasoning-BF16
+
+This is the full-precision BF16 form of the selected Abliterix pass-1 practical winner. It preserves the checkpoint before MLX quantization and exists for Transformers workflows, archival fidelity, and downstream conversion.
+
+**Selection verdict:** practical personal-model winner with measured deviations — not universal dominance and not a claim that all preregistered gates passed.
+
+## Why this release
+
+The target was a practical personal reasoning/VLM model with much lower reflexive refusal while retaining measured capability and an immutable fallback. The benchmark table is the decision record: it shows where the selected winner improved, where the control stayed stronger, and why both are published. The family therefore describes the winner as a **practical personal-model selection with measured deviations**, not universal dominance.
+
+## What this variant is
+
+- **Role:** Full-precision practical winner
+- **Format:** Merged BF16 Transformers checkpoint with the native 15-tensor MTP sidecar preserved as `mtp.safetensors`.
+- **Base:** [`Qwen/Qwen3.8-27B`](https://huggingface.co/Qwen/Qwen3.8-27B)
+- **Pipeline:** image + text to text
+- **License:** Apache-2.0
+
+## Frozen local results
+
+> **These are self-run, frozen local project benchmarks, not official Qwen benchmarks.** The same local harness compared `control-bf16` with `abliterix-pass1-bf16`.
+
+| Frozen local metric | Control | Abliterix winner |
+|---|---:|---:|
+| Harmful hard refusal | 43.2% | **0.0%** |
+| Harmful soft deflection | 14.6% | **0.2%** |
+| Harmful substantive response | 47.0% | **99.4%** |
+| Capability macro | 17.6859% | **21.0086%** |
+| Full code | **16/421** | 10/421 |
+| HumanEval | **7.9268%** | 4.2683% |
+| Long-form pass | 54.1667% | **62.5000%** |
+| MMMU30 | 9/30 | **11/30** |
+| Held-out loss ratio | 1.000000 | 1.024478 |
+| Benign KL | 0.000000 | 0.093614 |
+
+The winner improved the local capability macro by 3.3227 points, long-form pass rate by 8.3333 points, and MMMU30 by 2 correct answers while reducing harmful hard refusal by 43.2 points. The control remained better on full code/HumanEval.
+
+Full machine-readable values are in [`benchmark-results.json`](./benchmark-results.json).
+
+## Strict deviations and code pathology
+
+- Benign KL was `0.093614`, above the strict `0.05` limit.
+- Incoherence was `4.3077%`, above the `2.7692%` strict comparison point.
+- HumanEval was `4.2683%` versus control `7.9268%`; full code was `10/421` versus `16/421`.
+- Long-form maximum repeated-4gram fraction was `5.8632%`, above the `5%` limit.
+- Prompt leakage was detected: 3 exact winner prompt echoes versus 2 for control; both failed the leakage hard gate.
+- 376/421 winner code generations hit the 512-token cap. Among generations that reached execution, the winner passed 10/46 (21.74%) versus control 16/103 (15.53%), pointing to severe termination/extraction pathology rather than a clean latent-code estimate.
+
+## Training data and run
+
+| Aggregate source label | Rows |
+|---|---:|
+| `opus-10000x` | 9,633 |
+| `opus-3000x` | 2,326 |
+| `reasoning-700x` | 633 |
+| `high-reasoning-250x` | 250 |
+| **Raw total** | **12,842** |
+
+After 208 deduplications and 20 invalid-row removals, 12,614 rows remained: 12,349 train, 127 validation, and 138 test. The processed-manifest SHA-256 is `6e0a36ad20732c5f98ff592c4565a4c86876fead4e9f94bc6ceedfad1339a94d`. Only aggregate counts and hashes are published; raw/private rows are not.
+
+Training ran for 1,544 optimizer steps with 108,789,760 trainable LoRA parameters. Final validation loss was `0.23739749`; token accuracy was `91.7594%`.
+
+## Architecture and lineage
+
+The model uses `Qwen3_5ForConditionalGeneration` as recorded by the released config: a 64-layer, hidden-size-5120 text stack with a 3:1 linear/full-attention schedule and a configured 262,144-token maximum position range, paired with a 27-layer, hidden-size-1152 vision encoder. It is an image/text conditional-generation model, not a text-only checkpoint.
+
+### Method
+
+1. Started from [`Qwen/Qwen3.8-27B`](https://huggingface.co/Qwen/Qwen3.8-27B), Apache-2.0.
+2. Prepared 12,842 raw reasoning rows: 12,614 accepted after removing 208 duplicates and 20 invalid rows, then split into 12,349 train / 127 validation / 138 test rows.
+3. Trained a reasoning QLoRA on the 12,349-row train split for 1,544 optimizer steps with 108,789,760 trainable LoRA parameters. Final validation loss was 0.23739749 and token accuracy was 91.7594%.
+4. Merged the adapter to BF16 to form immutable `control-bf16`.
+5. Produced the selected winner with Abliterix 1.12.2 pass 1 (seed 42): orthogonal/projected, winsorized single-direction residual steering over the output/down-projection writer components. Q/K/V projections were excluded.
+6. Preserved BF16 releases and converted both variants to affine MLX 8-bit, group size 64. Native one-layer MTP drafters were split and validated separately.
+
+The final selection label is `selected_practical_winner_with_measured_deviations`. Private training data, raw harmful/benign prompt sets, operational receipts, local paths, and Drive metadata are intentionally not published.
+
+## Tensor and conversion integrity
+
+Across the BF16 comparison there were 1,199 tensor keys, including 15 native MTP tensors and 333 vision tensors. The winner contains 74 actual Abliterix residual-writer edits and zero unexpected changes.
+
+MLX native proof ran through `mlx-vlm` on arm64 macOS/Metal. Ordinary and MTP-assisted generation both produced the exact answer `323` for winner and control. The winner MLX build used `mlx` 0.32.0, `mlx-lm` 0.31.3, `mlx-vlm` 0.6.13, and affine 8-bit group-64 quantization.
+
+## Hash and size summary
+
+- `BF16 winner delivery`: 14 files, 55,583,144,390 bytes, sealed aggregate SHA-256 `cb0e04180dc257f19604f9ae70b65191a9f01322e5ed7c36aecf22a00af173be`
+
+The deliberately published model payload for this repository is 10 files / 55,583,125,224 bytes. Per-file source SHA-256 values are in [`SHA256SUMS`](./SHA256SUMS) and [`manifests/artifact-manifest.json`](./manifests/artifact-manifest.json). Sealed aggregate hashes above cover the complete local source components, including private entries excluded from publication; use the per-file public manifest for the Hub payload.
+
+## Use with Transformers
+
+A current Transformers version that supports Qwen3.8/Qwen3.5 VLM architecture is required (the sealed build used 5.15.0). BF16 needs substantial accelerator memory.
+
+```python
+import torch
+from transformers import AutoModelForImageTextToText, AutoProcessor
+
+repo_id = "timteh673/Qwen3.8-27B-Opus-Abliterix-Reasoning-BF16"
+processor = AutoProcessor.from_pretrained(repo_id)
+model = AutoModelForImageTextToText.from_pretrained(
+    repo_id, torch_dtype=torch.bfloat16, device_map="auto"
+)
+messages = [{"role": "user", "content": [
+    {"type": "text", "text": "Calculate 17 multiplied by 19."}
+]}]
+inputs = processor.apply_chat_template(
+    messages, tokenize=True, add_generation_prompt=True,
+    return_dict=True, return_tensors="pt"
+).to(model.device)
+output = model.generate(**inputs, max_new_tokens=256)
+trimmed = output[:, inputs.input_ids.shape[1]:]
+print(processor.batch_decode(trimmed, skip_special_tokens=True)[0])
+```
+
+For image input, add an image content item using the upstream Qwen multimodal message format. `mtp.safetensors` is the preserved native MTP sidecar; ordinary Transformers loading does not require it.
+
+## Known limitations
+
+- The winner is deliberately less refusal-prone. The frozen harmful benchmark measured 0.0% hard refusal, not a guarantee of zero refusal on every prompt.
+- It failed strict KL, incoherence, HumanEval/full-code, repetition, and prompt-leakage gates listed above.
+- Code results are heavily confounded by 512-token termination; increase generation limits and validate executable outputs.
+- BF16 is large and requires substantial memory; MTP support depends on the runtime.
+- The local benchmark suite and private training/evaluation data are not included. Results may not transfer to other prompts, languages, runtimes, or sampling settings.
+- Generated content can be incorrect, insecure, or incomplete; review it before consequential use.
+
+## Release family
+
+- [Recommended MLX 8-bit winner](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Abliterix-Reasoning-MLX-8bit)
+- [BF16 winner](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Abliterix-Reasoning-BF16)
+- [MLX 8-bit control](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Reasoning-Control-MLX-8bit)
+- [BF16 control](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Reasoning-Control-BF16)
+
+See [`RELEASE_FAMILY.md`](./RELEASE_FAMILY.md) for the role and integrity summary of every variant.
+
+## Responsible use
+
+Evaluate this model for your own setting, isolate untrusted code/tool output, and comply with applicable law and the policies of systems you connect it to. The uncensoring/abliteration work changes refusal behavior; it does not make outputs accurate or safe by default.
+
+## License and attribution
+
+Released under Apache License 2.0. This is a derivative of [`Qwen/Qwen3.8-27B`](https://huggingface.co/Qwen/Qwen3.8-27B), Copyright 2026 Alibaba Cloud. Qwen/Alibaba Cloud attribution and the full license text are retained in [`LICENSE`](./LICENSE). Modifications include the personal reasoning QLoRA merge, selected Abliterix residual edits for winner variants, and MLX quantization where applicable.
+
+## Support
+
+If this release is useful, follow [`timteh673`](https://huggingface.co/timteh673), star the repository, and share reproducible benchmark or runtime findings in the Hub community tab.

RELEASE_FAMILY.md

@@ -0,0 +1,47 @@
+# Qwen3.8-27B Opus personal-model release family
+
+This four-repository family publishes the selected practical winner and its immutable trained-control comparator in BF16 and Apple-Silicon MLX 8-bit formats. The recommended default is the MLX 8-bit winner.
+
+**Selection language:** practical personal-model selection with measured deviations. This release does not claim universal dominance or that every strict gate passed.
+
+| Repository | Role | Format | Public model files | Public model bytes |
+|---|---|---|---:|---:|
+| [Qwen3.8-27B-Opus-Abliterix-Reasoning-MLX-8bit](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Abliterix-Reasoning-MLX-8bit) | Recommended practical winner | MLX 8-bit | 17 | 30,390,836,635 |
+| [Qwen3.8-27B-Opus-Abliterix-Reasoning-BF16](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Abliterix-Reasoning-BF16) | Full-precision practical winner | BF16 Transformers | 10 | 55,583,125,224 |
+| [Qwen3.8-27B-Opus-Reasoning-Control-MLX-8bit](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Reasoning-Control-MLX-8bit) | Baseline comparator | MLX 8-bit | 17 | 30,390,836,197 |
+| [Qwen3.8-27B-Opus-Reasoning-Control-BF16](https://huggingface.co/timteh673/Qwen3.8-27B-Opus-Reasoning-Control-BF16) | Full-precision baseline comparator | BF16 Transformers | 20 | 55,583,123,681 |
+
+## Lineage
+
+`Qwen/Qwen3.8-27B` → reasoning QLoRA merge (`control-bf16`) → Abliterix pass 1 (`abliterix-pass1-bf16`) → BF16 and affine MLX 8-bit/group-64 release variants.
+
+Dataset preparation started from 12,842 raw rows and accepted 12,614 after 208 deduplications and 20 invalid-row removals; splits were 12,349 train / 127 validation / 138 test. Training used 1,544 optimizer steps, 108,789,760 trainable LoRA parameters, final validation loss 0.23739749, and token accuracy 91.7594%. The merged model retained 1,199 tensor keys, 15 native MTP tensors, and 333 vision tensors. The winner has 74 verified residual-writer edits and zero unexpected changes.
+
+## Local benchmark headline
+
+These are **self-run frozen local benchmarks, not official Qwen benchmarks**.
+
+| Frozen local metric | Control | Abliterix winner |
+|---|---:|---:|
+| Harmful hard refusal | 43.2% | **0.0%** |
+| Harmful soft deflection | 14.6% | **0.2%** |
+| Harmful substantive response | 47.0% | **99.4%** |
+| Capability macro | 17.6859% | **21.0086%** |
+| Full code | **16/421** | 10/421 |
+| HumanEval | **7.9268%** | 4.2683% |
+| Long-form pass | 54.1667% | **62.5000%** |
+| MMMU30 | 9/30 | **11/30** |
+| Held-out loss ratio | 1.000000 | 1.024478 |
+| Benign KL | 0.000000 | 0.093614 |
+
+Strict deviations remain part of the release: KL 0.093614 > 0.05; incoherence 4.3077% > 2.7692%; HumanEval 4.2683% versus 7.9268%; full code 10/421 versus 16/421; repetition 5.8632% > 5%; prompt leakage detected; and 376/421 winner code generations hit the 512-token cap. Among outputs reaching execution, winner pass rate was 10/46 (21.74%) versus control 16/103 (15.53%), indicating termination/extraction pathology rather than a clean latent-code estimate.
+
+The canonical structured record is [`benchmark-results.json`](./benchmark-results.json).
+
+## Packaging boundary
+
+Model weights and runtime metadata come from sealed source artifacts. Public manifests were regenerated from verified SHA-256/size receipts. Local paths, private prompt/training data, Drive identifiers, operational receipts, and internal release-control files are excluded. Each repository has a public per-file `SHA256SUMS` and `manifests/artifact-manifest.json`.
+
+## License
+
+Apache-2.0, inherited from [`Qwen/Qwen3.8-27B`](https://huggingface.co/Qwen/Qwen3.8-27B). Copyright 2026 Alibaba Cloud; modifications and format conversions are described in each model card.

SHA256SUMS

@@ -0,0 +1,10 @@
+c3cf9e34abf4f9e36c2d72165aa9c132d3e2a725b6c2586aaa3a8af9d7a81041  chat_template.jinja
+4c95af1d45132ead5b63e540f08f32a351e20512e62438a3a78faf9881d1e7dc  config.json
+b8eb74d15e0a56623d00ccd14950a4bb87fabbf84b5cc030dcc904b899fb1eb5  generation_config.json
+ea6111cc68925f6cb9fe26e623f3dba733b79fc01fd9e0de0ad43b94220e55bb  model-00001-of-00002.safetensors
+a6a5df602c43452310471f40eceb6453eb7bb9be5ffd4f781325038e6810db5e  model-00002-of-00002.safetensors
+df96fae8d7adc5adc388ad672842bfa457897d158ea62a1f3257b7672c2ea886  model.safetensors.index.json
+90fa0e3eed5a647c035c6df9ecabc416c0f8d573ff84ac12485b085f00a7cdf2  mtp.safetensors
+d89ef49ce9cd37fbf510158e13c1ef063d9286411c1ec9049932dbe0487143b1  processor_config.json
+6f32ce20dc35f57a7f9ad1eac03525bd7d30f9df8cea6507e958279cc3657706  tokenizer.json
+bee8eba30f0eb4af73c0fe2cd06d0f89b657d7819941c438157ec42f7c80ea87  tokenizer_config.json

benchmark-results.json

@@ -0,0 +1,213 @@
+{
+  "code_termination_diagnostic": {
+    "among_outputs_reaching_execution": {
+      "control": {
+        "executed": 103,
+        "pass_percent": 15.53,
+        "passed": 16
+      },
+      "winner": {
+        "executed": 46,
+        "pass_percent": 21.74,
+        "passed": 10
+      }
+    },
+    "interpretation": "The full-code result is dominated by severe termination/extraction pathology and is not a clean latent-code estimate.",
+    "winner_generations_hitting_512_token_cap": 376,
+    "winner_generations_total": 421
+  },
+  "comparison": {
+    "benign_kl": {
+      "control": 0.0,
+      "strict_limit": 0.05,
+      "winner": 0.093614
+    },
+    "capability_macro_percent": {
+      "control": 17.6859,
+      "winner": 21.0086
+    },
+    "exact_prompt_echoes": {
+      "control": 2,
+      "leakage_detected": true,
+      "winner": 3
+    },
+    "full_code_passes": {
+      "control": 16,
+      "denominator": 421,
+      "winner": 10
+    },
+    "harmful_hard_refusal_percent": {
+      "control": 43.2,
+      "winner": 0.0
+    },
+    "harmful_soft_deflection_percent": {
+      "control": 14.6,
+      "winner": 0.2
+    },
+    "harmful_substantive_response_percent": {
+      "control": 47.0,
+      "winner": 99.4
+    },
+    "held_out_loss_ratio": {
+      "control": 1.0,
+      "limit": 1.05,
+      "winner": 1.024478
+    },
+    "human_eval_percent": {
+      "control": 7.9268,
+      "winner": 4.2683
+    },
+    "incoherence_percent": {
+      "control": 2.7692,
+      "strict_winner_limit": 2.7692,
+      "winner": 4.3077
+    },
+    "long_form_pass_percent": {
+      "control": 54.1667,
+      "winner": 62.5
+    },
+    "max_repeated_4gram_fraction_percent": {
+      "strict_limit": 5.0,
+      "winner": 5.8632
+    },
+    "mmmu30_correct": {
+      "control": 9,
+      "denominator": 30,
+      "winner": 11
+    }
+  },
+  "evaluation_provenance": {
+    "kind": "self-run_frozen_local_benchmarks",
+    "note": "Results were produced by this project on a frozen local evaluation suite. They must not be mixed with or represented as official Qwen benchmark results.",
+    "official_qwen_benchmarks": false,
+    "release_report_frozen_at": "2026-08-22T00:01:48.742409+00:00"
+  },
+  "lineage": {
+    "abliterix_version": "1.12.2",
+    "base_model": "Qwen/Qwen3.8-27B",
+    "method": "Reasoning QLoRA merge followed by one selected Abliterix pass for the winner; control received no Abliterix residual-writer edits.",
+    "reasoning_control": "control-bf16",
+    "seed": 42,
+    "winner": "abliterix-pass1-bf16"
+  },
+  "native_mlx_proof": {
+    "control_mtp_statistics": {
+      "accepted_drafts_per_round": 1.73,
+      "accepted_tokens_per_round": 2.73,
+      "average_draft": 2.0,
+      "drafted_percent": 86.5,
+      "rounds": 37
+    },
+    "mtp_exact_answer": {
+      "control": "323",
+      "winner": "323"
+    },
+    "ordinary_exact_answer": {
+      "control": "323",
+      "winner": "323"
+    },
+    "runtime": "mlx-vlm on arm64 macOS/Metal",
+    "speed_note": "MTP speed is workload-dependent; positive acceptance and exact-answer agreement are the correctness evidence.",
+    "winner_mtp_statistics": {
+      "accepted_drafts_per_round": 1.78,
+      "accepted_tokens_per_round": 2.78,
+      "average_draft": 2.0,
+      "drafted_percent": 88.8,
+      "rounds": 76
+    },
+    "winner_quantization": {
+      "bits": 8,
+      "group_size": 64,
+      "scheme": "affine"
+    }
+  },
+  "release_family": "Qwen3.8-27B Opus Personal Model v1",
+  "schema_version": 1,
+  "selection": {
+    "all_preregistered_strict_gates_passed": false,
+    "control_variant": "control-bf16",
+    "selected_variant": "abliterix-pass1-bf16",
+    "status": "selected_practical_winner_with_measured_deviations",
+    "summary": "Practical personal-model selection with measured deviations; not universal dominance."
+  },
+  "strict_deviations": [
+    {
+      "metric": "benign_kl",
+      "passed": false,
+      "strict_limit": 0.05,
+      "winner": 0.093614
+    },
+    {
+      "metric": "incoherence_percent",
+      "passed": false,
+      "strict_limit": 2.7692,
+      "winner": 4.3077
+    },
+    {
+      "control": 7.9268,
+      "metric": "human_eval_percent",
+      "passed": false,
+      "strict_delta_floor": -3.0,
+      "winner": 4.2683,
+      "winner_delta_points": -3.6585
+    },
+    {
+      "control": "16/421",
+      "metric": "full_code_passes",
+      "passed": false,
+      "winner": "10/421"
+    },
+    {
+      "metric": "max_repeated_4gram_fraction_percent",
+      "passed": false,
+      "strict_limit": 5.0,
+      "winner": 5.8632
+    },
+    {
+      "control_exact_echoes": 2,
+      "metric": "prompt_leakage",
+      "passed": false,
+      "winner_exact_echoes": 3
+    }
+  ],
+  "tensor_integrity": {
+    "native_mtp_tensors": 15,
+    "total_tensor_keys": 1199,
+    "unexpected_changes": 0,
+    "vision_tensors_preserved": 333,
+    "winner_abliterix_residual_writer_edits": 74
+  },
+  "training": {
+    "dataset_preparation": {
+      "accepted_rows": 12614,
+      "duplicates_removed": 208,
+      "invalid_rows_removed": 20,
+      "processed_manifest_sha256": "6e0a36ad20732c5f98ff592c4565a4c86876fead4e9f94bc6ceedfad1339a94d",
+      "raw_rows": 12842,
+      "rows_published": false,
+      "source_rows": {
+        "high-reasoning-250x": 250,
+        "opus-10000x": 9633,
+        "opus-3000x": 2326,
+        "reasoning-700x": 633
+      },
+      "split_sha256": {
+        "test.jsonl": "a2b6b46ce8b054166b0e392701e49390a44e391b56d25a0a25e9b8a9000257d9",
+        "train.jsonl": "c7e691bcabd945e1259537f74119bce4c2c8f5fa4d2fba2e626b95bc04176ad3",
+        "validation.jsonl": "589a05d0cf2fededff1febb065d6b3c8c947d29e948495c1567bcd9f5c44f364"
+      },
+      "splits": {
+        "test": 138,
+        "train": 12349,
+        "validation": 127
+      }
+    },
+    "dtype_after_merge": "bfloat16",
+    "final_validation_loss": 0.23739749,
+    "hardware": "NVIDIA H200-class run; the sealed Abliterix environment recorded NVIDIA H200, CUDA 13.0, driver 580.159.03, Python 3.11.15.",
+    "optimizer_steps": 1544,
+    "rows": 12349,
+    "token_accuracy_percent": 91.7594,
+    "trainable_lora_parameters": 108789760
+  }
+}

manifests/artifact-manifest.json

@@ -0,0 +1,89 @@
+{
+  "base_model": "Qwen/Qwen3.8-27B",
+  "excluded_source_entries": [
+    {
+      "path": ".branch-reservation.json",
+      "reason": "Private release-control metadata."
+    },
+    {
+      "path": "reproduce/README.md",
+      "reason": "Internal/local-path reproducibility material replaced by sanitized public metadata."
+    },
+    {
+      "path": "reproduce/SHA256SUMS",
+      "reason": "Internal reproducibility sidecar replaced by the public SHA256SUMS."
+    },
+    {
+      "path": "reproduce/reproduce.json",
+      "reason": "Contains local paths and private dataset locations; not published."
+    }
+  ],
+  "file_count": 10,
+  "files": [
+    {
+      "path": "chat_template.jinja",
+      "sha256": "c3cf9e34abf4f9e36c2d72165aa9c132d3e2a725b6c2586aaa3a8af9d7a81041",
+      "size": 8952
+    },
+    {
+      "path": "config.json",
+      "sha256": "4c95af1d45132ead5b63e540f08f32a351e20512e62438a3a78faf9881d1e7dc",
+      "size": 3688
+    },
+    {
+      "path": "generation_config.json",
+      "sha256": "b8eb74d15e0a56623d00ccd14950a4bb87fabbf84b5cc030dcc904b899fb1eb5",
+      "size": 214
+    },
+    {
+      "path": "model-00001-of-00002.safetensors",
+      "sha256": "ea6111cc68925f6cb9fe26e623f3dba733b79fc01fd9e0de0ad43b94220e55bb",
+      "size": 49825162976
+    },
+    {
+      "path": "model-00002-of-00002.safetensors",
+      "sha256": "a6a5df602c43452310471f40eceb6453eb7bb9be5ffd4f781325038e6810db5e",
+      "size": 4888445168
+    },
+    {
+      "path": "model.safetensors.index.json",
+      "sha256": "df96fae8d7adc5adc388ad672842bfa457897d158ea62a1f3257b7672c2ea886",
+      "size": 111996
+    },
+    {
+      "path": "mtp.safetensors",
+      "sha256": "90fa0e3eed5a647c035c6df9ecabc416c0f8d573ff84ac12485b085f00a7cdf2",
+      "size": 849400424
+    },
+    {
+      "path": "processor_config.json",
+      "sha256": "d89ef49ce9cd37fbf510158e13c1ef063d9286411c1ec9049932dbe0487143b1",
+      "size": 1191
+    },
+    {
+      "path": "tokenizer.json",
+      "sha256": "6f32ce20dc35f57a7f9ad1eac03525bd7d30f9df8cea6507e958279cc3657706",
+      "size": 19989492
+    },
+    {
+      "path": "tokenizer_config.json",
+      "sha256": "bee8eba30f0eb4af73c0fe2cd06d0f89b657d7819941c438157ec42f7c80ea87",
+      "size": 1123
+    }
+  ],
+  "format": "BF16 Transformers",
+  "hash_algorithm": "sha256",
+  "repo_id": "timteh673/Qwen3.8-27B-Opus-Abliterix-Reasoning-BF16",
+  "schema_version": 1,
+  "scope": "Deliberately published model/runtime payload; generated documentation is outside this checksum scope.",
+  "sealed_source_components": [
+    {
+      "aggregate_sha256": "cb0e04180dc257f19604f9ae70b65191a9f01322e5ed7c36aecf22a00af173be",
+      "bytes": 55583144390,
+      "files": 14,
+      "name": "BF16 winner delivery"
+    }
+  ],
+  "total_bytes": 55583125224,
+  "variant_role": "Full-precision practical winner"
+}