Model card, EXPORT.md, attribution, small checkpoint (model/), inventory and weight digests

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+model/tokenizer.json filter=lfs diff=lfs merge=lfs -text

EXPORT.md

@@ -0,0 +1,66 @@
+# Converting this export to Core AI packages
+
+These steps rebuild the Core AI target published in [anemll/anemll-forge-qwen3.8-27B](https://huggingface.co/anemll/anemll-forge-qwen3.8-27B) (16 chunk packages, the output head and a manifest) from this repository, with [ANEMLL Forge](https://github.com/Anemll/anemll-forge). They were run on an M6 with macOS 27. About 25 minutes of export and, overlapped with it, the first ANE compile.
+
+## 1. Environment
+
+Clone ANEMLL Forge and create a **conversion** environment, separate from the inference one (the builder patches `coreai-opt` internals, so keep the pinned versions):
+
+```sh
+git clone https://github.com/Anemll/anemll-forge.git && cd anemll-forge
+uv venv --python 3.13 --seed .venv-convert     # or: python3.13 -m venv .venv-convert
+.venv-convert/bin/python -m pip install -r requirements-conversion.txt
+```
+
+## 2. Download this repository
+
+```sh
+export Q="$HOME/Models/anemll-quantized-qwen3.8-27b-for-CoreAI"
+hf download anemll/anemll-quantized-qwen3.8-27b-for-CoreAI --local-dir "$Q"       # about 18 GB
+```
+
+Check the weights the converter will read against the published digests (2,097 arrays; a few minutes):
+
+```sh
+MODEL="$Q/model" EXPORT_DIR="$Q/export/mix25in_mixr_lr64mix" \
+  .venv-convert/bin/python scripts/qwen38_weights_digest.py --check "$Q/weights_digest.json"
+```
+
+## 3. Export
+
+The published packages: 8-bit attention with M6 and M5 function sets in the same packages, transposed key cache, V8 cache, five context entries:
+
+```sh
+export OUT="$HOME/coreai-builds"                  # writes $OUT/mix25in_mixr_lr64mix_kvv8
+MODEL="$Q/model" EXPORT_DIR="$Q/export/mix25in_mixr_lr64mix" OUT="$OUT" \
+SILU=tanh MLP_SILU=tanh GDN_SQ=16 GDN_SV=64 MLP_DS=1 QCONV_INT8=0 \
+ATT_S8_UNIT=0.25 ATT_S8B_UNIT=0.25 ATT_INT8MM=s8,s8b,sm8,pvf8 ATT_INT8MM_M5=s8,s8b KV_KEYS_T=1 \
+  .venv-convert/bin/python coreai/qwen38_coreai_build.py all --kv-cache-dtype v8 \
+  --ctx 8192,16384,32768,49152,65536 --pctx 8192,16384,32768,49152,65536
+```
+
+To compile for this Mac's ANE while the export runs, start this first, in the **inference** environment (the compile cache is per Python):
+
+```sh
+python forge.py compile --follow --build "$OUT/mix25in_mixr_lr64mix_kvv8" &
+```
+
+Variants: drop `ATT_INT8MM_M5` for an M6-only build; drop the `ATT_*` and `KV_KEYS_T` settings for the plain V8 graph; `--ctx` / `--pctx` choose the context entries. The settings are recorded in each chunk's `numerics` in `manifest.json`, and the server prints them at startup.
+
+## 4. Run it
+
+The runtime also needs the tokenizer and FP16 embedding table (`model/`) and the paired DFlash2 drafter (`drafter/`) of the inference bundle. Download that bundle and point `BUILD` at the new packages:
+
+```sh
+export FORGE_BUNDLE="$HOME/Models/anemll-forge-qwen3.8-27B"
+python forge.py download --repo anemll/anemll-forge-qwen3.8-27B --revision main --runtime coreai --output "$FORGE_BUNDLE"
+BUILD="$OUT/mix25in_mixr_lr64mix_kvv8" CTX=64K scripts/qwen38_server.sh start
+```
+
+On an M5 the runtime derives its M5 function set once on first start (it needs `coreai-core`, included in `requirements-inference.txt`).
+
+## What to expect
+
+- The weights inside your packages equal the published ones (step 2's check). The package bytes differ slightly from the published files and from build to build: the Core AI converter's serialization is not byte-deterministic.
+- Disk: 18 GB for this repository, about 10 GB per full build, and 11 to 14 GB of compile cache per build in `~/Library/Caches/coreai-cache`.
+- Memory: the export runs one chunk at a time (several GB); keep other large jobs off the machine while it runs.

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.

MODIFICATIONS.md

@@ -0,0 +1,14 @@
+# Modification notice
+
+**ANEMLL has modified the weights of Qwen3.8-27B by quantizing them.** The files in `export/` are derivatives of [Qwen/Qwen3.8-27B at `1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0`](https://huggingface.co/Qwen/Qwen3.8-27B/tree/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0), whose weights carry Copyright 2026 Alibaba Cloud and the Apache-2.0 license preserved in [LICENSE](LICENSE).
+
+## Modified files
+
+- **`export/mix25in_mixr_lr64mix/layer_NN.safetensors`, `layer_NN_mixer.safetensors`, `lm_head.safetensors`:** the text model's linear weights replaced by quantized representations: GPTQ lookup-table indices and codebooks (vector 2x16 and scalar LUT4), per-channel scales, per-channel INT8 for the full-attention K / V projections, online Hadamard rotation seeds for the MLP basis, and rank-64 FP16 residual factors for the token-mixer projections. Recipe: [docs/QUANTIZATION.md](https://github.com/Anemll/anemll-forge/blob/main/docs/QUANTIZATION.md).
+
+## Unmodified files
+
+- **`model/small.safetensors`:** 449 small tensors (layer norms, DeltaNet conv1d, `A_log`, `dt_bias`, `in_proj_a` / `in_proj_b`, attention q / k norms, final norm) copied byte for byte in BF16 from the upstream checkpoint, with a new `model.safetensors.index.json` listing only them.
+- **`model/` configuration, tokenizer and chat template, root `config.json`, `LICENSE`:** copies of the upstream files.
+
+Vision weights, multi-token prediction and the embedding table are not included. Retain this notice, [NOTICE](NOTICE) and [LICENSE](LICENSE) when redistributing.

NOTICE

@@ -0,0 +1,23 @@
+ANEMLL quantized export of Qwen3.8-27B for Core AI conversion
+
+Original model: Qwen/Qwen3.8-27B
+Developed by the Qwen Team.
+Upstream weights: Copyright 2026 Alibaba Cloud.
+Reviewed revision: 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
+https://huggingface.co/Qwen/Qwen3.8-27B
+The upstream license is preserved unchanged in LICENSE.
+
+ANEMLL independently quantized the weights (mixed-precision GPTQ lookup tables,
+per-channel scales, online rotations, rank-64 residual corrections) and copied
+the small tensors, configuration and tokenizer files the converter needs without
+changes. See MODIFICATIONS.md and inventory.json.
+
+This NOTICE was added by ANEMLL; the reviewed upstream revision had no NOTICE
+file. It does not replace, amend, or add restrictions to the Apache License 2.0.
+
+Qwen, Alibaba Cloud and Apple are identified for credit and context. No official
+affiliation or endorsement is claimed.
+
+The model-weight license and the ANEMLL Forge source-code license are separate:
+ANEMLL Forge source code and documentation are MIT-licensed; these model-derived
+assets remain under the upstream Apache-2.0 license.

QWEN_SOURCE.json

@@ -0,0 +1,100 @@
+{
+  "upstream_model": {
+    "id": "Qwen/Qwen3.8-27B",
+    "revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+    "url": "https://huggingface.co/Qwen/Qwen3.8-27B/tree/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+    "license": "apache-2.0",
+    "license_url": "https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/LICENSE",
+    "copyright_notice": "Copyright 2026 Alibaba Cloud",
+    "notice_found": false
+  },
+  "checked_at": "2026-09-29",
+  "verification": {
+    "small_files": "Actual local bytes match pinned upstream Git-blob/LFS hashes.",
+    "cache_revision": "All 32 local Hugging Face download metadata files name this revision.",
+    "checkpoint_shards": "All 18 shard cache etags match upstream LFS SHA256 values; large shard contents were not rehashed in this review.",
+    "export_linkage": "The staged quantized/Core AI release artifacts still need their export and embedding provenance verified against this source."
+  },
+  "files": [
+    {
+      "path": "LICENSE",
+      "bytes": 11544,
+      "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a",
+      "upstream_hash": "f938136e3adacfd92be087f6e113b5d6d97f678f",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "README.md",
+      "bytes": 65012,
+      "sha256": "57e4bdb258ee1a7d2635c5174ebd4e56abe392505cdb5f8bbb356b0dc4293641",
+      "upstream_hash": "bc8aa0e396cd029c21cd773cca21830d0ded28ec",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "config.json",
+      "bytes": 4312,
+      "sha256": "191e0af232104ed8b65258cf3fb2b842e288008baca7633c11b82a1ac7203aab",
+      "upstream_hash": "706cebd746c4b6f2b1d1f892630867acfdfd3df8",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "tokenizer.json",
+      "bytes": 12809320,
+      "sha256": "0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3",
+      "upstream_hash": "0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3",
+      "upstream_hash_type": "sha256",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "tokenizer_config.json",
+      "bytes": 17928,
+      "sha256": "b11349aafa7cdc6a320767cf7ceb29ed82f7eda5d65e8e0819e76f0ce947bf27",
+      "upstream_hash": "5de744b3fca2129d7186979ae47c06be33903243",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "chat_template.jinja",
+      "bytes": 8952,
+      "sha256": "c3cf9e34abf4f9e36c2d72165aa9c132d3e2a725b6c2586aaa3a8af9d7a81041",
+      "upstream_hash": "c0c686f9c38d70d179fb7b5f5aa7530bc913dda3",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "generation_config.json",
+      "bytes": 202,
+      "sha256": "e70c136c1b78ddc1fb0905bac8e733a4dc448d4f852a5dd75143fffc70be550e",
+      "upstream_hash": "023756cfadf88e5bf69eefeee3e172f38c448d64",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "vocab.json",
+      "bytes": 6722759,
+      "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003",
+      "upstream_hash": "0aa0ce0658d60ac4a5d609f4eadb0e8e43514176",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "merges.txt",
+      "bytes": 3353259,
+      "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d",
+      "upstream_hash": "a494e019ca1502219fd0128658b979e5f05ae8e8",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    },
+    {
+      "path": "model.safetensors.index.json",
+      "bytes": 112216,
+      "sha256": "77042094076611b69791a610065f28b7013b8c621795fa86ddccc8bac7d1b9df",
+      "upstream_hash": "da35e3c564457dface7d138f0b6cac284ff8958c",
+      "upstream_hash_type": "git-blob-sha1",
+      "verified_against_upstream": true
+    }
+  ]
+}

README.md

@@ -0,0 +1,58 @@
+---
+base_model:
+  - Qwen/Qwen3.8-27B
+base_model_relation: quantized
+license: apache-2.0
+pipeline_tag: text-generation
+tags:
+  - coreai
+  - apple-neural-engine
+  - ane
+  - quantized
+  - gptq
+  - vector-quantization
+  - lut
+---
+
+# ANEMLL · Qwen3.8-27B, quantized for Core AI export
+
+**The quantized weights behind [ANEMLL Forge](https://github.com/Anemll/anemll-forge)'s Qwen3.8-27B for the Apple Neural Engine.** This repository holds the quantized export `mix25in_mixr_lr64mix` and the small part of the original checkpoint the converter needs, so the published Core AI packages ([anemll/anemll-forge-qwen3.8-27B](https://huggingface.co/anemll/anemll-forge-qwen3.8-27B)) can be rebuilt, checked or varied without the 52 GB original checkpoint. It is a research artifact for conversion: the files hold lookup-table indices, codebooks and scales in ANEMLL's own layout, not a checkpoint that transformers or other runtimes can load directly.
+
+To run the model, use the prepared Core AI bundle: [anemll/anemll-forge-qwen3.8-27B](https://huggingface.co/anemll/anemll-forge-qwen3.8-27B) and the [ANEMLL Forge README](https://github.com/Anemll/anemll-forge). To rebuild those packages from this repository, follow [EXPORT.md](EXPORT.md).
+
+## Contents
+
+| Path | What it is |
+| --- | --- |
+| `export/mix25in_mixr_lr64mix/layer_NN.safetensors` (64) | each layer's MLP (gate, up, down): LUT indices, codebooks, per-channel scales, online-rotation seeds in the header |
+| `export/mix25in_mixr_lr64mix/layer_NN_mixer.safetensors` (64) | each layer's token mixer: Gated DeltaNet projections (48 layers) or full-attention q / k / v / o (layers 3, 7, ..., 63), with rank-64 FP16 residual corrections |
+| `export/mix25in_mixr_lr64mix/lm_head.safetensors` | the output head (LUT4, per-output-channel scales) |
+| `model/` | the original `config.json`, tokenizer and chat template, and `small.safetensors`: the 449 small BF16 tensors the converter also reads (norms, DeltaNet conv1d, `A_log`, `dt_bias`, `in_proj_a` / `in_proj_b`, q / k norms, final norm), copied unchanged from the original checkpoint, with a `model.safetensors.index.json` |
+| `weights_digest.json` | SHA-256 of all 2,097 weight arrays the converter builds from this repository, to verify a download ([EXPORT.md](EXPORT.md)) |
+| `inventory.json` | size and SHA-256 of every file |
+| `config.json` | a copy of `model/config.json` at the root, for Hub discovery and download counting |
+| `LICENSE`, `NOTICE`, `MODIFICATIONS.md`, `QWEN_SOURCE.json` | the upstream license, attribution, modification notice and the pinned upstream source |
+
+## Quantization recipe
+
+Mixed precision with GPTQ, per weight matrix ([recipe and math](https://github.com/Anemll/anemll-forge/blob/main/docs/QUANTIZATION.md)):
+
+- **MLP:** `vector 2x16 + pcs` (vector lookup table: pairs of weights, 16 codebook entries, about 2 bits per weight, per-channel scaling) in 38 layers and `LUT4 per-tensor + pcs` in 26 layers, with online Hadamard rotations of the MLP basis.
+- **Token mixers:** vector LUTs in layers 0 to 23 and scalar LUT4 in layers 24 to 63; full-attention K / V projection weights use per-channel INT8. Rank-64 FP16 residual factors correct the DeltaNet and attention projections.
+- **Output head:** scalar LUT4 with per-output-channel scaling. Embeddings stay FP16 and are looked up on the host (the inference bundle's `model/embed_tokens_fp16.npy`).
+
+"2-bit model" is an incomplete description: activations, recurrent state and the KV cache stay in floating point (the Core AI graph adds an INT8 value cache and 8-bit attention arithmetic of its own).
+
+## Quality
+
+Teacher-forced KL divergence against the BF16 model, on a short trace (64 sequences, 40,023 positions), for the Core AI target built from this export: mean KL **0.184**, top-1 agreement **86.0%**, perplexity 2.414 against the teacher's 2.136. This is not a coding, reasoning, retrieval or long-context evaluation.
+
+## Reproducibility
+
+Converting this repository with the published settings gives the published packages' weights exactly: the converter consumes the same 2,097 arrays from this repository as from the original checkpoint (`weights_digest.json`). The package bytes still differ slightly from build to build, because the Core AI converter's serialization is not byte-deterministic. [EXPORT.md](EXPORT.md) lists the environment, the commands and the checks.
+
+## Attribution and licenses
+
+The original model is [Qwen/Qwen3.8-27B](https://huggingface.co/Qwen/Qwen3.8-27B) at revision [`1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0`](https://huggingface.co/Qwen/Qwen3.8-27B/tree/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0), developed by the **Qwen Team**; its weights carry **Copyright 2026 Alibaba Cloud** and the **Apache License 2.0**, preserved unchanged in [LICENSE](LICENSE). ANEMLL independently quantized these weights; see [MODIFICATIONS.md](MODIFICATIONS.md) and [NOTICE](NOTICE). Qwen, Alibaba Cloud and Apple are named for credit and context; no affiliation or endorsement is claimed.
+
+The weights remain under Apache-2.0. The [ANEMLL Forge](https://github.com/Anemll/anemll-forge) source code that produces and converts them is MIT-licensed. Retain LICENSE, NOTICE and MODIFICATIONS.md when redistributing.

config.json

@@ -0,0 +1,140 @@
+{
+    "architectures": [
+        "Qwen3_5ForConditionalGeneration"
+    ],
+    "image_token_id": 248056,
+    "language_model_only": false,
+    "model_type": "qwen3_5",
+    "text_config": {
+        "attention_bias": false,
+        "attention_dropout": 0.0,
+        "attn_output_gate": true,
+        "bos_token_id": 248044,
+        "dtype": "bfloat16",
+        "eos_token_id": 248044,
+        "full_attention_interval": 4,
+        "head_dim": 256,
+        "hidden_act": "silu",
+        "hidden_size": 5120,
+        "initializer_range": 0.02,
+        "intermediate_size": 17408,
+        "layer_types": [
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention"
+        ],
+        "linear_conv_kernel_dim": 4,
+        "linear_key_head_dim": 128,
+        "linear_num_key_heads": 16,
+        "linear_num_value_heads": 48,
+        "linear_value_head_dim": 128,
+        "mamba_ssm_dtype": "float32",
+        "max_position_embeddings": 262144,
+        "model_type": "qwen3_5_text",
+        "mtp_num_hidden_layers": 1,
+        "mtp_use_dedicated_embeddings": false,
+        "num_attention_heads": 24,
+        "num_hidden_layers": 64,
+        "num_key_value_heads": 4,
+        "output_gate_type": "swish",
+        "pad_token_id": null,
+        "partial_rotary_factor": 0.25,
+        "rms_norm_eps": 1e-06,
+        "rope_parameters": {
+            "mrope_interleaved": true,
+            "mrope_section": [
+                11,
+                11,
+                10
+            ],
+            "partial_rotary_factor": 0.25,
+            "rope_theta": 10000000,
+            "rope_type": "default"
+        },
+        "tie_word_embeddings": false,
+        "use_cache": true,
+        "vocab_size": 248320
+    },
+    "tie_word_embeddings": false,
+    "transformers_version": "5.8.0.dev0",
+    "video_token_id": 248057,
+    "vision_config": {
+        "deepstack_visual_indexes": [],
+        "depth": 27,
+        "hidden_act": "gelu_pytorch_tanh",
+        "hidden_size": 1152,
+        "in_channels": 3,
+        "initializer_range": 0.02,
+        "intermediate_size": 4304,
+        "model_type": "qwen3_5",
+        "num_heads": 16,
+        "num_position_embeddings": 2304,
+        "out_hidden_size": 5120,
+        "patch_size": 16,
+        "spatial_merge_size": 2,
+        "temporal_patch_size": 2
+    },
+    "vision_end_token_id": 248054,
+    "vision_start_token_id": 248053
+}

inventory.json

@@ -0,0 +1,739 @@
+{
+ "upstream_model": {
+  "id": "Qwen/Qwen3.8-27B",
+  "revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0"
+ },
+ "export": "mix25in_mixr_lr64mix",
+ "files": [
+  {
+   "path": "EXPORT.md",
+   "bytes": 3627,
+   "sha256": "5a72d0c1492b9f82e34b8113ea4e4f529317e6cbaac20b3a6c54b105a24aca79"
+  },
+  {
+   "path": "LICENSE",
+   "bytes": 11544,
+   "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a"
+  },
+  {
+   "path": "MODIFICATIONS.md",
+   "bytes": 1559,
+   "sha256": "f57bd42a4990e52fe8fb3c6b8758d406492b8d9beadebc0753c12507d2014d82"
+  },
+  {
+   "path": "NOTICE",
+   "bytes": 1115,
+   "sha256": "1cd63866fb82163cb72acc6e05d303f0ea569571ecf79d71976caba8833d42e4"
+  },
+  {
+   "path": "QWEN_SOURCE.json",
+   "bytes": 4052,
+   "sha256": "c207cff4ed6e5ede11629fce86f3a0292a83250e7976c1d501a17b5e8b3eb0f3"
+  },
+  {
+   "path": "README.md",
+   "bytes": 5226,
+   "sha256": "47855907887d983ae0412f65ad32f3ad536c8ce88fa513db26f7d8a411e2af21"
+  },
+  {
+   "path": "config.json",
+   "bytes": 4312,
+   "sha256": "191e0af232104ed8b65258cf3fb2b842e288008baca7633c11b82a1ac7203aab"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_00.safetensors",
+   "bytes": 133774360,
+   "sha256": "2efb9fd4ee68b117ea71d4ba8dd98900bd7547322e625547975a13f9bd68d313"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_00_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "5af75ce7035156f0320263d273839ab2889326b55b1460a5b195d8784f9520be"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_01.safetensors",
+   "bytes": 133774360,
+   "sha256": "be0877281b5d061d355791ea47c59907df0545623df10aad8b375bfe979c78b6"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_01_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "b16641848679b49a923253dff75ae09fc35974c6f99ea7f289b17e1c8291022f"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_02.safetensors",
+   "bytes": 133774360,
+   "sha256": "54abbb18e720b8c9b22e938e08b75dba8b22aae01041c6adcfe3c122a0a990e2"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_02_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "cf34407f1192c1e9a20c42e21e1195ccbcb08bf1dfba91cbdae1bd28e0831394"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_03.safetensors",
+   "bytes": 133774360,
+   "sha256": "fe77ed2e0af315af3b6873260de9fcfb45e3095c29afd00bde4995b92f5ee678"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_03_mixer.safetensors",
+   "bytes": 62955336,
+   "sha256": "d8dde2385f68e13ec3e96d420999e239b86aa7713c1239710ebb1f8ef1a0e35f"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_04.safetensors",
+   "bytes": 133774360,
+   "sha256": "a89627eacfc297c60f4d57ab275258badb132fcdd12aac9822522fba7d9c90b9"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_04_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "3d6474f816353afa27a2d114a89536d82728aced51ba8a0a58a76ea586a4b8a2"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_05.safetensors",
+   "bytes": 133774360,
+   "sha256": "614232d5415d0a27f7aae325cd74a11517e289be6b3a980b13e23fafa9e07fd7"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_05_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "b7c827996ab648a470897a63db3b1d5ffac905b631b7deab907284e33b00be7d"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_06.safetensors",
+   "bytes": 133774360,
+   "sha256": "aa837caac4aff114bc2c0cbae7264d793964dd0a8b2d86d930cccf0f8f08d60c"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_06_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "0ae2a6160cff3dbfb5c49724a16670e9d97c6431e45f1add5a83e663e20190c8"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_07.safetensors",
+   "bytes": 133774360,
+   "sha256": "6eb9820bb4d7b4f0cf30533a7b68642b5f940a2aae82eaabb1fcac2574c414ed"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_07_mixer.safetensors",
+   "bytes": 62955336,
+   "sha256": "baf847be8f772c82537dd4f01a35ede8f056d7c5d6499782bb30967bce196367"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_08.safetensors",
+   "bytes": 133774360,
+   "sha256": "dc78d10e793d00ff1bddb19810b626e88f18fbf2de0318148860eea504e17cef"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_08_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "c3a15dfa2b721a02f0518d7e80a7e7bacea4bd309c5ecd0dfe47f7a5fd2275b7"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_09.safetensors",
+   "bytes": 133774360,
+   "sha256": "c68cc3d21165c8205cbd3c0495c2557f5f11bc3f975544ed19ec0e9ddedab9eb"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_09_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "956661758896458eb390dc16ca165260953260237cc79e1b022ed208c0a3e4eb"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_10.safetensors",
+   "bytes": 133774360,
+   "sha256": "5ac57bf6edf6e3a76d17c601d321f42b7577491a54f188a4646af7afde8d2a32"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_10_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "acaa2fa3489c8cf982728c55d23c40a093aab1c1d4dcc4a94cd8534a745aa3ff"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_11.safetensors",
+   "bytes": 133774360,
+   "sha256": "c599c1c0bed65a53e3be587cd102f31c26712f34a0c79858ca9346307d89e185"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_11_mixer.safetensors",
+   "bytes": 62955336,
+   "sha256": "514a3662a38285155cb4bc163a562cecdaf335a4ca155e07aca08a0b5b3bf045"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_12.safetensors",
+   "bytes": 133774360,
+   "sha256": "f0852ec1c06212f5b89cd6f32dfce6fc935cd3365af1147e562c4714482766fb"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_12_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "251a2a1aa8875fe3af8edbdc192e2c7d78cb2096a88572a5177e4888f268242d"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_13.safetensors",
+   "bytes": 133774360,
+   "sha256": "43bb656e14f5554397e4a0a9e9ec51d5cda16e542ce84c409fecb2f3133f49a9"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_13_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "2c531d31094a3af920afcfa39aeca4cc4b827a62df8369857e6141c04e8c5899"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_14.safetensors",
+   "bytes": 133774360,
+   "sha256": "c575932d1855317e4306b3ae48606a00cd46ad7ed44dceae5d9d1c625f952f31"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_14_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "fdda060c4ec4ad858592b135668d0711929523485960444337433d02c9d14497"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_15.safetensors",
+   "bytes": 133774360,
+   "sha256": "030c27da5c3258fdee4d4d5c237a7b52c921e3f0b65e382394eb632dd2c566a6"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_15_mixer.safetensors",
+   "bytes": 62955336,
+   "sha256": "82a18435db7451eec2df7520bf5fd9f78a9c2659359b9cc95d93c95dbbe170a0"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_16.safetensors",
+   "bytes": 133774360,
+   "sha256": "d89930360027097a353cc650af3e21a95d90d273d631cc0261f01b0762837b6e"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_16_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "5c96df595100fa534c522126960978e445f5e3f3f6e6a2bed17e8e1d00a56b5a"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_17.safetensors",
+   "bytes": 133774360,
+   "sha256": "3b6fc0ef32ec0257fda9b3f1b988a5a004e337fb14fede11ef0db06da3735c1a"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_17_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "26aec4e1b87f3e8e387409531c91233af4dda72adb54457ef7ee993d11b0a8d2"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_18.safetensors",
+   "bytes": 133774360,
+   "sha256": "7792089cb80889aaec8ed3a30a46a99f11b034ae6fdc82e835ce081a62ae5f18"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_18_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "e73eef50dfa47b7d2baa0dd1d6313264417ffc83651fe3c97e581ee63136120a"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_19.safetensors",
+   "bytes": 133774360,
+   "sha256": "5a3f0746e30151130adc1ecd34bcc44fbc977bfd76c7f8f3599a2ced5b879072"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_19_mixer.safetensors",
+   "bytes": 62955336,
+   "sha256": "bb6983f17c5cc3bdb4fabee0ca9febe2e4061cbabbb4ab041a97369fdf262332"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_20.safetensors",
+   "bytes": 133774360,
+   "sha256": "ba8b6fc13efe55917c0013d1b317df6b46988028ec2328328c9c29eca5b28a7b"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_20_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "b5a0092f99a8725e73bd5ff18e59825b1eef0a6b82df4093f98331735ec441e0"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_21.safetensors",
+   "bytes": 133774360,
+   "sha256": "de247534adb81baec88237d108202fb9db69fb44bccfd33cc3d0ce30d6f09f2c"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_21_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "f49ca8bfc042bd53cddc3d17c4a5313c632bae40f535f4c9d5a9205d0623ed98"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_22.safetensors",
+   "bytes": 133774360,
+   "sha256": "395d209de1a9dd6ed5fdad5b261cfa7fe7658daf39e5b660eb1c8ffa543893d7"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_22_mixer.safetensors",
+   "bytes": 62566088,
+   "sha256": "90414a711228621b37241e48b3f87c5cccf5192762729d673aef54a2ddba471a"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_23.safetensors",
+   "bytes": 267467720,
+   "sha256": "bcd6c0af5a1d054baf748063a0737c4c9bbef104efff4c99df3a09a1ceca4970"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_23_mixer.safetensors",
+   "bytes": 62955336,
+   "sha256": "dc9f408b384284984c0fbc8a4e4815bb3ae5caaec47fbac6517b767c9d00e28d"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_24.safetensors",
+   "bytes": 267467720,
+   "sha256": "4f5183ae6190afaea5012c168e1e555ae4453825ceaca55e9b65b7c9c88ac14e"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_24_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "10609dfa21fc9694921698437bb6a177552ace5b1d3be74d2d7ce8ab10b56f18"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_25.safetensors",
+   "bytes": 267467720,
+   "sha256": "e00027984cf445e93d4b93a9b63620a65e0c9e2d9a9ce806f702f3a7631e3917"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_25_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "e4200f1547c187e8d81d2c6af9a9d6c81dd41407c99e2c36ed00df8b015b6bb0"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_26.safetensors",
+   "bytes": 133774360,
+   "sha256": "21483256c191fd0d9babcabdc9fd915cf057626dfb3e793fe368516a1258b890"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_26_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "c1143b8b399754cff29a4ea8f19970c539b611371a6432908cd17f6f9834773d"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_27.safetensors",
+   "bytes": 267467720,
+   "sha256": "e83ded67e799484eb62d88f1c3e1020a3314b02efe8e72e2dd62291a3e370b47"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_27_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "99a7d80d364fd0a77b1003ec8f00cac1db81b017b56c0a4beaec04c187bf30b4"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_28.safetensors",
+   "bytes": 267467720,
+   "sha256": "baa2ca17bfa6e0c3437191a277aaa65e0d24c3952111567ab97b8475aa395219"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_28_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "5e54dfdf176b5bcf5f712e5fe91305bd0470ae49bb33b6331fde1b3816184ce2"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_29.safetensors",
+   "bytes": 133774360,
+   "sha256": "b6e9e9470bdec6b9300f9b14c1bbb349bf4efd291349ddfdbcfbd9e2fd7e1c66"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_29_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "4f78710b973b8751e2d94b6ac7ca95b0b53137bd43a90e71bce567488a7bd231"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_30.safetensors",
+   "bytes": 267467720,
+   "sha256": "0ff8042c7f522f82ced1b4c28f29ac0ffe12904925619bc8c65ff19a404e54e1"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_30_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "285206c32cf8bc934c19c31b9ec8670d5c1075a1bf883be4fc360d0160458832"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_31.safetensors",
+   "bytes": 133774360,
+   "sha256": "17ae0e66bd8d3337a0e5212251bab41b0a64da46230c2e6f755688b5c6d4ca68"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_31_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "adf152ba93c2bb99405238f794d36b29ccb4f962b0b3cf0c5dd3b26ac6e3fb16"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_32.safetensors",
+   "bytes": 267467720,
+   "sha256": "acd8418dea3a6367e220bcddd993e7c6ca4b66a91365439c8b73cd0439f808b8"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_32_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "6e6503d5095ac82496e51f7cecf4ab9f78f094460b9d1d4782227efaf29c0f01"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_33.safetensors",
+   "bytes": 133774360,
+   "sha256": "b5c74fb8f1e848033f4fc467e5c8668fa3890e69bcee082b65446d9234c91c48"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_33_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "4bf8bc9dbb61f3dc7ac2f04b68d722d40a0abc6fbab8300ebb725230e880190f"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_34.safetensors",
+   "bytes": 133774360,
+   "sha256": "ec39be45c674872103f4b5de9d45980352ea10e4710f58be3d4a627123a3b9ff"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_34_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "6c2056a63557b64651a1bbbb5b41961950ddb5d68a737b6faa031c492ed62dbd"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_35.safetensors",
+   "bytes": 267467720,
+   "sha256": "e3b71e070c835fbf1995a6ba89e6b999d5f9fa9f1a263e05c3eb137ee634c58b"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_35_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "afa89fc5d76c3e3e6391a17b607d62f3ed0daefe747be25b511e97e6c4f3e34c"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_36.safetensors",
+   "bytes": 267467720,
+   "sha256": "e78b2fe703c06dbeb00fc7a09fb5112bf824af4f847dbbe2480d082d59ed0e6b"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_36_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "f8ada43f389f09479288e02062a6d4cae3e3e30fc02593fe7245c1e73b896c4d"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_37.safetensors",
+   "bytes": 267467720,
+   "sha256": "7e60a92edd010cc2aa04476d557827982b44a5a9ec84a03bc442b9815c6a45a7"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_37_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "84d28cb6922988a6c00950690c5a5676ab328c0f6c132cef05c0fb2783412c9d"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_38.safetensors",
+   "bytes": 133774360,
+   "sha256": "f835f74289b45cd6d628d57c1630c0eeafea82b6bdf3170339c7380ad683c98c"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_38_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "6ec6c6fefe5e0f367b84d1538a808f1daa582892ead538b5fd0171c2bc4e4c15"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_39.safetensors",
+   "bytes": 267467720,
+   "sha256": "effc9c7689c7f34c7133609ec7e6c98d2b7e15e09326cfe564db70033871d855"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_39_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "af7cc1a7e2d5cc177d6409e0ed1b0ef141fd950aa45369375242dfa745242db3"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_40.safetensors",
+   "bytes": 267467720,
+   "sha256": "edde34250a43f87fd0aa8d023aad8494a09e9970520be029eb734965e500d9a3"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_40_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "178a7f0065d9a1e964ffbfa073c09f23af3d8cc6ff33956e1a0aad4fe0a08aec"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_41.safetensors",
+   "bytes": 267467720,
+   "sha256": "924bfa4a91e7d6b097d0270aa5e7ffa4a129d2d4a82b689584aaeb2c3eca059e"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_41_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "262cec6f3355a33ece2ecffd083b72ab37e5056d47173ad2f7e80cfff3b629df"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_42.safetensors",
+   "bytes": 267467720,
+   "sha256": "99644d2148c68cec5cf43c9be212fe420a0c887e9b40c625d84ff1996e87ec2e"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_42_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "d01b765ddda7e511ff6a39db6256f996947abb283fd8a29de53bb31a9f91d7fa"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_43.safetensors",
+   "bytes": 133774360,
+   "sha256": "fcb1c84e9b80e6a08f30c864c81027a603f4746da192e3342706422945be4c33"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_43_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "90d07e7df46af600fa70e26204ff48dcc911872f64ea68fc802f81f4ef711892"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_44.safetensors",
+   "bytes": 267467720,
+   "sha256": "c75c2f24b9497a60ca91d5dddd201a574014ede816545287df2490bae2722ef1"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_44_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "3448baff6717970702f90ccc6b83e1f95f2018c583fd9bed236f5519aad83919"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_45.safetensors",
+   "bytes": 267467720,
+   "sha256": "08062e4e83384062d2005806e0e274bde1f217c284819d611f9d2a3c5b9b2695"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_45_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "71d6dbb9d63bfb038d3a32792937111abd6a6b7becaf84e0deffb0a5fc602200"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_46.safetensors",
+   "bytes": 133774360,
+   "sha256": "a45d7a7f5aee4c98a5c366b444749f69a951db5889011550e3b95956fa448fcc"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_46_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "1f8abd454f20a0bb7b508aa496facf01dac838c0a39a2ee53bd1e99618c1e86e"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_47.safetensors",
+   "bytes": 133774360,
+   "sha256": "0133e8c0c97b59c820dc2d76300179ad0639d3d1b00caea4a03a40758b8e93c4"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_47_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "17041e013214eb98e00a08a554ac8f7c6aff95ee3d2ad055808995ced6a29559"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_48.safetensors",
+   "bytes": 133774360,
+   "sha256": "b35dfb98eb718630719c217ca63ca9a8be15ef5fa9ae09ed7375fc73c6de7214"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_48_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "a5f9df48722d0eef7330dbb9cfbdc01df7cdc5973625a47b117442daf19190e4"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_49.safetensors",
+   "bytes": 267467720,
+   "sha256": "a68eaf063770407be5d96a97ba2a0ce7d3006bccff90c16b7e3c3eabea78fc71"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_49_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "3a71d162d586920e5ed447b4c3c86c3c4130515fee89f4af2e93f2e409f79430"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_50.safetensors",
+   "bytes": 133774360,
+   "sha256": "ca809b5583b510484c4ac7e3207f961a777ec36c2ce457b5eb7bffe9b5a0ef1f"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_50_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "d26e9e70c31b2d72c398faf2fd4014897f025ce5f756da1ede625fbe949adfc6"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_51.safetensors",
+   "bytes": 267467720,
+   "sha256": "fe8739bc133e878029ba5445372bf71d6de8a1b028d6bc721bced7dd7c416a1b"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_51_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "2f2bd253705e55901fe6e77e2b8d8a0353f790aba33373b03bafbd5588058ba0"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_52.safetensors",
+   "bytes": 133774360,
+   "sha256": "cff474dbc4c9a1dcbcc92affede49426321aa6120c0af7ea1e7df571e684c334"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_52_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "d24ba4e971fff681ad468632198e8fd5e9002b2412503f44778c590d6844e2ae"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_53.safetensors",
+   "bytes": 267467720,
+   "sha256": "89486521886f807e91865f613fb69c29eb242d23ed2ee94d346a5ddedcf4b559"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_53_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "b376fc3912923d09c8fb75e85fc62af263d36aba37ccc6734339d1eb6e5f6d54"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_54.safetensors",
+   "bytes": 267467720,
+   "sha256": "405d22e5c93bd7f4becdfc2a945649de463e12f73ed7ff7f990557a2d32f7e37"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_54_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "e4795c04f0fb6835d2d0fdd71595884418974c3db1c3560068d4d15626c974d7"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_55.safetensors",
+   "bytes": 267467720,
+   "sha256": "e7d98250e0f8a55a06538141d79190cd9259d294c1997cd4dea4edd628ea5d13"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_55_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "a5f49764a908906143084f25de09f8cedba95b6203a882a2674c3a0ddeb3d27a"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_56.safetensors",
+   "bytes": 133774360,
+   "sha256": "404fa4fe7dbcc5df4b9a4a3d98aff4fd75e0c9bb4909d3f5233ab510fd774421"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_56_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "91543234a4c8890ffd824746277665c916bbffbf1f6c0cbef908cd8530f4991f"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_57.safetensors",
+   "bytes": 133774360,
+   "sha256": "8d9fd84c74afce86a69387b3bc845d2a12b8fc8758ed6091723c382fe13e1eb5"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_57_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "ec9baa9770cc7ffc1b810f0fd4e94313f06e971b36e0826f25eddd2fb97d9f43"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_58.safetensors",
+   "bytes": 267467720,
+   "sha256": "f3945e9548aa57c3781b0b0b508c0c9755cf7b9d6b72ea79d331d11b7f417729"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_58_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "53f2469ee0f727b218461d5b07508cb757b1b325f0339d4090dcef8989f6bd4a"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_59.safetensors",
+   "bytes": 133774360,
+   "sha256": "894366876e67ed809fa8cfaaeb8651d5c7391fab44f95cb76515bec7397933bd"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_59_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "f6c6b79ce01b89bd2f3d9c3396a8da10d86583d56301aa7d034963816e2dd924"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_60.safetensors",
+   "bytes": 267467720,
+   "sha256": "a8f4fd929f6440a1eaf1796692b78c5adfc72c9ba7da90a593a5f31c1b4135ac"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_60_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "fe7542fa204f7aebe557d98c090958dd526db8a003e618d0e01b7fe207864840"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_61.safetensors",
+   "bytes": 267467720,
+   "sha256": "b666bf728d8fa5a6c02f2d49da612a7609ca3ef68c0575b5da04a3184ec80500"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_61_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "ad9632247049dc6cf7aef03072bb06e2fc4cec01344b7ececa43098d4a4ca5e8"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_62.safetensors",
+   "bytes": 267467720,
+   "sha256": "2098cdb75c4d442152305ef7d6bc5e62dd7ca952423f59b6b143e99c38332125"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_62_mixer.safetensors",
+   "bytes": 120237680,
+   "sha256": "69e525a6aca43ec4658971daf20633d0337eb095d6268c326b6329f231e6de15"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_63.safetensors",
+   "bytes": 267467720,
+   "sha256": "0cfa0e5974cb4ab017d0ce95e75c0f277125045bc5ae6f0ebe41c21c435baf0e"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/layer_63_mixer.safetensors",
+   "bytes": 110141200,
+   "sha256": "5275e9d65e3a78d103331d6483ac47925dacb2b402e82f11d0efc8d0c98d333c"
+  },
+  {
+   "path": "export/mix25in_mixr_lr64mix/lm_head.safetensors",
+   "bytes": 1271895360,
+   "sha256": "55f923bb4de720d032de117b414784e1191de387583f77ebd2786d37d16903dc"
+  },
+  {
+   "path": "model/chat_template.jinja",
+   "bytes": 8952,
+   "sha256": "c3cf9e34abf4f9e36c2d72165aa9c132d3e2a725b6c2586aaa3a8af9d7a81041"
+  },
+  {
+   "path": "model/config.json",
+   "bytes": 4312,
+   "sha256": "191e0af232104ed8b65258cf3fb2b842e288008baca7633c11b82a1ac7203aab"
+  },
+  {
+   "path": "model/generation_config.json",
+   "bytes": 202,
+   "sha256": "e70c136c1b78ddc1fb0905bac8e733a4dc448d4f852a5dd75143fffc70be550e"
+  },
+  {
+   "path": "model/merges.txt",
+   "bytes": 3353259,
+   "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d"
+  },
+  {
+   "path": "model/model.safetensors.index.json",
+   "bytes": 37789,
+   "sha256": "7f99ba4fff4aaa60ca0cdc7f5e076750258fb97f0f017a7b376b6181b0c63a14"
+  },
+  {
+   "path": "model/small.safetensors",
+   "bytes": 52533208,
+   "sha256": "d9bc791922fd0732cdf9480a501a31bb52bff5e68bb02f0fb34c85567d3559ef"
+  },
+  {
+   "path": "model/tokenizer.json",
+   "bytes": 12809320,
+   "sha256": "0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3"
+  },
+  {
+   "path": "model/tokenizer_config.json",
+   "bytes": 17928,
+   "sha256": "b11349aafa7cdc6a320767cf7ceb29ed82f7eda5d65e8e0819e76f0ce947bf27"
+  },
+  {
+   "path": "model/vocab.json",
+   "bytes": 6722759,
+   "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003"
+  },
+  {
+   "path": "weights_digest.json",
+   "bytes": 248562,
+   "sha256": "517edc20d805e662f2d7dae7f7a569c0becedfaaf14c9e318656bda22e96a4a0"
+  }
+ ]
+}

model/chat_template.jinja

@@ -0,0 +1,170 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- set reasoning_instructions = '' %}
+{%- if enable_thinking is undefined or enable_thinking is true %}
+    {%- set resolved_reasoning_effort = reasoning_effort|default('xhigh') %}
+    {%- if resolved_reasoning_effort not in ('xhigh', 'medium', 'low') %}
+        {{- raise_exception('Unexpected reasoning effort ' ~ reasoning_effort ~ '. Supported types are xhigh (default), medium, and low.') }}
+    {%- endif %}
+    {%- if resolved_reasoning_effort == 'xhigh' %}
+        {%- set reasoning_instructions = 'Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer.' %}
+    {%- elif resolved_reasoning_effort == 'low' %}
+        {%- set reasoning_instructions = 'Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration.' %}
+    {%- endif %}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {%- if reasoning_instructions %}
+        {{- reasoning_instructions + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '')  + content + '<|im_end|>\n' }}
+        {%- elif reasoning_instructions %}
+            {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' }}
+        {%- endif %}
+    {%- elif reasoning_instructions %}
+        {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined and tool_call.arguments != '' %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

model/config.json

@@ -0,0 +1,140 @@
+{
+    "architectures": [
+        "Qwen3_5ForConditionalGeneration"
+    ],
+    "image_token_id": 248056,
+    "language_model_only": false,
+    "model_type": "qwen3_5",
+    "text_config": {
+        "attention_bias": false,
+        "attention_dropout": 0.0,
+        "attn_output_gate": true,
+        "bos_token_id": 248044,
+        "dtype": "bfloat16",
+        "eos_token_id": 248044,
+        "full_attention_interval": 4,
+        "head_dim": 256,
+        "hidden_act": "silu",
+        "hidden_size": 5120,
+        "initializer_range": 0.02,
+        "intermediate_size": 17408,
+        "layer_types": [
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention",
+            "linear_attention",
+            "linear_attention",
+            "linear_attention",
+            "full_attention"
+        ],
+        "linear_conv_kernel_dim": 4,
+        "linear_key_head_dim": 128,
+        "linear_num_key_heads": 16,
+        "linear_num_value_heads": 48,
+        "linear_value_head_dim": 128,
+        "mamba_ssm_dtype": "float32",
+        "max_position_embeddings": 262144,
+        "model_type": "qwen3_5_text",
+        "mtp_num_hidden_layers": 1,
+        "mtp_use_dedicated_embeddings": false,
+        "num_attention_heads": 24,
+        "num_hidden_layers": 64,
+        "num_key_value_heads": 4,
+        "output_gate_type": "swish",
+        "pad_token_id": null,
+        "partial_rotary_factor": 0.25,
+        "rms_norm_eps": 1e-06,
+        "rope_parameters": {
+            "mrope_interleaved": true,
+            "mrope_section": [
+                11,
+                11,
+                10
+            ],
+            "partial_rotary_factor": 0.25,
+            "rope_theta": 10000000,
+            "rope_type": "default"
+        },
+        "tie_word_embeddings": false,
+        "use_cache": true,
+        "vocab_size": 248320
+    },
+    "tie_word_embeddings": false,
+    "transformers_version": "5.8.0.dev0",
+    "video_token_id": 248057,
+    "vision_config": {
+        "deepstack_visual_indexes": [],
+        "depth": 27,
+        "hidden_act": "gelu_pytorch_tanh",
+        "hidden_size": 1152,
+        "in_channels": 3,
+        "initializer_range": 0.02,
+        "intermediate_size": 4304,
+        "model_type": "qwen3_5",
+        "num_heads": 16,
+        "num_position_embeddings": 2304,
+        "out_hidden_size": 5120,
+        "patch_size": 16,
+        "spatial_merge_size": 2,
+        "temporal_patch_size": 2
+    },
+    "vision_end_token_id": 248054,
+    "vision_start_token_id": 248053
+}

model/generation_config.json

@@ -0,0 +1,12 @@
+{
+    "bos_token_id": 248044,
+    "do_sample": true,
+    "eos_token_id": [
+        248046,
+        248044
+    ],
+    "pad_token_id": 248044,
+    "temperature": 1.0,
+    "top_k": 20,
+    "top_p": 0.95
+}

model/model.safetensors.index.json

@@ -0,0 +1,456 @@
+{
+  "metadata": {
+    "total_size": 52476928
+  },
+  "weight_map": {
+    "model.language_model.layers.0.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.0.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.0.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.0.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.0.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.0.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.0.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.0.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.1.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.1.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.1.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.1.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.1.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.1.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.1.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.1.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.10.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.10.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.10.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.10.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.10.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.10.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.10.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.10.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.11.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.11.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.11.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.11.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.12.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.12.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.12.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.12.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.12.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.12.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.12.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.12.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.13.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.13.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.13.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.13.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.13.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.13.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.13.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.13.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.14.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.14.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.14.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.14.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.14.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.14.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.14.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.14.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.15.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.15.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.15.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.15.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.16.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.16.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.16.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.16.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.16.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.16.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.16.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.16.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.17.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.17.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.17.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.17.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.17.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.17.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.17.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.17.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.18.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.18.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.18.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.18.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.18.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.18.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.18.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.18.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.19.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.19.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.19.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.19.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.2.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.2.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.2.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.2.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.2.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.2.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.2.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.2.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.20.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.20.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.20.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.20.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.20.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.20.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.20.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.20.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.21.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.21.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.21.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.21.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.21.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.21.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.21.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.21.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.22.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.22.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.22.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.22.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.22.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.22.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.22.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.22.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.23.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.23.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.23.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.23.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.24.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.24.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.24.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.24.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.24.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.24.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.24.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.24.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.25.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.25.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.25.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.25.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.25.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.25.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.25.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.25.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.26.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.26.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.26.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.26.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.26.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.26.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.26.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.26.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.27.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.27.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.27.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.27.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.28.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.28.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.28.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.28.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.28.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.28.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.28.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.28.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.29.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.29.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.29.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.29.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.29.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.29.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.29.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.29.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.3.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.3.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.3.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.3.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.30.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.30.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.30.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.30.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.30.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.30.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.30.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.30.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.31.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.31.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.31.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.31.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.32.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.32.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.32.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.32.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.32.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.32.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.32.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.32.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.33.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.33.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.33.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.33.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.33.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.33.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.33.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.33.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.34.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.34.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.34.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.34.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.34.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.34.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.34.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.34.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.35.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.35.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.35.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.35.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.36.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.36.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.36.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.36.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.36.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.36.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.36.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.36.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.37.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.37.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.37.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.37.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.37.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.37.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.37.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.37.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.38.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.38.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.38.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.38.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.38.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.38.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.38.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.38.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.39.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.39.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.39.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.39.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.4.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.4.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.4.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.4.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.4.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.4.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.4.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.4.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.40.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.40.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.40.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.40.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.40.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.40.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.40.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.40.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.41.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.41.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.41.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.41.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.41.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.41.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.41.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.41.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.42.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.42.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.42.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.42.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.42.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.42.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.42.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.42.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.43.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.43.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.43.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.43.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.44.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.44.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.44.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.44.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.44.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.44.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.44.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.44.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.45.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.45.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.45.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.45.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.45.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.45.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.45.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.45.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.46.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.46.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.46.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.46.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.46.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.46.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.46.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.46.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.47.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.47.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.47.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.47.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.48.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.48.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.48.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.48.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.48.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.48.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.48.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.48.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.49.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.49.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.49.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.49.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.49.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.49.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.49.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.49.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.5.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.5.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.5.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.5.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.5.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.5.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.5.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.5.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.50.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.50.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.50.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.50.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.50.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.50.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.50.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.50.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.51.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.51.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.51.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.51.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.52.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.52.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.52.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.52.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.52.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.52.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.52.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.52.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.53.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.53.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.53.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.53.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.53.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.53.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.53.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.53.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.54.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.54.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.54.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.54.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.54.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.54.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.54.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.54.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.55.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.55.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.55.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.55.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.56.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.56.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.56.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.56.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.56.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.56.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.56.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.56.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.57.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.57.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.57.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.57.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.57.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.57.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.57.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.57.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.58.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.58.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.58.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.58.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.58.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.58.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.58.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.58.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.59.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.59.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.59.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.59.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.6.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.6.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.6.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.6.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.6.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.6.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.6.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.6.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.60.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.60.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.60.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.60.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.60.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.60.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.60.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.60.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.61.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.61.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.61.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.61.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.61.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.61.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.61.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.61.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.62.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.62.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.62.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.62.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.62.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.62.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.62.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.62.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.63.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.63.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.63.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.63.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.7.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.7.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.7.self_attn.k_norm.weight": "small.safetensors",
+    "model.language_model.layers.7.self_attn.q_norm.weight": "small.safetensors",
+    "model.language_model.layers.8.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.8.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.8.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.8.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.8.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.8.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.8.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.8.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.9.input_layernorm.weight": "small.safetensors",
+    "model.language_model.layers.9.linear_attn.A_log": "small.safetensors",
+    "model.language_model.layers.9.linear_attn.conv1d.weight": "small.safetensors",
+    "model.language_model.layers.9.linear_attn.dt_bias": "small.safetensors",
+    "model.language_model.layers.9.linear_attn.in_proj_a.weight": "small.safetensors",
+    "model.language_model.layers.9.linear_attn.in_proj_b.weight": "small.safetensors",
+    "model.language_model.layers.9.linear_attn.norm.weight": "small.safetensors",
+    "model.language_model.layers.9.post_attention_layernorm.weight": "small.safetensors",
+    "model.language_model.norm.weight": "small.safetensors"
+  }
+}

model/small.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9bc791922fd0732cdf9480a501a31bb52bff5e68bb02f0fb34c85567d3559ef
+size 52533208

model/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3
+size 12809320

model/tokenizer_config.json

@@ -0,0 +1,305 @@
+{
+    "add_prefix_space": false,
+    "added_tokens_decoder": {
+        "248044": {
+            "content": "<|endoftext|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248045": {
+            "content": "<|im_start|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248046": {
+            "content": "<|im_end|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248047": {
+            "content": "<|object_ref_start|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248048": {
+            "content": "<|object_ref_end|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248049": {
+            "content": "<|box_start|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248050": {
+            "content": "<|box_end|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248051": {
+            "content": "<|quad_start|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248052": {
+            "content": "<|quad_end|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248053": {
+            "content": "<|vision_start|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248054": {
+            "content": "<|vision_end|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248055": {
+            "content": "<|vision_pad|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248056": {
+            "content": "<|image_pad|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248057": {
+            "content": "<|video_pad|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248058": {
+            "content": "<tool_call>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248059": {
+            "content": "</tool_call>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248060": {
+            "content": "<|fim_prefix|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248061": {
+            "content": "<|fim_middle|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248062": {
+            "content": "<|fim_suffix|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248063": {
+            "content": "<|fim_pad|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248064": {
+            "content": "<|repo_name|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248065": {
+            "content": "<|file_sep|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248066": {
+            "content": "<tool_response>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248067": {
+            "content": "</tool_response>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248068": {
+            "content": "<think>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248069": {
+            "content": "</think>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": false
+        },
+        "248070": {
+            "content": "<|audio_start|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248071": {
+            "content": "<|audio_end|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248072": {
+            "content": "<tts_pad>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248073": {
+            "content": "<tts_text_bos>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248074": {
+            "content": "<tts_text_eod>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248075": {
+            "content": "<tts_text_bos_single>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        },
+        "248076": {
+            "content": "<|audio_pad|>",
+            "lstrip": false,
+            "normalized": false,
+            "rstrip": false,
+            "single_word": false,
+            "special": true
+        }
+    },
+    "additional_special_tokens": [
+        "<|im_start|>",
+        "<|im_end|>",
+        "<|object_ref_start|>",
+        "<|object_ref_end|>",
+        "<|box_start|>",
+        "<|box_end|>",
+        "<|quad_start|>",
+        "<|quad_end|>",
+        "<|vision_start|>",
+        "<|vision_end|>",
+        "<|vision_pad|>",
+        "<|image_pad|>",
+        "<|video_pad|>"
+    ],
+    "bos_token": null,
+    "chat_template": "{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- macro render_content(content, do_vision_count, is_system_content=false) %}\n    {%- if content is string %}\n        {{- content }}\n    {%- elif content is iterable and content is not mapping %}\n        {%- for item in content %}\n            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}\n                {%- if is_system_content %}\n                    {{- raise_exception('System message cannot contain images.') }}\n                {%- endif %}\n                {%- if do_vision_count %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                {%- endif %}\n                {%- if add_vision_id %}\n                    {{- 'Picture ' ~ image_count.value ~ ': ' }}\n                {%- endif %}\n                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}\n            {%- elif 'video' in item or item.type == 'video' %}\n                {%- if is_system_content %}\n                    {{- raise_exception('System message cannot contain videos.') }}\n                {%- endif %}\n                {%- if do_vision_count %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                {%- endif %}\n                {%- if add_vision_id %}\n                    {{- 'Video ' ~ video_count.value ~ ': ' }}\n                {%- endif %}\n                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}\n            {%- elif 'text' in item %}\n                {{- item.text }}\n            {%- else %}\n                {{- raise_exception('Unexpected item type in content.') }}\n            {%- endif %}\n        {%- endfor %}\n    {%- elif content is none or content is undefined %}\n        {{- '' }}\n    {%- else %}\n        {{- raise_exception('Unexpected content type.') }}\n    {%- endif %}\n{%- endmacro %}\n{%- if not messages %}\n    {{- raise_exception('No messages provided.') }}\n{%- endif %}\n{%- set reasoning_instructions = '' %}\n{%- if enable_thinking is undefined or enable_thinking is true %}\n    {%- set resolved_reasoning_effort = reasoning_effort|default('xhigh') %}\n    {%- if resolved_reasoning_effort not in ('xhigh', 'medium', 'low') %}\n        {{- raise_exception('Unexpected reasoning effort ' ~ reasoning_effort ~ '. Supported types are xhigh (default), medium, and low.') }}\n    {%- endif %}\n    {%- if resolved_reasoning_effort == 'xhigh' %}\n        {%- set reasoning_instructions = 'Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer.' %}\n    {%- elif resolved_reasoning_effort == 'low' %}\n        {%- set reasoning_instructions = 'Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration.' %}\n    {%- endif %}\n{%- endif %}\n{%- if tools and tools is iterable and tools is not mapping %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if reasoning_instructions %}\n        {{- reasoning_instructions + '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou have access to the following functions:\\n\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\" }}\n    {{- '\\n\\nIf you choose to call a function ONLY reply in the following format with NO suffix:\\n\\n<tool_call>\\n<function=example_function_name>\\n<parameter=example_parameter_1>\\nvalue_1\\n</parameter>\\n<parameter=example_parameter_2>\\nThis is the value for the second parameter\\nthat can span\\nmultiple lines\\n</parameter>\\n</function>\\n</tool_call>\\n\\n<IMPORTANT>\\nReminder:\\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\\n- Required parameters MUST be specified\\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\\n</IMPORTANT>' }}\n    {%- if messages[0].role == 'system' %}\n        {%- set content = render_content(messages[0].content, false, true)|trim %}\n        {%- if content %}\n            {{- '\\n\\n' + content }}\n        {%- endif %}\n    {%- endif %}\n    {{- '<|im_end|>\\n' }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {%- set content = render_content(messages[0].content, false, true)|trim %}\n        {%- if content %}\n            {{- '<|im_start|>system\\n' + (reasoning_instructions + '\\n\\n' if reasoning_instructions else '')  + content + '<|im_end|>\\n' }}\n        {%- elif reasoning_instructions %}\n            {{- '<|im_start|>system\\n' + reasoning_instructions + '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- elif reasoning_instructions %}\n        {{- '<|im_start|>system\\n' + reasoning_instructions + '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n    {%- set index = (messages|length - 1) - loop.index0 %}\n    {%- if ns.multi_step_tool and message.role == \"user\" %}\n        {%- set content = render_content(message.content, false)|trim %}\n        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}\n            {%- set ns.multi_step_tool = false %}\n            {%- set ns.last_query_index = index %}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if ns.multi_step_tool %}\n    {{- raise_exception('No user query found in messages.') }}\n{%- endif %}\n{%- for message in messages %}\n    {%- set content = render_content(message.content, true)|trim %}\n    {%- if message.role == \"system\" %}\n        {%- if not loop.first %}\n            {{- raise_exception('System message must be at the beginning.') }}\n        {%- endif %}\n    {%- elif message.role == \"user\" %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- endif %}\n        {%- set reasoning_content = reasoning_content|trim %}\n        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index %}\n            {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content + '\\n</think>\\n\\n' + content }}\n        {%- else %}\n            {{- '<|im_start|>' + message.role + '\\n' + content }}\n        {%- endif %}\n        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if tool_call.function is defined %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {%- if loop.first %}\n                    {%- if content|trim %}\n                        {{- '\\n\\n<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n                    {%- else %}\n                        {{- '<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n                    {%- endif %}\n                {%- else %}\n                    {{- '\\n<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n                {%- endif %}\n                {%- if tool_call.arguments is defined and tool_call.arguments != '' %}\n                    {%- for args_name, args_value in tool_call.arguments|items %}\n                        {{- '<parameter=' + args_name + '>\\n' }}\n                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}\n                        {{- args_value }}\n                        {{- '\\n</parameter>\\n' }}\n                    {%- endfor %}\n                {%- endif %}\n                {{- '</function>\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.previtem and loop.previtem.role != \"tool\" %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- content }}\n        {{- '\\n</tool_response>' }}\n        {%- if not loop.last and loop.nextitem.role != \"tool\" %}\n            {{- '<|im_end|>\\n' }}\n        {%- elif loop.last %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- else %}\n        {{- raise_exception('Unexpected message role.') }}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n    {%- if enable_thinking is defined and enable_thinking is false %}\n        {{- '<think>\\n\\n</think>\\n\\n' }}\n    {%- else %}\n        {{- '<think>\\n' }}\n    {%- endif %}\n{%- endif %}",
+    "clean_up_tokenization_spaces": false,
+    "eos_token": "<|im_end|>",
+    "errors": "replace",
+    "model_max_length": 262144,
+    "pad_token": "<|endoftext|>",
+    "split_special_tokens": false,
+    "tokenizer_class": "Qwen2Tokenizer",
+    "unk_token": null,
+    "add_bos_token": false,
+    "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+    "extra_special_tokens": {
+        "audio_bos_token": "<|audio_start|>",
+        "audio_eos_token": "<|audio_end|>",
+        "audio_token": "<|audio_pad|>",
+        "image_token": "<|image_pad|>",
+        "video_token": "<|video_pad|>",
+        "vision_bos_token": "<|vision_start|>",
+        "vision_eos_token": "<|vision_end|>"
+    }
+}