chiakelvin Adamiros commited on
Commit
683d42d
·
0 Parent(s):

Duplicate from bottlecapai/ThinkingCap-Qwen3.6-27B

Browse files

Co-authored-by: Adam Jirkovský <Adamiros@users.noreply.huggingface.co>

.eval_results/gpqa.yaml ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ - dataset:
2
+ id: Idavidrein/gpqa
3
+ task_id: diamond
4
+ value: 83.8
5
+ source:
6
+ url: https://huggingface.co/bottlecapai/ThinkingCap-Qwen3.6-27B
7
+ name: Model Card
.eval_results/gsm8k.yaml ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ - dataset:
2
+ id: openai/gsm8k
3
+ task_id: gsm8k
4
+ value: 96.5
5
+ source:
6
+ url: https://huggingface.co/bottlecapai/ThinkingCap-Qwen3.6-27B
7
+ name: Model Card
.eval_results/mmlu-pro.yaml ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ - dataset:
2
+ id: TIGER-Lab/MMLU-Pro
3
+ task_id: mmlu_pro
4
+ value: 85.4
5
+ source:
6
+ url: https://huggingface.co/bottlecapai/ThinkingCap-Qwen3.6-27B
7
+ name: Model Card
.gitattributes ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
37
+ bc_capybara.png filter=lfs diff=lfs merge=lfs -text
38
+ thinkingcap_6x.gif filter=lfs diff=lfs merge=lfs -text
39
+ cap_header.png filter=lfs diff=lfs merge=lfs -text
LICENSE ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ Apache License
3
+ Version 2.0, January 2004
4
+ http://www.apache.org/licenses/
5
+
6
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
7
+
8
+ 1. Definitions.
9
+
10
+ "License" shall mean the terms and conditions for use, reproduction,
11
+ and distribution as defined by Sections 1 through 9 of this document.
12
+
13
+ "Licensor" shall mean the copyright owner or entity authorized by
14
+ the copyright owner that is granting the License.
15
+
16
+ "Legal Entity" shall mean the union of the acting entity and all
17
+ other entities that control, are controlled by, or are under common
18
+ control with that entity. For the purposes of this definition,
19
+ "control" means (i) the power, direct or indirect, to cause the
20
+ direction or management of such entity, whether by contract or
21
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
22
+ outstanding shares, or (iii) beneficial ownership of such entity.
23
+
24
+ "You" (or "Your") shall mean an individual or Legal Entity
25
+ exercising permissions granted by this License.
26
+
27
+ "Source" form shall mean the preferred form for making modifications,
28
+ including but not limited to software source code, documentation
29
+ source, and configuration files.
30
+
31
+ "Object" form shall mean any form resulting from mechanical
32
+ transformation or translation of a Source form, including but
33
+ not limited to compiled object code, generated documentation,
34
+ and conversions to other media types.
35
+
36
+ "Work" shall mean the work of authorship, whether in Source or
37
+ Object form, made available under the License, as indicated by a
38
+ copyright notice that is included in or attached to the work
39
+ (an example is provided in the Appendix below).
40
+
41
+ "Derivative Works" shall mean any work, whether in Source or Object
42
+ form, that is based on (or derived from) the Work and for which the
43
+ editorial revisions, annotations, elaborations, or other modifications
44
+ represent, as a whole, an original work of authorship. For the purposes
45
+ of this License, Derivative Works shall not include works that remain
46
+ separable from, or merely link (or bind by name) to the interfaces of,
47
+ the Work and Derivative Works thereof.
48
+
49
+ "Contribution" shall mean any work of authorship, including
50
+ the original version of the Work and any modifications or additions
51
+ to that Work or Derivative Works thereof, that is intentionally
52
+ submitted to Licensor for inclusion in the Work by the copyright owner
53
+ or by an individual or Legal Entity authorized to submit on behalf of
54
+ the copyright owner. For the purposes of this definition, "submitted"
55
+ means any form of electronic, verbal, or written communication sent
56
+ to the Licensor or its representatives, including but not limited to
57
+ communication on electronic mailing lists, source code control systems,
58
+ and issue tracking systems that are managed by, or on behalf of, the
59
+ Licensor for the purpose of discussing and improving the Work, but
60
+ excluding communication that is conspicuously marked or otherwise
61
+ designated in writing by the copyright owner as "Not a Contribution."
62
+
63
+ "Contributor" shall mean Licensor and any individual or Legal Entity
64
+ on behalf of whom a Contribution has been received by Licensor and
65
+ subsequently incorporated within the Work.
66
+
67
+ 2. Grant of Copyright License. Subject to the terms and conditions of
68
+ this License, each Contributor hereby grants to You a perpetual,
69
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
70
+ copyright license to reproduce, prepare Derivative Works of,
71
+ publicly display, publicly perform, sublicense, and distribute the
72
+ Work and such Derivative Works in Source or Object form.
73
+
74
+ 3. Grant of Patent License. Subject to the terms and conditions of
75
+ this License, each Contributor hereby grants to You a perpetual,
76
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
77
+ (except as stated in this section) patent license to make, have made,
78
+ use, offer to sell, sell, import, and otherwise transfer the Work,
79
+ where such license applies only to those patent claims licensable
80
+ by such Contributor that are necessarily infringed by their
81
+ Contribution(s) alone or by combination of their Contribution(s)
82
+ with the Work to which such Contribution(s) was submitted. If You
83
+ institute patent litigation against any entity (including a
84
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
85
+ or a Contribution incorporated within the Work constitutes direct
86
+ or contributory patent infringement, then any patent licenses
87
+ granted to You under this License for that Work shall terminate
88
+ as of the date such litigation is filed.
89
+
90
+ 4. Redistribution. You may reproduce and distribute copies of the
91
+ Work or Derivative Works thereof in any medium, with or without
92
+ modifications, and in Source or Object form, provided that You
93
+ meet the following conditions:
94
+
95
+ (a) You must give any other recipients of the Work or
96
+ Derivative Works a copy of this License; and
97
+
98
+ (b) You must cause any modified files to carry prominent notices
99
+ stating that You changed the files; and
100
+
101
+ (c) You must retain, in the Source form of any Derivative Works
102
+ that You distribute, all copyright, patent, trademark, and
103
+ attribution notices from the Source form of the Work,
104
+ excluding those notices that do not pertain to any part of
105
+ the Derivative Works; and
106
+
107
+ (d) If the Work includes a "NOTICE" text file as part of its
108
+ distribution, then any Derivative Works that You distribute must
109
+ include a readable copy of the attribution notices contained
110
+ within such NOTICE file, excluding those notices that do not
111
+ pertain to any part of the Derivative Works, in at least one
112
+ of the following places: within a NOTICE text file distributed
113
+ as part of the Derivative Works; within the Source form or
114
+ documentation, if provided along with the Derivative Works; or,
115
+ within a display generated by the Derivative Works, if and
116
+ wherever such third-party notices normally appear. The contents
117
+ of the NOTICE file are for informational purposes only and
118
+ do not modify the License. You may add Your own attribution
119
+ notices within Derivative Works that You distribute, alongside
120
+ or as an addendum to the NOTICE text from the Work, provided
121
+ that such additional attribution notices cannot be construed
122
+ as modifying the License.
123
+
124
+ You may add Your own copyright statement to Your modifications and
125
+ may provide additional or different license terms and conditions
126
+ for use, reproduction, or distribution of Your modifications, or
127
+ for any such Derivative Works as a whole, provided Your use,
128
+ reproduction, and distribution of the Work otherwise complies with
129
+ the conditions stated in this License.
130
+
131
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
132
+ any Contribution intentionally submitted for inclusion in the Work
133
+ by You to the Licensor shall be under the terms and conditions of
134
+ this License, without any additional terms or conditions.
135
+ Notwithstanding the above, nothing herein shall supersede or modify
136
+ the terms of any separate license agreement you may have executed
137
+ with Licensor regarding such Contributions.
138
+
139
+ 6. Trademarks. This License does not grant permission to use the trade
140
+ names, trademarks, service marks, or product names of the Licensor,
141
+ except as required for reasonable and customary use in describing the
142
+ origin of the Work and reproducing the content of the NOTICE file.
143
+
144
+ 7. Disclaimer of Warranty. Unless required by applicable law or
145
+ agreed to in writing, Licensor provides the Work (and each
146
+ Contributor provides its Contributions) on an "AS IS" BASIS,
147
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
148
+ implied, including, without limitation, any warranties or conditions
149
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
150
+ PARTICULAR PURPOSE. You are solely responsible for determining the
151
+ appropriateness of using or redistributing the Work and assume any
152
+ risks associated with Your exercise of permissions under this License.
153
+
154
+ 8. Limitation of Liability. In no event and under no legal theory,
155
+ whether in tort (including negligence), contract, or otherwise,
156
+ unless required by applicable law (such as deliberate and grossly
157
+ negligent acts) or agreed to in writing, shall any Contributor be
158
+ liable to You for damages, including any direct, indirect, special,
159
+ incidental, or consequential damages of any character arising as a
160
+ result of this License or out of the use or inability to use the
161
+ Work (including but not limited to damages for loss of goodwill,
162
+ work stoppage, computer failure or malfunction, or any and all
163
+ other commercial damages or losses), even if such Contributor
164
+ has been advised of the possibility of such damages.
165
+
166
+ 9. Accepting Warranty or Additional Liability. While redistributing
167
+ the Work or Derivative Works thereof, You may choose to offer,
168
+ and charge a fee for, acceptance of support, warranty, indemnity,
169
+ or other liability obligations and/or rights consistent with this
170
+ License. However, in accepting such obligations, You may act only
171
+ on Your own behalf and on Your sole responsibility, not on behalf
172
+ of any other Contributor, and only if You agree to indemnify,
173
+ defend, and hold each Contributor harmless for any liability
174
+ incurred by, or claims asserted against, such Contributor by reason
175
+ of your accepting any such warranty or additional liability.
176
+
177
+ END OF TERMS AND CONDITIONS
178
+
179
+ APPENDIX: How to apply the Apache License to your work.
180
+
181
+ To apply the Apache License to your work, attach the following
182
+ boilerplate notice, with the fields enclosed by brackets "[]"
183
+ replaced with your own identifying information. (Don't include
184
+ the brackets!) The text should be enclosed in the appropriate
185
+ comment syntax for the file format. We also recommend that a
186
+ file or class name and description of purpose be included on the
187
+ same "printed page" as the copyright notice for easier
188
+ identification within third-party archives.
189
+
190
+ Copyright 2026 Alibaba Cloud
191
+
192
+ Licensed under the Apache License, Version 2.0 (the "License");
193
+ you may not use this file except in compliance with the License.
194
+ You may obtain a copy of the License at
195
+
196
+ http://www.apache.org/licenses/LICENSE-2.0
197
+
198
+ Unless required by applicable law or agreed to in writing, software
199
+ distributed under the License is distributed on an "AS IS" BASIS,
200
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
201
+ See the License for the specific language governing permissions and
202
+ limitations under the License.
README.md ADDED
@@ -0,0 +1,174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: "Qwen/Qwen3.6-27B"
3
+ base_model_relation: finetune
4
+ library_name: transformers
5
+ tags:
6
+ - qwen3_6
7
+ - token-efficient
8
+ - efficient-thinking
9
+ license: apache-2.0
10
+ ---
11
+
12
+ <a href="https://www.bottlecapai.com/"><img src="cap_header.png" alt="ThinkingCap — BottleCap AI" width="100%"></a>
13
+
14
+ # ThinkingCap: Qwen 3.6 27B
15
+
16
+ Capability of Qwen3.6-27B with **50% less** thinking tokens on average, and over **90% less** in best cases.
17
+ Achieved via finetuning [Qwen3.6-27B (Qwen Team, 2026)](https://huggingface.co/Qwen/Qwen3.6-27B) with state-of-the-art algorithms on a curated set of problems of various domains and difficulty. We designed the finetuning to be as minimally invasive as possible, preserving all of the original answer quality and style of Qwen, while being more token efficient. Check [the blogpost](https://www.bottlecapai.com/thinkingcap-qwen3-6-27b) for more details.
18
+
19
+ We rigorously evaluate the resulting checkpoint across general reasoning, non-reasoning multiple-choice question answering, everyday multi-turn conversations, system prompt adherence, safety, math, code and agentic use cases. Due to the high variability of reasoning quality at Qwen-recommended sampling temperature 1.0, we run each benchmark with multiple seeds and do statistical significance testing on all the results. We evaluate both in domain (holdout parts of selected datasets included in training) and out of domain.
20
+
21
+ <p align="center"><img src="thinkingcap_6x.gif" alt="ThinkingCap reasoning demo (6× speed)" width="480" style="display:block;margin:0 auto"></p>
22
+
23
+ ## Out-of-domain token efficiency
24
+
25
+ <table width="100%" style="border-collapse:collapse;width:100%;display:table;table-layout:auto;font-size:14px;line-height:1.4;">
26
+ <thead>
27
+ <tr><th rowspan="2" style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);vertical-align:bottom;color:#6b748a;font-weight:600;">Benchmark</th><th colspan="2" style="text-align:center;padding:6px 12px 3px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;">Accuracy</th><th colspan="3" style="text-align:center;padding:6px 12px 3px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;">Thinking tokens</th></tr>
28
+ <tr><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Base</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Ours</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Base</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Ours</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Reduction</th></tr>
29
+ </thead>
30
+ <tbody>
31
+ <tr><td colspan="6" style="text-align:left;padding:7px 12px;background:rgba(199,185,255,0.16);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;border-bottom:1px solid rgba(128,128,128,0.22);">Knowledge &amp; reasoning</td></tr>
32
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">GPQA-Diamond</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">85.5<span style="color:#6b748a;font-size:11px;"> ±1.4</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">83.8<span style="color:#6b748a;font-size:11px;"> ±1.9</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">10,777</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">3,351</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 67.8%</td></tr>
33
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">SuperGPQA</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">64.0<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">64.0<span style="color:#6b748a;font-size:11px;"> ±0.1</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">8,246</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">3,384</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 58.4%</td></tr>
34
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">MMLU-Pro</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">85.9<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">85.4<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">3,455</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,290</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 53.7%</td></tr>
35
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">MMLU-Redux</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">93.9<span style="color:#6b748a;font-size:11px;"> ±0.1</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">93.9<span style="color:#6b748a;font-size:11px;"> ±0.1</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">947</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">406</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 44.8%</td></tr>
36
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">C-Eval</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">90.6<span style="color:#6b748a;font-size:11px;"> ±0.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">90.3<span style="color:#6b748a;font-size:11px;"> ±0.6</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,279</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">663</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 47.1%</td></tr>
37
+ <tr><td colspan="6" style="text-align:left;padding:7px 12px;background:rgba(199,185,255,0.16);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;border-bottom:1px solid rgba(128,128,128,0.22);">Math &amp; code</td></tr>
38
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">HMMT (Nov 2025)</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">88.0<span style="color:#6b748a;font-size:11px;"> ±3.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">84.7<span style="color:#6b748a;font-size:11px;"> ±3.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">39,277</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">27,388</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 38.0%</td></tr>
39
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">LiveCodeBench</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">80.7<span style="color:#6b748a;font-size:11px;"> ±0.6</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">84.3<span style="color:#6b748a;font-size:11px;"> ±1.0</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">15,744</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">10,158</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 41.1%</td></tr>
40
+ <tr><td colspan="6" style="text-align:left;padding:7px 12px;background:rgba(199,185,255,0.16);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;border-bottom:1px solid rgba(128,128,128,0.22);">Long-context &amp; multimodal</td></tr>
41
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">LongBench v2</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">62.6<span style="color:#6b748a;font-size:11px;"> ±3.6</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">60.2<span style="color:#6b748a;font-size:11px;"> ±1.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,765</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,091</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 39.1%</td></tr>
42
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">RealWorldQA</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">82.4<span style="color:#6b748a;font-size:11px;"> ±0.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">81.9<span style="color:#6b748a;font-size:11px;"> ±1.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">2,959</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">913</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 48.5%</td></tr>
43
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">AA-LCR</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">76.2<span style="color:#6b748a;font-size:11px;"> ±3.0</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">74.2<span style="color:#6b748a;font-size:11px;"> ±2.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">2,455</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,337</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 45.5%</td></tr>
44
+ <tr><td colspan="6" style="text-align:left;padding:7px 12px;background:rgba(199,185,255,0.16);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;border-bottom:1px solid rgba(128,128,128,0.22);">Instruction following &amp; agentic</td></tr>
45
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">System-prompt adherence</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">80.6<span style="color:#6b748a;font-size:11px;"> ±1.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">81.5<span style="color:#6b748a;font-size:11px;"> ±1.8</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,737</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">976</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 40.0%</td></tr>
46
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">Claw-Eval <span style="color:#6b748a;font-size:11px;">think/task</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">87.0<span style="color:#6b748a;font-size:11px;"> ±1.9</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">84.4<span style="color:#6b748a;font-size:11px;"> ±1.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">919</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">689</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 25.2%</td></tr>
47
+ <tr><td style="text-align:left;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;">Macro average</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;">81.5</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;font-variant-numeric:tabular-nums;">80.7</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;border-left:1px solid rgba(128,128,128,0.22);">—</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;">—</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;color:#0a9d68;font-variant-numeric:tabular-nums;">↓ 45.8%</td></tr>
48
+ </tbody>
49
+ </table>
50
+
51
+ <sub>Claw-Eval thinking tokens are per-task (agentic; not a single-turn trace).</sub>
52
+
53
+ **Settings**
54
+
55
+ - **Models:** base `Qwen/Qwen3.6-27B` vs `bottlecapai/ThinkingCap-Qwen3.6-27B` (shown as `Ours` in the table).
56
+ - **Seeds:** 5 per condition; thinking on; cells are mean ± 95% CI across seeds.
57
+ - **Decoding:** thinking on; sampling `temperature=1.0, top_p=0.95, top_k=20, min_p=0.0` (`bottlecapai/ThinkingCap-Qwen3.6-27B` uses the base's sampling).
58
+ - **Max generation tokens:** 100,000 for the general suite (gpqa_diamond, mmlu_pro, longbench_v2, realworldqa) and AA-LCR; 250,000 for HMMT (Nov 2025); 32,768 for supergpqa and livecodebench; 16,384 for ceval and mmlu_redux; 15,000 for llm-system-prompts-benchmark; 49,152 for Claw-Eval.
59
+ - **Metrics** — the columns mirror the table:
60
+ - **Accuracy** (Base / Ours) — fraction correct (exact/regex match; soft compliance for llm-system-prompts-benchmark; judge task-score for Claw-Eval; judge CORRECT/INCORRECT for AA-LCR).
61
+ - **Thinking tokens** (Base / Ours) — mean length of the single-turn `<think>` trace (think-per-task for Claw-Eval).
62
+ - **Reduction** — the average per-question thinking-token saving: base and `Ours` are paired on the same question (each side seed-averaged), each question's `(base − cap)/base` is taken, then averaged over shared questions (a larger ↓ = a bigger saving).
63
+ - **Macro average** (bottom row) — equal-weight mean across benchmarks.
64
+
65
+ We separately track two trace-quality failure modes, reported only in aggregate: **looping** — the model gets stuck repeating the same reasoning chain (sometimes a single sentence), never finishing its thinking; detected from the fraction of repetitive n-grams — and **truncation** — the `<think>` trace never closes because the model hits the generation-token cap while still reasoning, so no answer is produced. Across all out-of-domain responses, truncation drops from **2.9% to 0.4%** while looping stays negligible (**~0.2%**).
66
+
67
+ ## In-domain evals
68
+
69
+ Holdout **test** splits of datasets whose train splits are part of the finetuning mix — quality retention on in-distribution tasks (in contrast to the out-of-domain benchmarks above).
70
+
71
+ <table width="100%" style="border-collapse:collapse;width:100%;display:table;table-layout:auto;font-size:14px;line-height:1.4;">
72
+ <thead>
73
+ <tr><th rowspan="2" style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);vertical-align:bottom;color:#6b748a;font-weight:600;">Benchmark</th><th colspan="2" style="text-align:center;padding:6px 12px 3px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;">Accuracy</th><th colspan="3" style="text-align:center;padding:6px 12px 3px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;">Thinking tokens</th></tr>
74
+ <tr><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Base</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Ours</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Base</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Ours</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Reduction</th></tr>
75
+ </thead>
76
+ <tbody>
77
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">GSM8K</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">93.3<span style="color:#6b748a;font-size:11px;"> ±1.5</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">96.5<span style="color:#6b748a;font-size:11px;"> ±0.3</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">3,175</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">648</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 74.1%</td></tr>
78
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">ARC-Challenge</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">97.0<span style="color:#6b748a;font-size:11px;"> ±0.3</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">97.6<span style="color:#6b748a;font-size:11px;"> ±0.4</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">966</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">335</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 51.5%</td></tr>
79
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">ARC-Easy</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">99.3<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">99.4<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">566</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">260</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 44.5%</td></tr>
80
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">CommonsenseQA</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">86.7<span style="color:#6b748a;font-size:11px;"> ±0.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">88.2<span style="color:#6b748a;font-size:11px;"> ±0.9</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,118</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">273</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 64.1%</td></tr>
81
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">OpenBookQA</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">96.0<span style="color:#6b748a;font-size:11px;"> ±0.5</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">96.7<span style="color:#6b748a;font-size:11px;"> ±0.6</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">858</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">248</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 59.5%</td></tr>
82
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">QASC</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">91.7<span style="color:#6b748a;font-size:11px;"> ±0.7</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">92.2<span style="color:#6b748a;font-size:11px;"> ±0.5</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,258</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">348</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 61.9%</td></tr>
83
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">SciQ</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">97.0<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">97.5<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">766</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">276</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 48.3%</td></tr>
84
+ <tr><td style="text-align:left;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;">Macro average</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;">94.4</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;font-variant-numeric:tabular-nums;">95.4</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;border-left:1px solid rgba(128,128,128,0.22);">—</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;">—</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;color:#0a9d68;font-variant-numeric:tabular-nums;">↓ 57.7%</td></tr>
85
+ </tbody>
86
+ </table>
87
+
88
+ **Settings**
89
+
90
+ - **Seeds:** 5 per condition; thinking on; cells are mean ± 95% CI across seeds.
91
+ - **Decoding:** sampling `temperature=1.0, top_p=0.95, top_k=20, min_p=0.0` (`bottlecapai/ThinkingCap-Qwen3.6-27B` uses the base's sampling).
92
+ - **Max generation tokens:** 15,000 for GSM8K; 8,192 for the MCQ sets.
93
+ - **Data:** GSM8K is the full 1,319-row test split; the MCQ sets are capped at 1,000 rows (OpenBookQA = 500 and QASC = 926 are smaller, so full).
94
+ - **Metrics:** **Accuracy** — exact-match on the final answer (GSM8K) / last-letter multiple-choice match (MCQ). **Thinking tokens**, **Reduction** and **Macro average** are as defined for the token-efficiency table above, as are the **looping** and **truncation** failure modes: across all in-domain responses, truncation drops from **1.6% to 0.03%** while looping is negligible for both (**≤0.01%**).
95
+
96
+ ## Guardrails preservation
97
+
98
+ Brevity finetuning leaves safety behaviour intact: on both safety sets `bottlecapai/ThinkingCap-Qwen3.6-27B` refuses harmful/jailbreak prompts at the base model's rate (statistically indistinguishable) while still spending fewer thinking tokens.
99
+
100
+ <table width="100%" style="border-collapse:collapse;width:100%;display:table;table-layout:auto;font-size:14px;line-height:1.4;">
101
+ <thead>
102
+ <tr><th rowspan="2" style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);vertical-align:bottom;color:#6b748a;font-weight:600;">Benchmark</th><th colspan="2" style="text-align:center;padding:6px 12px 3px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;">SAFE %</th><th colspan="3" style="text-align:center;padding:6px 12px 3px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#7a67e6;font-weight:700;font-size:11px;letter-spacing:0.06em;text-transform:uppercase;">Thinking tokens</th></tr>
103
+ <tr><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Base</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Ours</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Base</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Ours</th><th style="text-align:center;padding:3px 12px 8px;border-bottom:1px solid rgba(128,128,128,0.22);color:#6b748a;font-weight:600;font-size:12px;">Reduction</th></tr>
104
+ </thead>
105
+ <tbody>
106
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">Nemotron-Safety</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">98.9<span style="color:#6b748a;font-size:11px;"> ±0.8</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">99.0<span style="color:#6b748a;font-size:11px;"> ±0.4</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">1,242</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">933</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 23.8%</td></tr>
107
+ <tr><td style="text-align:left;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;font-size:13px;white-space:nowrap;">HEx-PHI</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">99.9<span style="color:#6b748a;font-size:11px;"> ±0.2</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">100.0<span style="color:#6b748a;font-size:11px;"> ±0.0</span></td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">693</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;">543</td><td style="text-align:center;padding:8px 12px;border-bottom:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;white-space:nowrap;color:#0a9d68;font-weight:700;">↓ 20.0%</td></tr>
108
+ <tr><td style="text-align:left;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;">Macro average</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;border-left:1px solid rgba(128,128,128,0.22);font-variant-numeric:tabular-nums;">99.4</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;font-variant-numeric:tabular-nums;">99.5</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;border-left:1px solid rgba(128,128,128,0.22);">—</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;">—</td><td style="text-align:center;padding:9px 12px;border-top:2px solid rgba(128,128,128,0.22);background:rgba(199,185,255,0.10);font-weight:750;color:#0a9d68;font-variant-numeric:tabular-nums;">↓ 21.9%</td></tr>
109
+ </tbody>
110
+ </table>
111
+
112
+ **Settings**
113
+
114
+ - **Seeds:** 5 per condition; thinking on; cells are mean ± 95% CI across seeds.
115
+ - **Benchmarks:** `Nemotron-Safety` — a 300-prompt sample of the `safety` split of `nvidia/Llama-Nemotron-Post-Training-Dataset`; `HEx-PHI` — the 300 HEx-PHI harmful instructions (`jkazdan/guardrail-llama-3-8b-refusal-hexphi`). Both are safety-relevant / jailbreak prompts.
116
+
117
+ Both datasets only contain a train subset and given their near-perfect scores, it is hard to rule out possible contimination in Qwen's original training data. These results are intended purely as a demonstration of preserved guardrail capabilities, and are in line with our internal usage testing.
118
+
119
+ - **Metrics:** **SAFE %** — fraction of replies an LLM safety judge marks SAFE (upholds guardrails — refuses or safely deflects the harmful request); higher is better. **Thinking tokens** and **Reduction** are as defined for the token-efficiency table above.
120
+
121
+ ## Usage
122
+
123
+ ### HuggingFace Transformers
124
+
125
+ ```python
126
+ from transformers import AutoModelForImageTextToText, AutoProcessor
127
+ model = AutoModelForImageTextToText.from_pretrained("bottlecapai/ThinkingCap-Qwen3.6-27B", dtype="bfloat16")
128
+ proc = AutoProcessor.from_pretrained("bottlecapai/ThinkingCap-Qwen3.6-27B")
129
+ ```
130
+
131
+ Check https://huggingface.co/Qwen/Qwen3.6-27B for recommended usage, sampling params etc.
132
+
133
+ ### FP8 (vLLM / SGLang)
134
+
135
+ An official FP8 quantization for GPU serving lives in the sibling repo [bottlecapai/ThinkingCap-Qwen3.6-27B-FP8](https://huggingface.co/bottlecapai/ThinkingCap-Qwen3.6-27B-FP8) — half the memory of bf16 at near-lossless quality, in the compressed-tensors format vLLM/SGLang load natively, with the MTP (multi-token-prediction) speculative-decoding head kept in bf16.
136
+
137
+ ```bash
138
+ vllm serve bottlecapai/ThinkingCap-Qwen3.6-27B-FP8
139
+ ```
140
+
141
+ ### GGUF (llama.cpp)
142
+
143
+ Quantized [GGUF](https://github.com/ggml-org/ggml/blob/master/docs/gguf.md) builds of this model live in the sibling repo [bottlecapai/ThinkingCap-Qwen3.6-27B-GGUF](https://huggingface.co/bottlecapai/ThinkingCap-Qwen3.6-27B-GGUF), for local inference with [llama.cpp](https://github.com/ggml-org/llama.cpp) and compatible runtimes (Ollama, LM Studio, …).
144
+
145
+ Quantization stores the weights at reduced precision — e.g. ~4.7 bits per weight for `Q4_K_M` instead of 16-bit bf16 — cutting download size and memory severalfold at a small quality cost. `Q4_K_M` is the recommended size/quality balance, `Q8_0` is near-lossless.
146
+
147
+ ```bash
148
+ llama-cli -hf bottlecapai/ThinkingCap-Qwen3.6-27B-GGUF:Q4_K_M -p "Hi"
149
+ ```
150
+
151
+ ## Where to find us
152
+
153
+ <table style="border-collapse:collapse;border:0;margin:0"><tbody><tr>
154
+ <td style="border:0;padding:0 18px 0 0"><a href="https://www.bottlecapai.com/"><img src="social-web.png" alt="Website" width="34" height="34"></a></td>
155
+ <td style="border:0;padding:0 18px 0 0"><a href="https://www.linkedin.com/company/bottlecap-ai/"><img src="social-linkedin.png" alt="LinkedIn" width="34" height="34"></a></td>
156
+ <td style="border:0;padding:0 18px 0 0"><a href="https://www.instagram.com/bottlecapai/"><img src="social-instagram.png" alt="Instagram" width="34" height="34"></a></td>
157
+ <td style="border:0;padding:0"><a href="https://x.com/BottleCapAI"><img src="social-x.png" alt="X" width="34" height="34"></a></td>
158
+ </tr></tbody></table>
159
+
160
+ ## Citation
161
+
162
+ If you use this model, please cite:
163
+
164
+ ```bibtex
165
+ @misc{ThinkingCap-Qwen3.6-27B,
166
+ title = {bottlecapai/ThinkingCap-Qwen3.6-27B},
167
+ author = {Lasocki, Karol and Osusky, Adam and Lindauer, Jan and Jirkovsky, Adam and Mihal, Filip and Platek, Ondrej and Herel, David and Ihnatchenko, Luka and Bartek, Vojtech and Jirak, Jiri and Mikolov, Tomas},
168
+ year = {2026},
169
+ }
170
+ ```
171
+
172
+ ## Acknowledgements
173
+
174
+ We acknowledge EuroHPC Joint Undertaking for awarding the project ID EHPC-AIF-2025SC03-029 access to Leonardo at CINECA, Italy.
bc_capybara.png ADDED

Git LFS Details

  • SHA256: 4571093445d0e52feef28174f4c596502c0dcb546998a824236693f0640c8796
  • Pointer size: 132 Bytes
  • Size of remote file: 2.98 MB
bottlecap-logo.png ADDED
budget_aggregate.png ADDED
cap_header.png ADDED

Git LFS Details

  • SHA256: 82841e4c6cc4299fafb4eef8a7175e787924f9d7de3273d0c5eaa225059ea2a1
  • Pointer size: 132 Bytes
  • Size of remote file: 1.06 MB
chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\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>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if (preserve_thinking is defined and preserve_thinking is true) or (loop.index0 > ns.last_query_index) %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
config.json ADDED
@@ -0,0 +1,142 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "image_token_id": 248056,
7
+ "language_model_only": false,
8
+ "model_type": "qwen3_5",
9
+ "text_config": {
10
+ "attention_bias": false,
11
+ "attention_dropout": 0.0,
12
+ "attn_output_gate": true,
13
+ "bos_token_id": 248044,
14
+ "dtype": "bfloat16",
15
+ "eos_token_id": 248044,
16
+ "full_attention_interval": 4,
17
+ "head_dim": 256,
18
+ "hidden_act": "silu",
19
+ "hidden_size": 5120,
20
+ "initializer_range": 0.02,
21
+ "intermediate_size": 17408,
22
+ "layer_types": [
23
+ "linear_attention",
24
+ "linear_attention",
25
+ "linear_attention",
26
+ "full_attention",
27
+ "linear_attention",
28
+ "linear_attention",
29
+ "linear_attention",
30
+ "full_attention",
31
+ "linear_attention",
32
+ "linear_attention",
33
+ "linear_attention",
34
+ "full_attention",
35
+ "linear_attention",
36
+ "linear_attention",
37
+ "linear_attention",
38
+ "full_attention",
39
+ "linear_attention",
40
+ "linear_attention",
41
+ "linear_attention",
42
+ "full_attention",
43
+ "linear_attention",
44
+ "linear_attention",
45
+ "linear_attention",
46
+ "full_attention",
47
+ "linear_attention",
48
+ "linear_attention",
49
+ "linear_attention",
50
+ "full_attention",
51
+ "linear_attention",
52
+ "linear_attention",
53
+ "linear_attention",
54
+ "full_attention",
55
+ "linear_attention",
56
+ "linear_attention",
57
+ "linear_attention",
58
+ "full_attention",
59
+ "linear_attention",
60
+ "linear_attention",
61
+ "linear_attention",
62
+ "full_attention",
63
+ "linear_attention",
64
+ "linear_attention",
65
+ "linear_attention",
66
+ "full_attention",
67
+ "linear_attention",
68
+ "linear_attention",
69
+ "linear_attention",
70
+ "full_attention",
71
+ "linear_attention",
72
+ "linear_attention",
73
+ "linear_attention",
74
+ "full_attention",
75
+ "linear_attention",
76
+ "linear_attention",
77
+ "linear_attention",
78
+ "full_attention",
79
+ "linear_attention",
80
+ "linear_attention",
81
+ "linear_attention",
82
+ "full_attention",
83
+ "linear_attention",
84
+ "linear_attention",
85
+ "linear_attention",
86
+ "full_attention"
87
+ ],
88
+ "linear_conv_kernel_dim": 4,
89
+ "linear_key_head_dim": 128,
90
+ "linear_num_key_heads": 16,
91
+ "linear_num_value_heads": 48,
92
+ "linear_value_head_dim": 128,
93
+ "mamba_ssm_dtype": "float32",
94
+ "max_position_embeddings": 262144,
95
+ "model_type": "qwen3_5_text",
96
+ "mtp_num_hidden_layers": 1,
97
+ "mtp_use_dedicated_embeddings": false,
98
+ "num_attention_heads": 24,
99
+ "num_hidden_layers": 64,
100
+ "num_key_value_heads": 4,
101
+ "output_gate_type": "swish",
102
+ "pad_token_id": null,
103
+ "partial_rotary_factor": 0.25,
104
+ "rms_norm_eps": 1e-06,
105
+ "rope_parameters": {
106
+ "mrope_interleaved": true,
107
+ "mrope_section": [
108
+ 11,
109
+ 11,
110
+ 10
111
+ ],
112
+ "partial_rotary_factor": 0.25,
113
+ "rope_theta": 10000000,
114
+ "rope_type": "default"
115
+ },
116
+ "tie_word_embeddings": false,
117
+ "use_cache": true,
118
+ "vocab_size": 248320
119
+ },
120
+ "tie_word_embeddings": false,
121
+ "transformers_version": "5.7.0",
122
+ "video_token_id": 248057,
123
+ "vision_config": {
124
+ "deepstack_visual_indexes": [],
125
+ "depth": 27,
126
+ "dtype": "bfloat16",
127
+ "hidden_act": "gelu_pytorch_tanh",
128
+ "hidden_size": 1152,
129
+ "in_channels": 3,
130
+ "initializer_range": 0.02,
131
+ "intermediate_size": 4304,
132
+ "model_type": "qwen3_5_vision",
133
+ "num_heads": 16,
134
+ "num_position_embeddings": 2304,
135
+ "out_hidden_size": 5120,
136
+ "patch_size": 16,
137
+ "spatial_merge_size": 2,
138
+ "temporal_patch_size": 2
139
+ },
140
+ "vision_end_token_id": 248054,
141
+ "vision_start_token_id": 248053
142
+ }
configuration.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"framework":"Pytorch","task":"image-text-to-text"}
generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 248044,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 248046,
6
+ 248044
7
+ ],
8
+ "pad_token_id": 248044,
9
+ "temperature": 1.0,
10
+ "top_k": 20,
11
+ "top_p": 0.95,
12
+ "transformers_version": "5.7.0"
13
+ }
merge_verify_report.json ADDED
@@ -0,0 +1,2225 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "adapter_config": {
3
+ "alora_invocation_tokens": null,
4
+ "alpha_pattern": {},
5
+ "arrow_config": null,
6
+ "auto_mapping": null,
7
+ "base_model_name_or_path": "Qwen/Qwen3.6-27B",
8
+ "bias": "none",
9
+ "corda_config": null,
10
+ "ensure_weight_tying": false,
11
+ "eva_config": null,
12
+ "exclude_modules": null,
13
+ "fan_in_fan_out": false,
14
+ "inference_mode": true,
15
+ "init_lora_weights": true,
16
+ "layer_replication": null,
17
+ "layers_pattern": null,
18
+ "layers_to_transform": null,
19
+ "loftq_config": {},
20
+ "lora_alpha": 128,
21
+ "lora_bias": false,
22
+ "lora_dropout": 0.05,
23
+ "lora_ga_config": null,
24
+ "megatron_config": null,
25
+ "megatron_core": "megatron.core",
26
+ "modules_to_save": null,
27
+ "peft_type": "LORA",
28
+ "peft_version": "0.19.1",
29
+ "qalora_group_size": 16,
30
+ "r": 64,
31
+ "rank_pattern": {},
32
+ "revision": null,
33
+ "target_modules": "(?!.*visual).*\\.(q_proj|k_proj|v_proj|o_proj|gate_proj|up_proj|down_proj)",
34
+ "target_parameters": null,
35
+ "task_type": "CAUSAL_LM",
36
+ "trainable_token_indices": null,
37
+ "use_bdlora": null,
38
+ "use_dora": false,
39
+ "use_qalora": false,
40
+ "use_rslora": false
41
+ },
42
+ "hf": {
43
+ "timings": {
44
+ "hf_load_base": 32.24131433299044,
45
+ "hf_apply_adapter": 9.03372309403494,
46
+ "hf_decode_adapter": 111.14524953003274,
47
+ "hf_logits_adapter": 1.4711488810135052,
48
+ "hf_merge": 0.42533229797845706,
49
+ "hf_save_merged": 132.39111311297165,
50
+ "hf_restore_base_only": 1.0795407029800117,
51
+ "hf_load_merged": 15.514058025029954,
52
+ "hf_decode_merged": 55.286309406976216,
53
+ "hf_logits_merged": 1.0814605370396748
54
+ },
55
+ "hf_adapter_vs_merged": {
56
+ "all_token_ids_exact_match": false,
57
+ "logit_max_abs_diff": 6.109375,
58
+ "logit_mean_abs_diff": 0.033503563340466246,
59
+ "argmax_agreement": 0.9929140832595217,
60
+ "per_prompt": [
61
+ {
62
+ "prompt": "What is the capital of France, and why is it historically significant?",
63
+ "token_ids_exact_match": false,
64
+ "first_divergence_idx": 20,
65
+ "logit_max_abs_diff": 6.109375,
66
+ "logit_mean_abs_diff": 0.03354217857122421,
67
+ "argmax_agreement": 0.9857142567634583
68
+ },
69
+ {
70
+ "prompt": "Solve step by step: a train travels 120 km in 1.5 hours. What is its average speed?",
71
+ "token_ids_exact_match": false,
72
+ "first_divergence_idx": 138,
73
+ "logit_max_abs_diff": 2.8125,
74
+ "logit_mean_abs_diff": 0.03629594296216965,
75
+ "argmax_agreement": 0.9931740760803223
76
+ },
77
+ {
78
+ "prompt": "Write a short Python function that returns the n-th Fibonacci number.",
79
+ "token_ids_exact_match": true,
80
+ "first_divergence_idx": null,
81
+ "logit_max_abs_diff": 2.78125,
82
+ "logit_mean_abs_diff": 0.03149561583995819,
83
+ "argmax_agreement": 0.9964157938957214
84
+ },
85
+ {
86
+ "prompt": "Explain in two sentences why the sky appears blue.",
87
+ "token_ids_exact_match": true,
88
+ "first_divergence_idx": null,
89
+ "logit_max_abs_diff": 3.09375,
90
+ "logit_mean_abs_diff": 0.032533302903175354,
91
+ "argmax_agreement": 0.9963898658752441
92
+ }
93
+ ]
94
+ }
95
+ },
96
+ "weight_check": {
97
+ "scaling": 2.0,
98
+ "n_modules": 256,
99
+ "n_modules_total": 256,
100
+ "max_abs_diff_vs_fp32_ref": 0.00022973120212554932,
101
+ "max_rel_diff_vs_fp32_ref": 0.0038910505827516317,
102
+ "worst_module": "model.language_model.layers.47.self_attn.v_proj.weight",
103
+ "non_target_checked": [
104
+ [
105
+ "model.language_model.embed_tokens.weight",
106
+ true
107
+ ],
108
+ [
109
+ "model.visual.patch_embed.proj.bias",
110
+ true
111
+ ],
112
+ [
113
+ "model.visual.patch_embed.proj.weight",
114
+ true
115
+ ],
116
+ [
117
+ "model.visual.pos_embed.weight",
118
+ true
119
+ ],
120
+ [
121
+ "mtp.pre_fc_norm_embedding.weight",
122
+ true
123
+ ],
124
+ [
125
+ "lm_head.weight",
126
+ true
127
+ ],
128
+ [
129
+ "model.language_model.layers.0.input_layernorm.weight",
130
+ true
131
+ ],
132
+ [
133
+ "model.language_model.layers.0.linear_attn.norm.weight",
134
+ true
135
+ ]
136
+ ],
137
+ "non_target_identical": 8,
138
+ "bf16_ulp_rel": 0.00390625
139
+ },
140
+ "engines": {
141
+ "A": {
142
+ "engine": "HF base+LoRA",
143
+ "gens": [
144
+ [
145
+ 8160,
146
+ 579,
147
+ 264,
148
+ 7047,
149
+ 1817,
150
+ 25,
151
+ 271,
152
+ 16,
153
+ 13,
154
+ 220,
155
+ 2972,
156
+ 27382,
157
+ 1386,
158
+ 279,
159
+ 6007,
160
+ 3296,
161
+ 92217,
162
+ 561,
163
+ 1156,
164
+ 16561,
165
+ 364,
166
+ 279,
167
+ 6511,
168
+ 314,
169
+ 9338,
170
+ 321,
171
+ 1141,
172
+ 13263,
173
+ 24556,
174
+ 13,
175
+ 198,
176
+ 17,
177
+ 13,
178
+ 220,
179
+ 2972,
180
+ 15666,
181
+ 279,
182
+ 1118,
183
+ 919,
184
+ 92217,
185
+ 561,
186
+ 6511,
187
+ 314,
188
+ 9338,
189
+ 369,
190
+ 11751,
191
+ 13,
192
+ 198,
193
+ 18,
194
+ 13,
195
+ 220,
196
+ 2972,
197
+ 4146,
198
+ 279,
199
+ 2018,
200
+ 919,
201
+ 92217,
202
+ 8169,
203
+ 369,
204
+ 11751,
205
+ 33649,
206
+ 4927,
207
+ 30,
208
+ 353,
209
+ 1144,
210
+ 310,
211
+ 3300,
212
+ 264,
213
+ 61446,
214
+ 694,
215
+ 15352,
216
+ 22527,
217
+ 314,
218
+ 1141,
219
+ 13263,
220
+ 12287,
221
+ 13,
222
+ 198,
223
+ 19,
224
+ 13,
225
+ 220,
226
+ 2972,
227
+ 1536,
228
+ 13263,
229
+ 3387,
230
+ 883,
231
+ 11751,
232
+ 92217,
233
+ 198,
234
+ 256,
235
+ 471,
236
+ 35725,
237
+ 14,
238
+ 39836,
239
+ 3712,
240
+ 25,
241
+ 75053,
242
+ 539,
243
+ 279,
244
+ 11751,
245
+ 3680,
246
+ 32459,
247
+ 11,
248
+ 2843,
249
+ 12386,
250
+ 3177,
251
+ 318,
252
+ 43,
253
+ 330,
254
+ 294,
255
+ 667,
256
+ 8,
257
+ 198,
258
+ 256,
259
+ 471,
260
+ 74509,
261
+ 4031,
262
+ 25,
263
+ 5651,
264
+ 314,
265
+ 6618,
266
+ 318,
267
+ 29187,
268
+ 314,
269
+ 11751,
270
+ 11,
271
+ 17508,
272
+ 3835,
273
+ 16,
274
+ 16,
275
+ 20,
276
+ 15,
277
+ 681,
278
+ 41679,
279
+ 17120,
280
+ 318,
281
+ 2536,
282
+ 265,
283
+ 9141,
284
+ 368,
285
+ 11,
286
+ 199949,
287
+ 28589,
288
+ 385,
289
+ 6502,
290
+ 681,
291
+ 4788,
292
+ 3990,
293
+ 314,
294
+ 279,
295
+ 8323,
296
+ 24284,
297
+ 198,
298
+ 256,
299
+ 471,
300
+ 52459,
301
+ 14,
302
+ 39836,
303
+ 17722,
304
+ 25,
305
+ 38901,
306
+ 321,
307
+ 19034,
308
+ 18156,
309
+ 11,
310
+ 8323,
311
+ 83079,
312
+ 579,
313
+ 10407,
314
+ 11,
315
+ 4257,
316
+ 314,
317
+ 18001,
318
+ 321,
319
+ 34508,
320
+ 198,
321
+ 256,
322
+ 471,
323
+ 220,
324
+ 16,
325
+ 23,
326
+ 337,
327
+ 9018,
328
+ 25,
329
+ 88664,
330
+ 3990,
331
+ 11,
332
+ 2660,
333
+ 314,
334
+ 279,
335
+ 8323,
336
+ 21440,
337
+ 318,
338
+ 16,
339
+ 22,
340
+ 23,
341
+ 24,
342
+ 681,
343
+ 43563,
344
+ 4369,
345
+ 11,
346
+ 78082,
347
+ 314,
348
+ 35925,
349
+ 11,
350
+ 3478,
351
+ 4788,
352
+ 92305,
353
+ 21090,
354
+ 198,
355
+ 256,
356
+ 471,
357
+ 220,
358
+ 16,
359
+ 24,
360
+ 337,
361
+ 9018,
362
+ 25,
363
+ 23770,
364
+ 1954,
365
+ 11,
366
+ 13697,
367
+ 1790,
368
+ 17039,
369
+ 579,
370
+ 74910,
371
+ 11,
372
+ 6590,
373
+ 2792,
374
+ 314,
375
+ 220,
376
+ 16,
377
+ 23,
378
+ 22,
379
+ 16,
380
+ 11,
381
+ 12387,
382
+ 94869,
383
+ 318,
384
+ 24415,
385
+ 10860,
386
+ 2074,
387
+ 11,
388
+ 16698,
389
+ 11,
390
+ 18550,
391
+ 8,
392
+ 198,
393
+ 256,
394
+ 471,
395
+ 220,
396
+ 17,
397
+ 15,
398
+ 337,
399
+ 9018,
400
+ 25
401
+ ],
402
+ [
403
+ 8160,
404
+ 579,
405
+ 264,
406
+ 7047,
407
+ 1817,
408
+ 25,
409
+ 271,
410
+ 16,
411
+ 13,
412
+ 220,
413
+ 2972,
414
+ 2014,
415
+ 53983,
416
+ 2570,
417
+ 5396,
418
+ 64700,
419
+ 198,
420
+ 256,
421
+ 471,
422
+ 21397,
423
+ 25,
424
+ 357,
425
+ 5257,
426
+ 33060,
427
+ 220,
428
+ 16,
429
+ 17,
430
+ 15,
431
+ 12754,
432
+ 303,
433
+ 220,
434
+ 16,
435
+ 13,
436
+ 20,
437
+ 3979,
438
+ 13,
439
+ 198,
440
+ 256,
441
+ 471,
442
+ 15380,
443
+ 25,
444
+ 3437,
445
+ 369,
446
+ 1141,
447
+ 5289,
448
+ 4478,
449
+ 30,
450
+ 198,
451
+ 256,
452
+ 471,
453
+ 5952,
454
+ 25,
455
+ 61148,
456
+ 2923,
457
+ 539,
458
+ 2923,
459
+ 13,
460
+ 271,
461
+ 17,
462
+ 13,
463
+ 220,
464
+ 2972,
465
+ 27382,
466
+ 1386,
467
+ 5141,
468
+ 73182,
469
+ 64700,
470
+ 198,
471
+ 256,
472
+ 471,
473
+ 22830,
474
+ 4478,
475
+ 14377,
476
+ 25,
477
+ 14739,
478
+ 283,
479
+ 30116,
480
+ 593,
481
+ 3984,
482
+ 198,
483
+ 256,
484
+ 471,
485
+ 15767,
486
+ 25,
487
+ 30116,
488
+ 283,
489
+ 220,
490
+ 16,
491
+ 17,
492
+ 15,
493
+ 12754,
494
+ 11,
495
+ 3984,
496
+ 283,
497
+ 220,
498
+ 16,
499
+ 13,
500
+ 20,
501
+ 3979,
502
+ 198,
503
+ 256,
504
+ 471,
505
+ 14235,
506
+ 310,
507
+ 10724,
508
+ 25,
509
+ 14739,
510
+ 303,
511
+ 12754,
512
+ 7294,
513
+ 271,
514
+ 18,
515
+ 13,
516
+ 220,
517
+ 2972,
518
+ 8046,
519
+ 13909,
520
+ 12,
521
+ 8046,
522
+ 71690,
523
+ 64700,
524
+ 198,
525
+ 256,
526
+ 471,
527
+ 14392,
528
+ 220,
529
+ 16,
530
+ 25,
531
+ 9357,
532
+ 1441,
533
+ 279,
534
+ 14377,
535
+ 364,
536
+ 5289,
537
+ 4478,
538
+ 13,
539
+ 198,
540
+ 257,
541
+ 14739,
542
+ 283,
543
+ 30116,
544
+ 1624,
545
+ 115,
546
+ 3984,
547
+ 198,
548
+ 256,
549
+ 471,
550
+ 14392,
551
+ 220,
552
+ 17,
553
+ 25,
554
+ 51032,
555
+ 303,
556
+ 279,
557
+ 2574,
558
+ 2663,
559
+ 13,
560
+ 198,
561
+ 257,
562
+ 14739,
563
+ 283,
564
+ 220,
565
+ 16,
566
+ 17,
567
+ 15,
568
+ 12754,
569
+ 1624,
570
+ 115,
571
+ 220,
572
+ 16,
573
+ 13,
574
+ 20,
575
+ 3979,
576
+ 198,
577
+ 256,
578
+ 471,
579
+ 14392,
580
+ 220,
581
+ 18,
582
+ 25,
583
+ 24209,
584
+ 279,
585
+ 12436,
586
+ 13,
587
+ 198,
588
+ 257,
589
+ 220,
590
+ 16,
591
+ 17,
592
+ 15,
593
+ 1624,
594
+ 115,
595
+ 220,
596
+ 16,
597
+ 13,
598
+ 20,
599
+ 283,
600
+ 220,
601
+ 23,
602
+ 15,
603
+ 198,
604
+ 256,
605
+ 471,
606
+ 14392,
607
+ 220,
608
+ 19,
609
+ 25,
610
+ 3130,
611
+ 279,
612
+ 1534,
613
+ 4087,
614
+ 440,
615
+ 7896,
616
+ 13,
617
+ 198,
618
+ 257,
619
+ 22830,
620
+ 4478,
621
+ 283,
622
+ 220,
623
+ 23,
624
+ 15,
625
+ 12754,
626
+ 7294,
627
+ 271,
628
+ 19,
629
+ 13,
630
+ 220,
631
+ 2972,
632
+ 31547,
633
+ 64700,
634
+ 198,
635
+ 256,
636
+ 471,
637
+ 4109,
638
+ 25,
639
+ 220,
640
+ 23,
641
+ 15,
642
+ 12754,
643
+ 7294,
644
+ 23985,
645
+ 220,
646
+ 16,
647
+ 13,
648
+ 20,
649
+ 304,
650
+ 283,
651
+ 220,
652
+ 16,
653
+ 17,
654
+ 15,
655
+ 12754,
656
+ 13,
657
+ 38643,
658
+ 13
659
+ ],
660
+ [
661
+ 8160,
662
+ 579,
663
+ 264,
664
+ 7047,
665
+ 1817,
666
+ 25,
667
+ 271,
668
+ 16,
669
+ 13,
670
+ 220,
671
+ 2972,
672
+ 15771,
673
+ 2598,
674
+ 2570,
675
+ 5952,
676
+ 92217,
677
+ 561,
678
+ 1156,
679
+ 6587,
680
+ 264,
681
+ 2716,
682
+ 12654,
683
+ 709,
684
+ 421,
685
+ 4523,
686
+ 279,
687
+ 307,
688
+ 7324,
689
+ 76938,
690
+ 1324,
691
+ 13,
692
+ 271,
693
+ 17,
694
+ 13,
695
+ 220,
696
+ 2972,
697
+ 27382,
698
+ 1386,
699
+ 5141,
700
+ 32666,
701
+ 92217,
702
+ 198,
703
+ 256,
704
+ 471,
705
+ 11106,
706
+ 25,
707
+ 12654,
708
+ 198,
709
+ 256,
710
+ 471,
711
+ 8984,
712
+ 25,
713
+ 307,
714
+ 7324,
715
+ 76938,
716
+ 1324,
717
+ 198,
718
+ 256,
719
+ 471,
720
+ 42794,
721
+ 25,
722
+ 10382,
723
+ 30211,
724
+ 66,
725
+ 1029,
726
+ 271,
727
+ 18,
728
+ 13,
729
+ 220,
730
+ 2972,
731
+ 35944,
732
+ 76938,
733
+ 19107,
734
+ 92217,
735
+ 198,
736
+ 256,
737
+ 471,
738
+ 426,
739
+ 7,
740
+ 15,
741
+ 8,
742
+ 283,
743
+ 220,
744
+ 15,
745
+ 198,
746
+ 256,
747
+ 471,
748
+ 426,
749
+ 7,
750
+ 16,
751
+ 8,
752
+ 283,
753
+ 220,
754
+ 16,
755
+ 198,
756
+ 256,
757
+ 471,
758
+ 426,
759
+ 1393,
760
+ 8,
761
+ 283,
762
+ 426,
763
+ 1393,
764
+ 12,
765
+ 16,
766
+ 8,
767
+ 478,
768
+ 426,
769
+ 1393,
770
+ 12,
771
+ 17,
772
+ 8,
773
+ 364,
774
+ 307,
775
+ 2518,
776
+ 220,
777
+ 17,
778
+ 198,
779
+ 256,
780
+ 471,
781
+ 6817,
782
+ 25,
783
+ 4188,
784
+ 16954,
785
+ 1151,
786
+ 440,
787
+ 426,
788
+ 7,
789
+ 16,
790
+ 11393,
791
+ 16,
792
+ 11,
793
+ 426,
794
+ 7,
795
+ 17,
796
+ 11393,
797
+ 16,
798
+ 11,
799
+ 694,
800
+ 279,
801
+ 5129,
802
+ 15019,
803
+ 20624,
804
+ 369,
805
+ 426,
806
+ 7,
807
+ 15,
808
+ 11393,
809
+ 15,
810
+ 11,
811
+ 426,
812
+ 7,
813
+ 16,
814
+ 11393,
815
+ 16,
816
+ 13,
817
+ 353,
818
+ 3172,
819
+ 8941,
820
+ 440,
821
+ 421,
822
+ 694,
823
+ 628,
824
+ 6088,
825
+ 424,
826
+ 413,
827
+ 4221,
828
+ 13,
829
+ 271,
830
+ 19,
831
+ 13,
832
+ 220,
833
+ 2972,
834
+ 79503,
835
+ 29806,
836
+ 14147,
837
+ 92217,
838
+ 198,
839
+ 256,
840
+ 471,
841
+ 348,
842
+ 75853,
843
+ 77340,
844
+ 8722,
845
+ 694,
846
+ 61240,
847
+ 318,
848
+ 46,
849
+ 7,
850
+ 17,
851
+ 83193,
852
+ 578,
853
+ 198,
854
+ 256,
855
+ 471,
856
+ 348,
857
+ 8275,
858
+ 1340,
859
+ 77340,
860
+ 85943,
861
+ 318,
862
+ 46,
863
+ 1393,
864
+ 8,
865
+ 854,
866
+ 11,
867
+ 496,
868
+ 7,
869
+ 16,
870
+ 8,
871
+ 3433,
872
+ 8,
873
+ 198,
874
+ 256,
875
+ 471,
876
+ 348,
877
+ 6479,
878
+ 26793,
879
+ 6880,
880
+ 77340,
881
+ 496,
882
+ 12171,
883
+ 307,
884
+ 8,
885
+ 694,
886
+ 777,
887
+ 6150,
888
+ 198,
889
+ 256,
890
+ 471,
891
+ 348,
892
+ 27852,
893
+ 294,
894
+ 579,
895
+ 14377,
896
+ 77340,
897
+ 496,
898
+ 7,
899
+ 16,
900
+ 8,
901
+ 694,
902
+ 18484,
903
+ 1406,
904
+ 15579,
905
+ 4562,
906
+ 198,
907
+ 256,
908
+ 471,
909
+ 348,
910
+ 44856,
911
+ 1954,
912
+ 77340,
913
+ 7441,
914
+ 694,
915
+ 11039,
916
+ 22373
917
+ ],
918
+ [
919
+ 8160,
920
+ 579,
921
+ 264,
922
+ 7047,
923
+ 1817,
924
+ 25,
925
+ 271,
926
+ 16,
927
+ 13,
928
+ 220,
929
+ 2972,
930
+ 2014,
931
+ 53983,
932
+ 2570,
933
+ 5396,
934
+ 64700,
935
+ 198,
936
+ 256,
937
+ 471,
938
+ 2972,
939
+ 14162,
940
+ 64700,
941
+ 8169,
942
+ 1503,
943
+ 279,
944
+ 12515,
945
+ 4833,
946
+ 6105,
947
+ 30,
948
+ 198,
949
+ 256,
950
+ 471,
951
+ 2972,
952
+ 17355,
953
+ 64700,
954
+ 79091,
955
+ 303,
956
+ 6681,
957
+ 1330,
958
+ 22157,
959
+ 13,
960
+ 271,
961
+ 17,
962
+ 13,
963
+ 220,
964
+ 2972,
965
+ 27382,
966
+ 1386,
967
+ 5141,
968
+ 35801,
969
+ 73182,
970
+ 64700,
971
+ 198,
972
+ 256,
973
+ 471,
974
+ 7806,
975
+ 4009,
976
+ 5435,
977
+ 660,
978
+ 7736,
979
+ 314,
980
+ 279,
981
+ 9155,
982
+ 19142,
983
+ 13,
984
+ 198,
985
+ 256,
986
+ 471,
987
+ 8964,
988
+ 579,
989
+ 16078,
990
+ 1100,
991
+ 9872,
992
+ 22602,
993
+ 89661,
994
+ 318,
995
+ 11855,
996
+ 5285,
997
+ 29899,
998
+ 8,
999
+ 777,
1000
+ 1056,
1001
+ 4860,
1002
+ 89661,
1003
+ 318,
1004
+ 1114,
1005
+ 25643,
1006
+ 5131,
1007
+ 8,
1008
+ 4016,
1009
+ 310,
1010
+ 12870,
1011
+ 60845,
1012
+ 69377,
1013
+ 13,
1014
+ 198,
1015
+ 256,
1016
+ 471,
1017
+ 5456,
1018
+ 6213,
1019
+ 513,
1020
+ 777,
1021
+ 15739,
1022
+ 310,
1023
+ 6105,
1024
+ 1056,
1025
+ 76990,
1026
+ 11,
1027
+ 748,
1028
+ 567,
1029
+ 42918,
1030
+ 279,
1031
+ 12515,
1032
+ 430,
1033
+ 6105,
1034
+ 13,
1035
+ 271,
1036
+ 18,
1037
+ 13,
1038
+ 220,
1039
+ 2972,
1040
+ 48401,
1041
+ 471,
1042
+ 76802,
1043
+ 220,
1044
+ 16,
1045
+ 318,
1046
+ 13423,
1047
+ 383,
1048
+ 69377,
1049
+ 188912,
1050
+ 198,
1051
+ 256,
1052
+ 7806,
1053
+ 4009,
1054
+ 5435,
1055
+ 660,
1056
+ 7736,
1057
+ 314,
1058
+ 279,
1059
+ 9155,
1060
+ 19142,
1061
+ 11,
1062
+ 694,
1063
+ 8964,
1064
+ 579,
1065
+ 16078,
1066
+ 1100,
1067
+ 9872,
1068
+ 22602,
1069
+ 6105,
1070
+ 89661,
1071
+ 777,
1072
+ 13057,
1073
+ 1056,
1074
+ 4860,
1075
+ 2438,
1076
+ 89661,
1077
+ 1472,
1078
+ 264,
1079
+ 1817,
1080
+ 2512,
1081
+ 12870,
1082
+ 60845,
1083
+ 69377,
1084
+ 13,
1085
+ 271,
1086
+ 19,
1087
+ 13,
1088
+ 220,
1089
+ 2972,
1090
+ 48401,
1091
+ 471,
1092
+ 76802,
1093
+ 220,
1094
+ 17,
1095
+ 318,
1096
+ 13423,
1097
+ 383,
1098
+ 19808,
1099
+ 188912,
1100
+ 198,
1101
+ 256,
1102
+ 10022,
1103
+ 76990,
1104
+ 3000,
1105
+ 1100,
1106
+ 9872,
1107
+ 1442,
1108
+ 777,
1109
+ 11,
1110
+ 1004,
1111
+ 6213,
1112
+ 513,
1113
+ 777,
1114
+ 15739,
1115
+ 310,
1116
+ 6105,
1117
+ 11,
1118
+ 864,
1119
+ 369,
1120
+ 3069,
1121
+ 567,
1122
+ 42918,
1123
+ 279,
1124
+ 12515,
1125
+ 430,
1126
+ 45239,
1127
+ 6105,
1128
+ 13,
1129
+ 271,
1130
+ 20,
1131
+ 13,
1132
+ 220,
1133
+ 2972,
1134
+ 3840,
1135
+ 84457,
1136
+ 64700,
1137
+ 198,
1138
+ 256,
1139
+ 471,
1140
+ 66172,
1141
+ 1330,
1142
+ 22157,
1143
+ 30,
1144
+ 7179,
1145
+ 13,
1146
+ 198,
1147
+ 256,
1148
+ 471,
1149
+ 17467,
1150
+ 1674,
1151
+ 3069,
1152
+ 279,
1153
+ 12515,
1154
+ 7701,
1155
+ 6105,
1156
+ 30,
1157
+ 7179,
1158
+ 13,
1159
+ 198,
1160
+ 256,
1161
+ 471,
1162
+ 35801,
1163
+ 722,
1164
+ 12995,
1165
+ 30,
1166
+ 7179,
1167
+ 13,
1168
+ 271,
1169
+ 21,
1170
+ 13,
1171
+ 220,
1172
+ 2972,
1173
+ 3812,
1174
+ 472
1175
+ ]
1176
+ ],
1177
+ "load_s": 41.27503742702538,
1178
+ "decode_s": 111.14524953003274,
1179
+ "gen_tokens": 1024
1180
+ },
1181
+ "B": {
1182
+ "engine": "HF merged",
1183
+ "gens": [
1184
+ [
1185
+ 8160,
1186
+ 579,
1187
+ 264,
1188
+ 7047,
1189
+ 1817,
1190
+ 25,
1191
+ 271,
1192
+ 16,
1193
+ 13,
1194
+ 220,
1195
+ 2972,
1196
+ 27382,
1197
+ 1386,
1198
+ 279,
1199
+ 6007,
1200
+ 3296,
1201
+ 92217,
1202
+ 561,
1203
+ 1156,
1204
+ 16561,
1205
+ 1330,
1206
+ 2433,
1207
+ 25,
1208
+ 198,
1209
+ 256,
1210
+ 471,
1211
+ 3437,
1212
+ 369,
1213
+ 279,
1214
+ 6511,
1215
+ 314,
1216
+ 9338,
1217
+ 30,
1218
+ 198,
1219
+ 256,
1220
+ 471,
1221
+ 8169,
1222
+ 369,
1223
+ 424,
1224
+ 33649,
1225
+ 4927,
1226
+ 30,
1227
+ 271,
1228
+ 17,
1229
+ 13,
1230
+ 220,
1231
+ 2972,
1232
+ 15666,
1233
+ 279,
1234
+ 1118,
1235
+ 919,
1236
+ 92217,
1237
+ 561,
1238
+ 6511,
1239
+ 314,
1240
+ 9338,
1241
+ 369,
1242
+ 11751,
1243
+ 13,
1244
+ 271,
1245
+ 18,
1246
+ 13,
1247
+ 220,
1248
+ 2972,
1249
+ 4146,
1250
+ 279,
1251
+ 2018,
1252
+ 919,
1253
+ 318,
1254
+ 42005,
1255
+ 908,
1256
+ 24556,
1257
+ 31230,
1258
+ 25,
1259
+ 353,
1260
+ 1144,
1261
+ 310,
1262
+ 3300,
1263
+ 264,
1264
+ 61446,
1265
+ 694,
1266
+ 15352,
1267
+ 22527,
1268
+ 314,
1269
+ 11751,
1270
+ 579,
1271
+ 13263,
1272
+ 24556,
1273
+ 13,
1274
+ 5141,
1275
+ 3387,
1276
+ 310,
1277
+ 3311,
1278
+ 25,
1279
+ 198,
1280
+ 256,
1281
+ 471,
1282
+ 35725,
1283
+ 14,
1284
+ 1993,
1285
+ 30839,
1286
+ 31304,
1287
+ 318,
1288
+ 43,
1289
+ 330,
1290
+ 294,
1291
+ 667,
1292
+ 11,
1293
+ 12386,
1294
+ 16576,
1295
+ 11,
1296
+ 39654,
1297
+ 6310,
1298
+ 8,
1299
+ 198,
1300
+ 256,
1301
+ 471,
1302
+ 29501,
1303
+ 2805,
1304
+ 42011,
1305
+ 3990,
1306
+ 314,
1307
+ 9338,
1308
+ 318,
1309
+ 3243,
1310
+ 278,
1311
+ 6511,
1312
+ 11,
1313
+ 83079,
1314
+ 11,
1315
+ 8323,
1316
+ 21440,
1317
+ 8,
1318
+ 198,
1319
+ 256,
1320
+ 471,
1321
+ 88664,
1322
+ 321,
1323
+ 19034,
1324
+ 18156,
1325
+ 318,
1326
+ 18642,
1327
+ 2305,
1328
+ 11,
1329
+ 58641,
1330
+ 11,
1331
+ 11991,
1332
+ 79835,
1333
+ 8,
1334
+ 198,
1335
+ 256,
1336
+ 471,
1337
+ 5001,
1338
+ 321,
1339
+ 7431,
1340
+ 318,
1341
+ 49,
1342
+ 6933,
1343
+ 23217,
1344
+ 11,
1345
+ 13977,
1346
+ 10860,
1347
+ 2074,
1348
+ 11,
1349
+ 6278,
1350
+ 1880,
1351
+ 11,
1352
+ 16698,
1353
+ 8,
1354
+ 198,
1355
+ 256,
1356
+ 471,
1357
+ 23269,
1358
+ 4038,
1359
+ 57902,
1360
+ 318,
1361
+ 36,
1362
+ 2992,
1363
+ 300,
1364
+ 21262,
1365
+ 11,
1366
+ 42027,
1367
+ 9141,
1368
+ 368,
1369
+ 11,
1370
+ 9440,
1371
+ 46890,
1372
+ 11,
1373
+ 4831,
1374
+ 5946,
1375
+ 198,
1376
+ 256,
1377
+ 471,
1378
+ 14955,
1379
+ 303,
1380
+ 3478,
1381
+ 13263,
1382
+ 4216,
1383
+ 318,
1384
+ 41772,
1385
+ 21440,
1386
+ 11,
1387
+ 4196,
1388
+ 14534,
1389
+ 11,
1390
+ 220,
1391
+ 16,
1392
+ 24,
1393
+ 21,
1394
+ 23,
1395
+ 21231,
1396
+ 11,
1397
+ 4831,
1398
+ 5946,
1399
+ 198,
1400
+ 256,
1401
+ 471,
1402
+ 7711,
1403
+ 9870,
1404
+ 318,
1405
+ 69,
1406
+ 8517,
1407
+ 11,
1408
+ 33847,
1409
+ 11,
1410
+ 59174,
1411
+ 11,
1412
+ 77989,
1413
+ 4196,
1414
+ 32043,
1415
+ 8,
1416
+ 271,
1417
+ 19,
1418
+ 13,
1419
+ 220,
1420
+ 2972,
1421
+ 22238,
1422
+ 279,
1423
+ 1965,
1424
+ 92217,
1425
+ 198,
1426
+ 256,
1427
+ 471,
1428
+ 6916,
1429
+ 4087,
1430
+ 1118,
1431
+ 198,
1432
+ 256,
1433
+ 471,
1434
+ 10446,
1435
+ 537,
1436
+ 13263,
1437
+ 24556,
1438
+ 1083,
1439
+ 2708,
1440
+ 11
1441
+ ],
1442
+ [
1443
+ 8160,
1444
+ 579,
1445
+ 264,
1446
+ 7047,
1447
+ 1817,
1448
+ 25,
1449
+ 271,
1450
+ 16,
1451
+ 13,
1452
+ 220,
1453
+ 2972,
1454
+ 2014,
1455
+ 53983,
1456
+ 2570,
1457
+ 5396,
1458
+ 64700,
1459
+ 198,
1460
+ 256,
1461
+ 471,
1462
+ 21397,
1463
+ 25,
1464
+ 357,
1465
+ 5257,
1466
+ 33060,
1467
+ 220,
1468
+ 16,
1469
+ 17,
1470
+ 15,
1471
+ 12754,
1472
+ 303,
1473
+ 220,
1474
+ 16,
1475
+ 13,
1476
+ 20,
1477
+ 3979,
1478
+ 13,
1479
+ 198,
1480
+ 256,
1481
+ 471,
1482
+ 15380,
1483
+ 25,
1484
+ 3437,
1485
+ 369,
1486
+ 1141,
1487
+ 5289,
1488
+ 4478,
1489
+ 30,
1490
+ 198,
1491
+ 256,
1492
+ 471,
1493
+ 5952,
1494
+ 25,
1495
+ 61148,
1496
+ 2923,
1497
+ 539,
1498
+ 2923,
1499
+ 13,
1500
+ 271,
1501
+ 17,
1502
+ 13,
1503
+ 220,
1504
+ 2972,
1505
+ 27382,
1506
+ 1386,
1507
+ 5141,
1508
+ 73182,
1509
+ 64700,
1510
+ 198,
1511
+ 256,
1512
+ 471,
1513
+ 22830,
1514
+ 4478,
1515
+ 14377,
1516
+ 25,
1517
+ 14739,
1518
+ 283,
1519
+ 30116,
1520
+ 593,
1521
+ 3984,
1522
+ 198,
1523
+ 256,
1524
+ 471,
1525
+ 15767,
1526
+ 25,
1527
+ 30116,
1528
+ 283,
1529
+ 220,
1530
+ 16,
1531
+ 17,
1532
+ 15,
1533
+ 12754,
1534
+ 11,
1535
+ 3984,
1536
+ 283,
1537
+ 220,
1538
+ 16,
1539
+ 13,
1540
+ 20,
1541
+ 3979,
1542
+ 198,
1543
+ 256,
1544
+ 471,
1545
+ 14235,
1546
+ 310,
1547
+ 10724,
1548
+ 25,
1549
+ 14739,
1550
+ 303,
1551
+ 12754,
1552
+ 7294,
1553
+ 271,
1554
+ 18,
1555
+ 13,
1556
+ 220,
1557
+ 2972,
1558
+ 8046,
1559
+ 13909,
1560
+ 12,
1561
+ 8046,
1562
+ 71690,
1563
+ 64700,
1564
+ 198,
1565
+ 256,
1566
+ 471,
1567
+ 14392,
1568
+ 220,
1569
+ 16,
1570
+ 25,
1571
+ 9357,
1572
+ 1441,
1573
+ 279,
1574
+ 14377,
1575
+ 364,
1576
+ 5289,
1577
+ 4478,
1578
+ 13,
1579
+ 198,
1580
+ 257,
1581
+ 471,
1582
+ 14739,
1583
+ 283,
1584
+ 30116,
1585
+ 1624,
1586
+ 115,
1587
+ 3984,
1588
+ 198,
1589
+ 256,
1590
+ 471,
1591
+ 14392,
1592
+ 220,
1593
+ 17,
1594
+ 25,
1595
+ 51032,
1596
+ 303,
1597
+ 279,
1598
+ 2574,
1599
+ 2663,
1600
+ 13,
1601
+ 198,
1602
+ 257,
1603
+ 471,
1604
+ 14739,
1605
+ 283,
1606
+ 220,
1607
+ 16,
1608
+ 17,
1609
+ 15,
1610
+ 12754,
1611
+ 1624,
1612
+ 115,
1613
+ 220,
1614
+ 16,
1615
+ 13,
1616
+ 20,
1617
+ 3979,
1618
+ 198,
1619
+ 256,
1620
+ 471,
1621
+ 14392,
1622
+ 220,
1623
+ 18,
1624
+ 25,
1625
+ 24209,
1626
+ 279,
1627
+ 12436,
1628
+ 13,
1629
+ 198,
1630
+ 257,
1631
+ 471,
1632
+ 220,
1633
+ 16,
1634
+ 17,
1635
+ 15,
1636
+ 1624,
1637
+ 115,
1638
+ 220,
1639
+ 16,
1640
+ 13,
1641
+ 20,
1642
+ 283,
1643
+ 220,
1644
+ 23,
1645
+ 15,
1646
+ 198,
1647
+ 256,
1648
+ 471,
1649
+ 14392,
1650
+ 220,
1651
+ 19,
1652
+ 25,
1653
+ 3130,
1654
+ 279,
1655
+ 1534,
1656
+ 4087,
1657
+ 440,
1658
+ 7896,
1659
+ 13,
1660
+ 198,
1661
+ 257,
1662
+ 471,
1663
+ 22830,
1664
+ 4478,
1665
+ 283,
1666
+ 220,
1667
+ 23,
1668
+ 15,
1669
+ 12754,
1670
+ 7294,
1671
+ 271,
1672
+ 19,
1673
+ 13,
1674
+ 220,
1675
+ 2972,
1676
+ 31547,
1677
+ 64700,
1678
+ 198,
1679
+ 256,
1680
+ 471,
1681
+ 4109,
1682
+ 25,
1683
+ 220,
1684
+ 23,
1685
+ 15,
1686
+ 12754,
1687
+ 7294,
1688
+ 23985,
1689
+ 220,
1690
+ 16,
1691
+ 13,
1692
+ 20,
1693
+ 304,
1694
+ 283,
1695
+ 220,
1696
+ 16,
1697
+ 17,
1698
+ 15
1699
+ ],
1700
+ [
1701
+ 8160,
1702
+ 579,
1703
+ 264,
1704
+ 7047,
1705
+ 1817,
1706
+ 25,
1707
+ 271,
1708
+ 16,
1709
+ 13,
1710
+ 220,
1711
+ 2972,
1712
+ 15771,
1713
+ 2598,
1714
+ 2570,
1715
+ 5952,
1716
+ 92217,
1717
+ 561,
1718
+ 1156,
1719
+ 6587,
1720
+ 264,
1721
+ 2716,
1722
+ 12654,
1723
+ 709,
1724
+ 421,
1725
+ 4523,
1726
+ 279,
1727
+ 307,
1728
+ 7324,
1729
+ 76938,
1730
+ 1324,
1731
+ 13,
1732
+ 271,
1733
+ 17,
1734
+ 13,
1735
+ 220,
1736
+ 2972,
1737
+ 27382,
1738
+ 1386,
1739
+ 5141,
1740
+ 32666,
1741
+ 92217,
1742
+ 198,
1743
+ 256,
1744
+ 471,
1745
+ 11106,
1746
+ 25,
1747
+ 12654,
1748
+ 198,
1749
+ 256,
1750
+ 471,
1751
+ 8984,
1752
+ 25,
1753
+ 307,
1754
+ 7324,
1755
+ 76938,
1756
+ 1324,
1757
+ 198,
1758
+ 256,
1759
+ 471,
1760
+ 42794,
1761
+ 25,
1762
+ 10382,
1763
+ 30211,
1764
+ 66,
1765
+ 1029,
1766
+ 271,
1767
+ 18,
1768
+ 13,
1769
+ 220,
1770
+ 2972,
1771
+ 35944,
1772
+ 76938,
1773
+ 19107,
1774
+ 92217,
1775
+ 198,
1776
+ 256,
1777
+ 471,
1778
+ 426,
1779
+ 7,
1780
+ 15,
1781
+ 8,
1782
+ 283,
1783
+ 220,
1784
+ 15,
1785
+ 198,
1786
+ 256,
1787
+ 471,
1788
+ 426,
1789
+ 7,
1790
+ 16,
1791
+ 8,
1792
+ 283,
1793
+ 220,
1794
+ 16,
1795
+ 198,
1796
+ 256,
1797
+ 471,
1798
+ 426,
1799
+ 1393,
1800
+ 8,
1801
+ 283,
1802
+ 426,
1803
+ 1393,
1804
+ 12,
1805
+ 16,
1806
+ 8,
1807
+ 478,
1808
+ 426,
1809
+ 1393,
1810
+ 12,
1811
+ 17,
1812
+ 8,
1813
+ 364,
1814
+ 307,
1815
+ 2518,
1816
+ 220,
1817
+ 17,
1818
+ 198,
1819
+ 256,
1820
+ 471,
1821
+ 6817,
1822
+ 25,
1823
+ 4188,
1824
+ 16954,
1825
+ 1151,
1826
+ 440,
1827
+ 426,
1828
+ 7,
1829
+ 16,
1830
+ 11393,
1831
+ 16,
1832
+ 11,
1833
+ 426,
1834
+ 7,
1835
+ 17,
1836
+ 11393,
1837
+ 16,
1838
+ 11,
1839
+ 694,
1840
+ 279,
1841
+ 5129,
1842
+ 15019,
1843
+ 20624,
1844
+ 369,
1845
+ 426,
1846
+ 7,
1847
+ 15,
1848
+ 11393,
1849
+ 15,
1850
+ 11,
1851
+ 426,
1852
+ 7,
1853
+ 16,
1854
+ 11393,
1855
+ 16,
1856
+ 13,
1857
+ 353,
1858
+ 3172,
1859
+ 8941,
1860
+ 440,
1861
+ 421,
1862
+ 694,
1863
+ 628,
1864
+ 6088,
1865
+ 424,
1866
+ 413,
1867
+ 4221,
1868
+ 13,
1869
+ 271,
1870
+ 19,
1871
+ 13,
1872
+ 220,
1873
+ 2972,
1874
+ 79503,
1875
+ 29806,
1876
+ 14147,
1877
+ 92217,
1878
+ 198,
1879
+ 256,
1880
+ 471,
1881
+ 348,
1882
+ 75853,
1883
+ 77340,
1884
+ 8722,
1885
+ 694,
1886
+ 61240,
1887
+ 318,
1888
+ 46,
1889
+ 7,
1890
+ 17,
1891
+ 83193,
1892
+ 578,
1893
+ 198,
1894
+ 256,
1895
+ 471,
1896
+ 348,
1897
+ 8275,
1898
+ 1340,
1899
+ 77340,
1900
+ 85943,
1901
+ 318,
1902
+ 46,
1903
+ 1393,
1904
+ 8,
1905
+ 854,
1906
+ 11,
1907
+ 496,
1908
+ 7,
1909
+ 16,
1910
+ 8,
1911
+ 3433,
1912
+ 8,
1913
+ 198,
1914
+ 256,
1915
+ 471,
1916
+ 348,
1917
+ 6479,
1918
+ 26793,
1919
+ 6880,
1920
+ 77340,
1921
+ 496,
1922
+ 12171,
1923
+ 307,
1924
+ 8,
1925
+ 694,
1926
+ 777,
1927
+ 6150,
1928
+ 198,
1929
+ 256,
1930
+ 471,
1931
+ 348,
1932
+ 27852,
1933
+ 294,
1934
+ 579,
1935
+ 14377,
1936
+ 77340,
1937
+ 496,
1938
+ 7,
1939
+ 16,
1940
+ 8,
1941
+ 694,
1942
+ 18484,
1943
+ 1406,
1944
+ 15579,
1945
+ 4562,
1946
+ 198,
1947
+ 256,
1948
+ 471,
1949
+ 348,
1950
+ 44856,
1951
+ 1954,
1952
+ 77340,
1953
+ 7441,
1954
+ 694,
1955
+ 11039,
1956
+ 22373
1957
+ ],
1958
+ [
1959
+ 8160,
1960
+ 579,
1961
+ 264,
1962
+ 7047,
1963
+ 1817,
1964
+ 25,
1965
+ 271,
1966
+ 16,
1967
+ 13,
1968
+ 220,
1969
+ 2972,
1970
+ 2014,
1971
+ 53983,
1972
+ 2570,
1973
+ 5396,
1974
+ 64700,
1975
+ 198,
1976
+ 256,
1977
+ 471,
1978
+ 2972,
1979
+ 14162,
1980
+ 64700,
1981
+ 8169,
1982
+ 1503,
1983
+ 279,
1984
+ 12515,
1985
+ 4833,
1986
+ 6105,
1987
+ 30,
1988
+ 198,
1989
+ 256,
1990
+ 471,
1991
+ 2972,
1992
+ 17355,
1993
+ 64700,
1994
+ 79091,
1995
+ 303,
1996
+ 6681,
1997
+ 1330,
1998
+ 22157,
1999
+ 13,
2000
+ 271,
2001
+ 17,
2002
+ 13,
2003
+ 220,
2004
+ 2972,
2005
+ 27382,
2006
+ 1386,
2007
+ 5141,
2008
+ 35801,
2009
+ 73182,
2010
+ 64700,
2011
+ 198,
2012
+ 256,
2013
+ 471,
2014
+ 7806,
2015
+ 4009,
2016
+ 5435,
2017
+ 660,
2018
+ 7736,
2019
+ 314,
2020
+ 279,
2021
+ 9155,
2022
+ 19142,
2023
+ 13,
2024
+ 198,
2025
+ 256,
2026
+ 471,
2027
+ 8964,
2028
+ 579,
2029
+ 16078,
2030
+ 1100,
2031
+ 9872,
2032
+ 22602,
2033
+ 89661,
2034
+ 318,
2035
+ 11855,
2036
+ 5285,
2037
+ 29899,
2038
+ 8,
2039
+ 777,
2040
+ 1056,
2041
+ 4860,
2042
+ 89661,
2043
+ 318,
2044
+ 1114,
2045
+ 25643,
2046
+ 5131,
2047
+ 8,
2048
+ 4016,
2049
+ 310,
2050
+ 12870,
2051
+ 60845,
2052
+ 69377,
2053
+ 13,
2054
+ 198,
2055
+ 256,
2056
+ 471,
2057
+ 5456,
2058
+ 6213,
2059
+ 513,
2060
+ 777,
2061
+ 15739,
2062
+ 310,
2063
+ 6105,
2064
+ 1056,
2065
+ 76990,
2066
+ 11,
2067
+ 748,
2068
+ 567,
2069
+ 42918,
2070
+ 279,
2071
+ 12515,
2072
+ 430,
2073
+ 6105,
2074
+ 13,
2075
+ 271,
2076
+ 18,
2077
+ 13,
2078
+ 220,
2079
+ 2972,
2080
+ 48401,
2081
+ 471,
2082
+ 76802,
2083
+ 220,
2084
+ 16,
2085
+ 318,
2086
+ 13423,
2087
+ 383,
2088
+ 69377,
2089
+ 188912,
2090
+ 198,
2091
+ 256,
2092
+ 7806,
2093
+ 4009,
2094
+ 5435,
2095
+ 660,
2096
+ 7736,
2097
+ 314,
2098
+ 279,
2099
+ 9155,
2100
+ 19142,
2101
+ 11,
2102
+ 694,
2103
+ 8964,
2104
+ 579,
2105
+ 16078,
2106
+ 1100,
2107
+ 9872,
2108
+ 22602,
2109
+ 6105,
2110
+ 89661,
2111
+ 777,
2112
+ 13057,
2113
+ 1056,
2114
+ 4860,
2115
+ 2438,
2116
+ 89661,
2117
+ 1472,
2118
+ 264,
2119
+ 1817,
2120
+ 2512,
2121
+ 12870,
2122
+ 60845,
2123
+ 69377,
2124
+ 13,
2125
+ 271,
2126
+ 19,
2127
+ 13,
2128
+ 220,
2129
+ 2972,
2130
+ 48401,
2131
+ 471,
2132
+ 76802,
2133
+ 220,
2134
+ 17,
2135
+ 318,
2136
+ 13423,
2137
+ 383,
2138
+ 19808,
2139
+ 188912,
2140
+ 198,
2141
+ 256,
2142
+ 10022,
2143
+ 76990,
2144
+ 3000,
2145
+ 1100,
2146
+ 9872,
2147
+ 1442,
2148
+ 777,
2149
+ 11,
2150
+ 1004,
2151
+ 6213,
2152
+ 513,
2153
+ 777,
2154
+ 15739,
2155
+ 310,
2156
+ 6105,
2157
+ 11,
2158
+ 864,
2159
+ 369,
2160
+ 3069,
2161
+ 567,
2162
+ 42918,
2163
+ 279,
2164
+ 12515,
2165
+ 430,
2166
+ 45239,
2167
+ 6105,
2168
+ 13,
2169
+ 271,
2170
+ 20,
2171
+ 13,
2172
+ 220,
2173
+ 2972,
2174
+ 3840,
2175
+ 84457,
2176
+ 64700,
2177
+ 198,
2178
+ 256,
2179
+ 471,
2180
+ 66172,
2181
+ 1330,
2182
+ 22157,
2183
+ 30,
2184
+ 7179,
2185
+ 13,
2186
+ 198,
2187
+ 256,
2188
+ 471,
2189
+ 17467,
2190
+ 1674,
2191
+ 3069,
2192
+ 279,
2193
+ 12515,
2194
+ 7701,
2195
+ 6105,
2196
+ 30,
2197
+ 7179,
2198
+ 13,
2199
+ 198,
2200
+ 256,
2201
+ 471,
2202
+ 35801,
2203
+ 722,
2204
+ 12995,
2205
+ 30,
2206
+ 7179,
2207
+ 13,
2208
+ 271,
2209
+ 21,
2210
+ 13,
2211
+ 220,
2212
+ 2972,
2213
+ 3812,
2214
+ 472
2215
+ ]
2216
+ ],
2217
+ "load_s": 15.514058025029954,
2218
+ "decode_s": 55.286309406976216,
2219
+ "gen_tokens": 1024
2220
+ }
2221
+ },
2222
+ "argmax_matrix": {
2223
+ "A_vs_B": 0.9937777777777778
2224
+ }
2225
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3efb6055936d242c0a481fc0e0b917780c798b91c7f0dcbfdfdb6f6fc7c129a
3
+ size 49825162976
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c824fd15b693cd7e3c1272ededb0c4de46f9f7ecbf1516e35939227e8644255c
3
+ size 4888445168
model-base-aux.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ecba8ef1d0c35b401090ea7e6beee462a738d745d86959e965c54bb4eaf6e4b6
3
+ size 849400424
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
social-instagram.png ADDED
social-linkedin.png ADDED
social-web.png ADDED
social-x.png ADDED
thinkingcap_6x.gif ADDED

Git LFS Details

  • SHA256: 347285925ef7b6a8daf2cc3dcf129008ee5c55068b205785f9eb5e20d19920cd
  • Pointer size: 132 Bytes
  • Size of remote file: 6.69 MB
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523
3
+ size 19989325
tokenizer_config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "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+",
26
+ "split_special_tokens": false,
27
+ "tokenizer_class": "Qwen2Tokenizer",
28
+ "unk_token": null,
29
+ "video_token": "<|video_pad|>",
30
+ "vision_bos_token": "<|vision_start|>",
31
+ "vision_eos_token": "<|vision_end|>"
32
+ }
video_preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 25165824,
4
+ "shortest_edge": 4096
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "video_processor_type": "Qwen3VLVideoProcessor"
21
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
weight_check_report.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "scaling": 2.0,
3
+ "n_modules": 256,
4
+ "n_bitwise_exact_bf16_replicate": 0,
5
+ "max_abs_diff_vs_fp32_ref": 0.00024040043354034424,
6
+ "max_rel_diff_vs_fp32_ref": 0.0038910505827516317,
7
+ "worst_module": "model.language_model.layers.47.self_attn.v_proj.weight",
8
+ "non_target_checked": [
9
+ {
10
+ "key": "model.language_model.embed_tokens.weight",
11
+ "identical": true
12
+ },
13
+ {
14
+ "key": "model.visual.patch_embed.proj.bias",
15
+ "identical": true
16
+ },
17
+ {
18
+ "key": "model.visual.patch_embed.proj.weight",
19
+ "identical": true
20
+ },
21
+ {
22
+ "key": "model.visual.pos_embed.weight",
23
+ "identical": true
24
+ },
25
+ {
26
+ "key": "lm_head.weight",
27
+ "identical": true
28
+ },
29
+ {
30
+ "key": "model.language_model.layers.0.input_layernorm.weight",
31
+ "identical": true
32
+ },
33
+ {
34
+ "key": "model.language_model.layers.0.linear_attn.norm.weight",
35
+ "identical": true
36
+ },
37
+ {
38
+ "key": "model.language_model.layers.0.post_attention_layernorm.weight",
39
+ "identical": true
40
+ }
41
+ ],
42
+ "non_target_identical": 8
43
+ }