Hugging Face's logo

unsloth
/
Ornith-1.0-397B

Text Generation
Safetensors
qwen3_5_moe
unsloth
conversational
Model card Files
xet
 Community
1
danielhanchen commited on
Commit
d6e0cf1
·
verified ·
1 Parent(s): dfc12c8

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +3 -0
  2. README.md +486 -0
  3. assets/ornith_397b_eval.png +3 -0
  4. assets/ornith_logo.png +3 -0
  5. chat_template.jinja +155 -0
  6. config.json +145 -0
  7. generation_config.json +13 -0
  8. model-00001-of-00122.safetensors +3 -0
  9. model-00002-of-00122.safetensors +3 -0
  10. model-00003-of-00122.safetensors +3 -0
  11. model-00004-of-00122.safetensors +3 -0
  12. model-00005-of-00122.safetensors +3 -0
  13. model-00006-of-00122.safetensors +3 -0
  14. model-00007-of-00122.safetensors +3 -0
  15. model-00008-of-00122.safetensors +3 -0
  16. model-00009-of-00122.safetensors +3 -0
  17. model-00010-of-00122.safetensors +3 -0
  18. model-00011-of-00122.safetensors +3 -0
  19. model-00012-of-00122.safetensors +3 -0
  20. model-00013-of-00122.safetensors +3 -0
  21. model-00014-of-00122.safetensors +3 -0
  22. model-00015-of-00122.safetensors +3 -0
  23. model-00016-of-00122.safetensors +3 -0
  24. model-00017-of-00122.safetensors +3 -0
  25. model-00018-of-00122.safetensors +3 -0
  26. model-00019-of-00122.safetensors +3 -0
  27. model-00020-of-00122.safetensors +3 -0
  28. model-00021-of-00122.safetensors +3 -0
  29. model-00022-of-00122.safetensors +3 -0
  30. model-00023-of-00122.safetensors +3 -0
  31. model-00024-of-00122.safetensors +3 -0
  32. model-00025-of-00122.safetensors +3 -0
  33. model-00026-of-00122.safetensors +3 -0
  34. model-00027-of-00122.safetensors +3 -0
  35. model-00028-of-00122.safetensors +3 -0
  36. model-00029-of-00122.safetensors +3 -0
  37. model-00030-of-00122.safetensors +3 -0
  38. model-00031-of-00122.safetensors +3 -0
  39. model-00032-of-00122.safetensors +3 -0
  40. model-00033-of-00122.safetensors +3 -0
  41. model-00034-of-00122.safetensors +3 -0
  42. model-00035-of-00122.safetensors +3 -0
  43. model-00036-of-00122.safetensors +3 -0
  44. model-00037-of-00122.safetensors +3 -0
  45. model-00038-of-00122.safetensors +3 -0
  46. model-00039-of-00122.safetensors +3 -0
  47. model-00040-of-00122.safetensors +3 -0
  48. model-00041-of-00122.safetensors +3 -0
  49. model-00042-of-00122.safetensors +3 -0
  50. model-00043-of-00122.safetensors +3 -0
.gitattributes CHANGED
@@ -33,3 +33,6 @@ saved_model/**/* 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
 
 
 
 
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
+ assets/ornith_397b_eval.png filter=lfs diff=lfs merge=lfs -text
37
+ assets/ornith_logo.png filter=lfs diff=lfs merge=lfs -text
38
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,486 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - unsloth
4
+ base_model:
5
+ - deepreinforce-ai/Ornith-1.0-397B
6
+ library_name: transformers
7
+ license: mit
8
+ license_link: https://huggingface.co/deepreinforce-ai/Ornith-1.0-397B/blob/main/LICENSE
9
+ pipeline_tag: text-generation
10
+ ---
11
+ <div>
12
+ <p style="margin-top: 0;margin-bottom: 0;">
13
+ <em><a href="https://docs.unsloth.ai/basics/unsloth-dynamic-v2.0-gguf">Unsloth Dynamic 2.0</a> achieves superior accuracy & outperforms other leading quants.</em>
14
+ </p>
15
+ <div style="display: flex; gap: 5px; align-items: center; ">
16
+ <a href="https://github.com/unslothai/unsloth/">
17
+ <img src="https://github.com/unslothai/unsloth/raw/main/images/unsloth%20new%20logo.png" width="133">
18
+ </a>
19
+ <a href="https://discord.gg/unsloth">
20
+ <img src="https://github.com/unslothai/unsloth/raw/main/images/Discord%20button.png" width="173">
21
+ </a>
22
+ <a href="https://docs.unsloth.ai/">
23
+ <img src="https://raw.githubusercontent.com/unslothai/unsloth/refs/heads/main/images/documentation%20green%20button.png" width="143">
24
+ </a>
25
+ </div>
26
+ </div>
27
+
28
+
29
+
30
+ <img width="600px" src="assets/ornith_logo.png">
31
+
32
+ [![Ornith Blog](https://img.shields.io/badge/%F0%9F%A6%A2%EF%B8%8F%20Ornith%20Blog%20-FD8E5B)](https://deep-reinforce.com/ornith.html)
33
+
34
+ # Ornith-1.0-397B
35
+
36
+ Aloha! 🌺 Today, we are releasing Ornith-1.0, a self-improving family of open-source models for agentic coding.
37
+
38
+ Highlights:
39
+
40
+ - **State-of-the-Art Coding Agents**: Available in 9B-Dense, 31B-Dense, 35B-MoE, and 397B-MoE (post-trained on top of Gemma 4 and Qwen 3.5), achieving state-of-the-art performance among open-source models of comparable size on coding benchmarks such as Terminal-Bench 2.1, SWE-Bench, NL2Repo and OpenClaw.
41
+ - **Self-Improving Training Framework**: Ornith-1.0 employs RL to learn to generate not only solution rollouts, but also the scallfold that drive those rollouts. By jointly optimizing the scaffold and the resulting solution, the model discovers better search trajectories and generates higher-quality solutions.
42
+ - **Licence**: MIT licensed, globally accessible, and free from regional limitations.
43
+
44
+ <img style="width: 100%; max-width: 900px;" src="assets/ornith_397b_eval.png" alt="Ornith 35B Benchmark Results" title="Ornith 35B Benchmark Results">
45
+
46
+ ## Ornith 1.0 397B
47
+
48
+ This model card documents **Ornith-1.0-397B**, the lightweight member of the Ornith family, designed for efficient single-GPU deployment.
49
+
50
+
51
+ ### Benchmarks
52
+
53
+ <div style="font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif;width:100%;margin:0 auto;padding:16px 0">
54
+ <table style="width:100%;table-layout:fixed;border-collapse:collapse;font-size:13px">
55
+ <thead><tr>
56
+ <th style="width:24%;padding:10px 7px;text-align:left;font-weight:600;border-bottom:2px solid #FD8E5B;color:#FD8E5B"></th>
57
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:700;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px;background:rgba(253, 142, 91, 0.12)">Ornith-1.0-397B</th>
58
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">Qwen3.5-397B</th>
59
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">Qwen3.7-Max</th>
60
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">GLM-5.2-744B</th>
61
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">Minimax-M3-428B</th>
62
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">DeepSeek-V4-Pro-1.6T</th>
63
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">Claude Opus 4.7</th>
64
+ <th style="width:9.50%;padding:10px 7px;text-align:center;font-weight:500;border-bottom:2px solid #FD8E5B;color:#FD8E5B;font-size:14px">Claude Opus 4.8</th>
65
+ </tr></thead>
66
+ <tbody>
67
+ <tr><td colspan="9" style="padding:8px 12px;font-weight:600;color:#FD8E5B;border-bottom:1px solid rgba(253, 142, 91, 0.2);background:rgba(253, 142, 91, 0.1)">Agentic Coding</td></tr>
68
+ <tr>
69
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">Terminal-Bench 2.1 <sub><small>(Terminus-2)</small></sub></td>
70
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">77.5</td>
71
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">53.5</td>
72
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">73.5</td>
73
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">81.0</td>
74
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">64</td>
75
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">64</td>
76
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">70.3</td>
77
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">85</td>
78
+ </tr>
79
+ <tr>
80
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">Terminal-Bench 2.1 <sub><small>(Claude Code)</small></sub></td>
81
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">78.2</td>
82
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">48.6</td>
83
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">69.8</td>
84
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">82.7</td>
85
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
86
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">66.5</td>
87
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">69.7</td>
88
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">78.9</td>
89
+ </tr>
90
+ <tr>
91
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">SWE-bench Verified</td>
92
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">82.4</td>
93
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">76.4</td>
94
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">80.4</td>
95
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
96
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
97
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">80.6</td>
98
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">80.8</td>
99
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">87.6</td>
100
+ </tr>
101
+ <tr>
102
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">SWE-bench Pro</td>
103
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">62.2</td>
104
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">51.6</td>
105
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">60.6</td>
106
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">62.1</td>
107
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">59</td>
108
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">55.4</td>
109
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">64.3</td>
110
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">69.2</td>
111
+ </tr>
112
+ <tr>
113
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">SWE-bench Multilingual</td>
114
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">78.9</td>
115
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">69.3</td>
116
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">78.3</td>
117
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
118
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
119
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">76.2</td>
120
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
121
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
122
+ </tr>
123
+ <tr>
124
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">NL2Repo</td>
125
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">48.2</td>
126
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">36.8</td>
127
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">47.2</td>
128
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">48.9</td>
129
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">42.1</td>
130
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
131
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
132
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">69.7</td>
133
+ </tr>
134
+ <tr>
135
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">Claw-eval Avg</td>
136
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">77.1</td>
137
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">70.7</td>
138
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">65.2</td>
139
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
140
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
141
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">75.8</td>
142
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">78.2</td>
143
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
144
+ </tr>
145
+ <tr>
146
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">SWE Atlas - QnA</td>
147
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">41.2</td>
148
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">20.4</td>
149
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
150
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
151
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">37.9</td>
152
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">27.2</td>
153
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">40.3</td>
154
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">48.8</td>
155
+ </tr>
156
+ <tr>
157
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">SWE Atlas - RF</td>
158
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">42.6</td>
159
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">18.4</td>
160
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
161
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
162
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
163
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
164
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">48.6</td>
165
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">46.7</td>
166
+ </tr>
167
+ <tr>
168
+ <td style="padding:7px 7px;padding-left:20px;border-bottom:1px solid rgba(128, 128, 128, 0.15);">SWE Atlas - TW</td>
169
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);font-weight:600;color:#FD8E5B;background:rgba(253, 142, 91, 0.06)">39.1</td>
170
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">18.5</td>
171
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
172
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
173
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">30.8</td>
174
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
175
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">38.5</td>
176
+ <td style="padding:7px 7px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15)">-</td>
177
+ </tr>
178
+ </tbody>
179
+ </table>
180
+
181
+ <p style="margin-top:12px;font-size:10px;opacity:0.7">
182
+ * Terminal-Bench 2.1 (Terminus-2): We evaluate Terminal-Bench 2.1 using the Harbor/Terminus-2 framework with parser=json, temperature=1.0, top_p=1.0, and a 128K context window. Each run uses a 4-hour timeout with 32 CPU cores and 48GB RAM, and results are averaged over 5 runs. We adjust the Qwen chat template to ensure consistency between training and inference (https://huggingface.co/deepreinforce-ai/Ornith-1.0-397B/blob/main/chat_template.jinja), and modify Harbor to align with vLLM's reasoning_content key.<br/>
183
+ * Terminal-Bench 2.1 (Claude Code): We evaluate Terminal-Bench 2.1 using Claude Code 2.1.126 with parser=json, temperature=1.0, top_p=1.0, max_new_tokens=131072. Results are averaged over 5 runs. Again, Qwen chat template needs to be modified.<br/>
184
+ * SWE-Bench Verified, Pro and Multilingual: using OpenHands harness with temp=1.0, top_p=0.95, 256k context window.<br/>
185
+ * SWE Atlas QnA, RF, TW: using mini SWE agent harness with temp=1.0, top_p=0.95, 128K context window. Results are averaged over 5 runs.<br/>
186
+ * NL2Repo: with temperature=1.0, top_p=1.0, 400K context, 48K output and anti-hacking filters.<br/>
187
+ * ClawEval: An agentic code benchmark over real-user task distributions; temp=0.6 and 256K context.<br/>
188
+ </p>
189
+
190
+ </div>
191
+
192
+
193
+
194
+
195
+ ## Quickstart
196
+
197
+
198
+ <div style="border-left:4px solid #FD8E5B;background:rgba(253,142,91,0.1);border-radius:6px;padding:12px 16px;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif;font-size:14px;line-height:1.6">
199
+ <div style="font-weight:700;color:#FD8E5B;margin-bottom:6px">📝 NOTE</div>
200
+ <p style="margin:0 0 10px"><b>Ornith-1.0-397B</b> is a <b>reasoning model</b>: by default the assistant turn opens with a <code style="background:rgba(253,142,91,0.15);padding:1px 5px;border-radius:4px">&lt;think&gt; … &lt;/think&gt;</code> block before the final answer. The serving recipes below enable a reasoning parser so the chain-of-thought is returned in a separate <code style="background:rgba(253,142,91,0.15);padding:1px 5px;border-radius:4px">reasoning_content</code> field, and a tool-call parser so the model's <code style="background:rgba(253,142,91,0.15);padding:1px 5px;border-radius:4px">&lt;tool_call&gt;</code> blocks are surfaced as OpenAI-style <code style="background:rgba(253,142,91,0.15);padding:1px 5px;border-radius:4px">tool_calls</code>.</p>
201
+ <p style="margin:0 0 6px">Serving Ornith-1.0-397B requires recent runtimes:</p>
202
+ <ul style="margin:0;padding-left:20px">
203
+ <li><b>Transformers</b> ≥ 5.8.1</li>
204
+ <li><b>vLLM</b> ≥ 0.19.1</li>
205
+ <li><b>SGLang</b> ≥ 0.5.9</li>
206
+ </ul>
207
+ </div>
208
+
209
+ ### Serving Ornith-1.0-397B
210
+
211
+ The two recipes below stand up an OpenAI-compatible server on a single 8×80GB GPU node (tensor-parallel 8). Adjust `--tensor-parallel-size` / `--tp` to the number of GPUs you have.
212
+
213
+ #### vLLM
214
+
215
+ ```bash
216
+ vllm serve deepreinforce-ai/Ornith-1.0-397B \
217
+ --served-model-name Ornith-1.0-397B \
218
+ --tensor-parallel-size 8 \
219
+ --host 0.0.0.0 --port 8000 \
220
+ --max-model-len 262144 \
221
+ --gpu-memory-utilization 0.90 \
222
+ --enable-prefix-caching \
223
+ --enable-auto-tool-choice --tool-call-parser qwen3_xml \
224
+ --reasoning-parser qwen3 \
225
+ --trust-remote-code
226
+ ```
227
+
228
+ #### SGLang
229
+
230
+ ```bash
231
+ python -m sglang.launch_server \
232
+ --model-path deepreinforce-ai/Ornith-1.0-397B \
233
+ --served-model-name Ornith-1.0-397B \
234
+ --tp 8 \
235
+ --host 0.0.0.0 --port 8000 \
236
+ --context-length 262144 \
237
+ --mem-fraction-static 0.85 \
238
+ --tool-call-parser qwen3_coder \
239
+ --reasoning-parser qwen3
240
+ ```
241
+
242
+ #### Hugging Face Transformers
243
+
244
+ For a quick local test (or to script offline generation), load the model directly with Transformers. Make sure you have a recent release installed — see the [Transformers installation guide](https://huggingface.co/docs/transformers/installation); Ornith-1.0-397B requires `transformers >= 5.8.1`.
245
+
246
+ ```python
247
+ from transformers import AutoModelForCausalLM, AutoTokenizer
248
+
249
+ model_name = "deepreinforce-ai/Ornith-1.0-397B"
250
+
251
+ tokenizer = AutoTokenizer.from_pretrained(model_name)
252
+ model = AutoModelForCausalLM.from_pretrained(
253
+ model_name,
254
+ dtype="auto",
255
+ device_map="auto",
256
+ )
257
+
258
+ messages = [
259
+ {"role": "user", "content": "Write a Python function is_prime(n). Keep it short."}
260
+ ]
261
+ text = tokenizer.apply_chat_template(
262
+ messages,
263
+ tokenize=False,
264
+ add_generation_prompt=True,
265
+ )
266
+
267
+ inputs = tokenizer(text, return_tensors="pt").to(model.device)
268
+ generated = model.generate(
269
+ **inputs,
270
+ max_new_tokens=512,
271
+ do_sample=True,
272
+ temperature=0.6,
273
+ top_p=0.95,
274
+ top_k=20,
275
+ )
276
+ output_ids = generated[0][inputs.input_ids.shape[1]:]
277
+
278
+ # The reply contains a <think> ... </think> reasoning block followed by the answer.
279
+ content = tokenizer.decode(output_ids, skip_special_tokens=True)
280
+ print(content)
281
+ ```
282
+
283
+ To split the reasoning trace from the final answer, parse on the `</think>` marker:
284
+
285
+ ```python
286
+ text = tokenizer.decode(output_ids, skip_special_tokens=True)
287
+ if "</think>" in text:
288
+ reasoning, answer = text.split("</think>", 1)
289
+ reasoning = reasoning.replace("<think>", "").strip()
290
+ answer = answer.strip()
291
+ else:
292
+ reasoning, answer = "", text.strip()
293
+ ```
294
+
295
+ ### Using Ornith-1.0-397B via the Chat Completions API
296
+
297
+ Once a vLLM or SGLang server is running, talk to it with any OpenAI-compatible client.
298
+
299
+ #### Basic Usage
300
+
301
+ ```python
302
+ from openai import OpenAI
303
+
304
+ client = OpenAI(
305
+ base_url="http://localhost:8000/v1",
306
+ api_key="EMPTY", # any non-empty string works for a local server
307
+ )
308
+
309
+ response = client.chat.completions.create(
310
+ model="Ornith-1.0-397B",
311
+ messages=[
312
+ {"role": "user", "content": "Write a one-line Python lambda that squares a number."}
313
+ ],
314
+ temperature=0.6,
315
+ top_p=0.95,
316
+ max_tokens=1024,
317
+ )
318
+
319
+ message = response.choices[0].message
320
+ # reasoning_content holds the <think> trace; content holds the final answer.
321
+ print("reasoning:", getattr(message, "reasoning_content", None))
322
+ print("answer:", message.content)
323
+ ```
324
+
325
+ You can also stream tokens, or hand the model tools — Ornith-1.0-397B emits well-formed function calls that the server parses into the standard `tool_calls` field:
326
+
327
+ ```python
328
+ tools = [
329
+ {
330
+ "type": "function",
331
+ "function": {
332
+ "name": "get_weather",
333
+ "description": "Get the current weather for a city",
334
+ "parameters": {
335
+ "type": "object",
336
+ "properties": {"city": {"type": "string"}},
337
+ "required": ["city"],
338
+ },
339
+ },
340
+ }
341
+ ]
342
+
343
+ response = client.chat.completions.create(
344
+ model="Ornith-1.0-397B",
345
+ messages=[{"role": "user", "content": "What is the weather in Paris right now?"}],
346
+ tools=tools,
347
+ tool_choice="auto",
348
+ temperature=0.6,
349
+ max_tokens=2048,
350
+ )
351
+
352
+ tool_call = response.choices[0].message.tool_calls[0]
353
+ print(tool_call.function.name, tool_call.function.arguments)
354
+ # -> get_weather {"city": "Paris"}
355
+ ```
356
+
357
+ You can point any OpenAI-compatible SDK (Python, Node.js, etc.) or `curl` at the same `/v1/chat/completions` endpoint.
358
+
359
+ ## Agentic Usage
360
+
361
+ Ornith-1.0-397B excels in tool-calling and agentic coding capabilities.
362
+
363
+ ### Agent Frameworks
364
+
365
+ Because Ornith-1.0-397B exposes an OpenAI-compatible endpoint with tool calling, it works out of the box with standard agent frameworks. Below is a minimal example that connects Ornith-1.0-397B to tools through an MCP server.
366
+
367
+ ```python
368
+ import os
369
+ from openai import OpenAI
370
+
371
+ client = OpenAI(
372
+ base_url=os.getenv("OPENAI_BASE_URL", "http://localhost:8000/v1"),
373
+ api_key=os.getenv("OPENAI_API_KEY", "EMPTY"),
374
+ )
375
+
376
+ tools = [
377
+ {
378
+ "type": "function",
379
+ "function": {
380
+ "name": "run_shell",
381
+ "description": "Run a shell command and return its output.",
382
+ "parameters": {
383
+ "type": "object",
384
+ "properties": {
385
+ "command": {"type": "string", "description": "The command to run"}
386
+ },
387
+ "required": ["command"],
388
+ },
389
+ },
390
+ }
391
+ ]
392
+
393
+ messages = [{"role": "user", "content": "List the Python files in the current directory."}]
394
+
395
+ response = client.chat.completions.create(
396
+ model="deepreinforce-ai/Ornith-1.0-397B",
397
+ messages=messages,
398
+ tools=tools,
399
+ temperature=0.6,
400
+ top_p=0.95,
401
+ )
402
+ print(response.choices[0].message)
403
+ ```
404
+
405
+ **Examples of using Ornith with agent harness:**
406
+
407
+ #### Hermes Agent
408
+ ```bash
409
+ # Hermes talks to any OpenAI-compatible endpoint — point it at your Ornith server.
410
+ export OPENAI_BASE_URL="http://localhost:8000/v1"
411
+ export OPENAI_API_KEY="EMPTY"
412
+ export MODEL="deepreinforce-ai/Ornith-1.0-397B"
413
+ ```
414
+
415
+ #### OpenClaw
416
+
417
+ ```bash
418
+ # OpenClaw talks to any OpenAI-compatible endpoint — point it at your Ornith server.
419
+ export OPENAI_BASE_URL="http://localhost:8000/v1"
420
+ export OPENAI_API_KEY="EMPTY"
421
+ export OPENAI_MODEL="deepreinforce-ai/Ornith-1.0-397B"
422
+ ```
423
+
424
+ #### Unsloth Studio
425
+
426
+ ```bash
427
+ pip install unsloth
428
+
429
+ # Load Ornith for fast local inference or fine-tuning (Python):
430
+ # from unsloth import FastLanguageModel
431
+ # model, tokenizer = FastLanguageModel.from_pretrained(
432
+ # "deepreinforce-ai/Ornith-1.0-397B",
433
+ # max_seq_length=262144,
434
+ # load_in_4bit=True,
435
+ # )
436
+ ```
437
+
438
+ #### OpenHands
439
+ ```bash
440
+ pip install openhands-ai
441
+
442
+ # OpenHands routes through LiteLLM; the "openai/" prefix selects the OpenAI-compatible path.
443
+ export LLM_MODEL="openai/deepreinforce-ai/Ornith-1.0-397B"
444
+ export LLM_BASE_URL="http://localhost:8000/v1"
445
+ export LLM_API_KEY="EMPTY"
446
+
447
+ # Launch the CLI (or run the official OpenHands Docker image with the same env vars).
448
+ openhands
449
+ ```
450
+
451
+ ### Coding CLIs
452
+
453
+ Ornith-1.0-397B is optimized for terminal-based coding agents. Point any OpenAI-compatible coding CLI at your Ornith-1.0-397B endpoint (set `OPENAI_BASE_URL` and `OPENAI_API_KEY`) to understand large codebases, automate tedious work, and ship faster.
454
+
455
+ #### OpenCode
456
+ ```bash
457
+ # Register your local Ornith endpoint as a provider in ~/.config/opencode/opencode.json:
458
+ #
459
+ # {
460
+ # "$schema": "https://opencode.ai/config.json",
461
+ # "provider": {
462
+ # "ornith": {
463
+ # "npm": "@ai-sdk/openai-compatible",
464
+ # "name": "Ornith (local)",
465
+ # "options": { "baseURL": "http://localhost:8000/v1", "apiKey": "EMPTY" },
466
+ # "models": { "deepreinforce-ai/Ornith-1.0-397B": { "name": "Ornith-1.0-397B" } }
467
+ # }
468
+ # }
469
+ # }
470
+
471
+ opencode
472
+ ```
473
+
474
+
475
+ ### Citation
476
+
477
+ If you find our work helpful, feel free to give us a cite.
478
+
479
+ ```bibtex
480
+ @misc{ornith_397b,
481
+ title = {{Ornith-1.0-397B}: Agentic Coding, Open to All},
482
+ url = {https://deep-reinforce.com/ornith_1_0.html},
483
+ author = {{DeepReinforce Team}},
484
+ year = {2026}
485
+ }
486
+ ```
assets/ornith_397b_eval.png ADDED

Git LFS Details

  • SHA256: 0815e9a86f41a63c7c18593b73bf5c5a8929479bf974960ce58dd34266bd7c88
  • Pointer size: 131 Bytes
  • Size of remote file: 585 kB
assets/ornith_logo.png ADDED

Git LFS Details

  • SHA256: 458ee0d85baea4d1fc0b099245f243b21006f696921f896a5b59980a6eadef6a
  • Pointer size: 131 Bytes
  • Size of remote file: 962 kB
chat_template.jinja ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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 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 mapping %}
120
+ {%- for args_name in tool_call.arguments %}
121
+ {%- set args_value = tool_call.arguments[args_name] %}
122
+ {{- '<parameter=' + args_name + '>\n' }}
123
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
124
+ {{- args_value }}
125
+ {{- '\n</parameter>\n' }}
126
+ {%- endfor %}
127
+ {%- endif %}
128
+ {{- '</function>\n</tool_call>' }}
129
+ {%- endfor %}
130
+ {%- endif %}
131
+ {{- '<|im_end|>\n' }}
132
+ {%- elif message.role == "tool" %}
133
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
134
+ {{- '<|im_start|>user' }}
135
+ {%- endif %}
136
+ {{- '\n<tool_response>\n' }}
137
+ {{- content }}
138
+ {{- '\n</tool_response>' }}
139
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
140
+ {{- '<|im_end|>\n' }}
141
+ {%- elif loop.last %}
142
+ {{- '<|im_end|>\n' }}
143
+ {%- endif %}
144
+ {%- else %}
145
+ {{- raise_exception('Unexpected message role.') }}
146
+ {%- endif %}
147
+ {%- endfor %}
148
+ {%- if add_generation_prompt %}
149
+ {{- '<|im_start|>assistant\n' }}
150
+ {%- if enable_thinking is defined and enable_thinking is false %}
151
+ {{- '<think>\n\n</think>\n\n' }}
152
+ {%- else %}
153
+ {{- '<think>\n' }}
154
+ {%- endif %}
155
+ {%- endif %}
config.json ADDED
@@ -0,0 +1,145 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5MoeForConditionalGeneration"
4
+ ],
5
+ "torch_dtype": "bfloat16",
6
+ "hidden_size": 4096,
7
+ "image_token_id": 248056,
8
+ "model_type": "qwen3_5_moe",
9
+ "pad_token_id": 248055,
10
+ "text_config": {
11
+ "attention_bias": false,
12
+ "attention_dropout": 0.0,
13
+ "attn_output_gate": true,
14
+ "bos_token_id": null,
15
+ "torch_dtype": "bfloat16",
16
+ "eos_token_id": 248044,
17
+ "full_attention_interval": 4,
18
+ "head_dim": 256,
19
+ "hidden_act": "silu",
20
+ "hidden_size": 4096,
21
+ "initializer_range": 0.02,
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
+ ],
84
+ "linear_conv_kernel_dim": 4,
85
+ "linear_key_head_dim": 128,
86
+ "linear_num_key_heads": 16,
87
+ "linear_num_value_heads": 64,
88
+ "linear_value_head_dim": 128,
89
+ "mamba_ssm_dtype": "float32",
90
+ "max_position_embeddings": 262144,
91
+ "mlp_only_layers": [],
92
+ "model_type": "qwen3_5_moe_text",
93
+ "moe_intermediate_size": 1024,
94
+ "mtp_num_hidden_layers": 1,
95
+ "mtp_use_dedicated_embeddings": false,
96
+ "num_attention_heads": 32,
97
+ "num_experts": 512,
98
+ "num_experts_per_tok": 10,
99
+ "num_hidden_layers": 60,
100
+ "num_key_value_heads": 2,
101
+ "output_router_logits": false,
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
+ "router_aux_loss_coef": 0.001,
117
+ "shared_expert_intermediate_size": 1024,
118
+ "tie_word_embeddings": false,
119
+ "use_cache": true,
120
+ "vocab_size": 248320
121
+ },
122
+ "tie_word_embeddings": false,
123
+ "transformers_version": "5.15.0.dev0",
124
+ "unsloth_fixed": true,
125
+ "video_token_id": 248057,
126
+ "vision_config": {
127
+ "deepstack_visual_indexes": [],
128
+ "depth": 27,
129
+ "torch_dtype": "bfloat16",
130
+ "hidden_act": "gelu_pytorch_tanh",
131
+ "hidden_size": 1152,
132
+ "in_channels": 3,
133
+ "initializer_range": 0.02,
134
+ "intermediate_size": 4304,
135
+ "model_type": "qwen3_5_moe_vision",
136
+ "num_heads": 16,
137
+ "num_position_embeddings": 2304,
138
+ "out_hidden_size": 4096,
139
+ "patch_size": 16,
140
+ "spatial_merge_size": 2,
141
+ "temporal_patch_size": 2
142
+ },
143
+ "vision_end_token_id": 248054,
144
+ "vision_start_token_id": 248053
145
+ }
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": 0.6,
10
+ "top_k": 20,
11
+ "top_p": 0.95,
12
+ "transformers_version": "4.57.0.dev0"
13
+ }
model-00001-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aabf8ec90ddf7cbe332dfdf6807480c09f623a5560f2f19b43a326a6778c307b
3
+ size 3211703200
model-00002-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9ba68cbd3b48155e4cf277436e39cea26abba86ee20a59749a8d8088a4a38178
3
+ size 8589934736
model-00003-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:821b56fc0d15ed955a6548303e8e6dd97ddd0bfd079acec45a8c379adfbef6f1
3
+ size 4560382680
model-00004-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d55c91ab44d970b2dd5c22de683f4ea440bd4d7b1a4684012bde371c74454dc
3
+ size 8589934736
model-00005-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0048441e741c84c0cb76fd13bf0027492935c42833c81f82fd41ce9baf2139b
3
+ size 4560382680
model-00006-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7659556392190b06af86644c028228c092862f9f7cabe5ddba80840deeac6770
3
+ size 8589934736
model-00007-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba40bd1ec549a76208991589cc452672b463ea34d60c03e6f02b9b8336b89fbe
3
+ size 4534070088
model-00008-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ee38622f02482efbf5de34cceaf01f3305c3680ee05674374d895119acd24ab
3
+ size 8589934736
model-00009-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:febdc39e6260617e6e20ce85ddaf60fdc761e4d3786eef6d45d52b53334cb2d0
3
+ size 4560382680
model-00010-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cad0ec5c37788975aa75458cb83bb7ada0b6099360184e016f615963ab5a2c35
3
+ size 8589934736
model-00011-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c32be779b30d7dfcb84eb2fc5b99d5523d6a44f22967afc2388f581f03c7c672
3
+ size 4560382680
model-00012-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:058586e3a825e8d5f9993faa4c7b102d809f2bf791257ce548af5fa6f6c7cc4e
3
+ size 8589934736
model-00013-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4694938ea71956d58db5668dd9d1ab0e4a92296b9039fd258f394b693b982a33
3
+ size 4560382680
model-00014-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e2a5ff5794bd6134aa6e1c1d10ba01c6a378f7783068785f0a69e9e890b6294
3
+ size 8589934736
model-00015-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:122ab40c8b23eacb977027a8925afab5c632e9d7785657bcf298ff7ff4a76c60
3
+ size 4534070088
model-00016-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28a09397dcb6d887946351b2e0f2f8456342b2c11a73361a4ad7b15f0ed2bc53
3
+ size 8589934736
model-00017-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:986a7beff2236575baf8b3f1192c7a8118c8d84213e475958372fd3576d5351e
3
+ size 4560382680
model-00018-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:677d2a7f008f739ff2b2824bd254cc1b7ed17afc4728539ee973700084213ecc
3
+ size 8589934736
model-00019-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:147e79b095b5f7c24d4c31de402011030b2618af51976facffd9a6143b981c1a
3
+ size 4560382680
model-00020-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6041117e937175cbbe0b1289bcb1031a79266e60ffd497721bf73b24922d35b
3
+ size 8589934736
model-00021-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9ea150d9daecafde24735d645dc0d48c146aaff81e8085188b18387385c24b8
3
+ size 4560382624
model-00022-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56ddb5354a38e58a94db23aaf32570db6912203e3aab7aed55e31245887dce33
3
+ size 8589934736
model-00023-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6182149887dedb93b971fa717400d921f45368a749365ca38891e221c21a52b
3
+ size 4534070104
model-00024-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:520dc122c4165c0983b3bc6f12ef8756ea511fc36c873c4988d7ff003ef980a9
3
+ size 8589934736
model-00025-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3dfe447aa38d1b15d7dc2d6610d047bff09ce1e5f036d69251c6f28f7051c5c7
3
+ size 4560382696
model-00026-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0bb6a9ba9f21a638085ca4fde13238eb4c2b956ed16a368b72d9d18aff743e9
3
+ size 8589934736
model-00027-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4299994b59365817815c25e8d4c3a789b31d4827a5beaed340e8005018d7e6b5
3
+ size 4560382696
model-00028-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b79438707babbb0536d7a96b2916c6b2d2842653869a549eab4dc88bbdf6a22
3
+ size 8589934736
model-00029-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64f8361dc0c2430a1df72304a067248330ba84165899d1233bd0742c193c93e1
3
+ size 4560382696
model-00030-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ceba78a436237656b5bb3ec8ef7ce30132dc1f0a4771ed3fc4e420545226d73
3
+ size 8589934736
model-00031-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:644f2cc0cfa0ab1695fe7979a59d61589c3630310fb7ac72e3fd6ced2474dc81
3
+ size 4534070104
model-00032-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3ecce2e4a448e18b57ef2c44de41e295761b9878c6478d125c074e7c5e07d67
3
+ size 8589934736
model-00033-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fed7ca1d0c0f8d13652c73a146ba10baa18ca225bf10e896282dfe60928c8a2b
3
+ size 4560382696
model-00034-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc6c9e8f1e65dbc5ec6e9976bc6a4a92e9bbf173c9b6909c05ca567ea9e6aa30
3
+ size 8589934736
model-00035-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0cc4ac67286d6891c782f85d1321837f577b3bb34fd33ffc4ace11f1e6cc8ad
3
+ size 4560382696
model-00036-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:691e0aee30af8210973be5aae3554a60fdc367175e7bf5367d53ff81504395cd
3
+ size 8589934736
model-00037-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12682047bcecf91864f145ddf2f1eaca4e6b1c820a879a620257c4df87ae0874
3
+ size 4560382696
model-00038-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74feedc0f48fc88c167373cba1cb18b7776cc54239803d277b617dc977d0783f
3
+ size 8589934736
model-00039-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:96c1c87b52c7953c7543aa719d0ed8bd99195905c476376cb613c367ad57bde3
3
+ size 4534070104
model-00040-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b1f46a8d99396c72c0c61ccac37e4bc7d8432667031a99987c33e25371cbd9e
3
+ size 8589934736
model-00041-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:916b1a4ce9474db5f6421fa4a01492e2e0d909f4e3b142ca1b00ad78ab3a41a0
3
+ size 4560382696
model-00042-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a18051f2fbbc5105710a56db2c5f4a6ecc551bc57eed3862fda33e3e8e0339db
3
+ size 8589934736
model-00043-of-00122.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:672abdfb2ce6042b996a45af606253c5d5b347c299ea3a896a667d8a22ef39c0
3
+ size 4560382696