Encoder-Guard v3 int8: int8-quantized backbone (heads/threshold identical to v3)

.gitattributes

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

LICENSE

@@ -0,0 +1,125 @@
+Encoder-Guard 模型许可说明
+==========================
+
+本模型为**分组件许可**（component-wise licensing）作品：不同组成部分适用不同
+许可条款，**不是**"二选一"式的双许可（dual licensing）。使用某一组成部分时，
+仅需遵守该部分对应的条款；但分发整个模型时，须同时满足全部适用条款。
+分发或使用本模型前，请完整阅读本文件及所附 `LICENSE-APACHE-2.0` 全文。
+
+--------------------------------------------------------------------------------
+一、组成部分与适用许可
+--------------------------------------------------------------------------------
+
+1. backbone 权重（`backbone/model.safetensors`、`backbone/tokenizer.json`、
+   `backbone/config.json`、`backbone/chat_template.jinja` 等）
+
+   来源：Qwen 系列基座模型
+     - 0.6B 线：Qwen/Qwen3Guard-Stream-0.6B
+     - 4B 线：  Qwen/Qwen3Guard-Stream-4B
+     - v1 线：  Qwen/Qwen2.5-0.5B
+
+   许可：**Apache License 2.0**
+   全文：见本仓库 `LICENSE-APACHE-2.0`
+
+   依据 Apache-2.0 第 4(a) 条，任何再分发必须随附该许可全文。
+
+2. 分类头与训练产物（`heads.safetensors`、`threshold.json`、
+   `model_config.json`、`heads_meta.json`、`train_log.jsonl`、`data_card.json`）
+
+   许可：**Encoder-Guard Non-Commercial License 1.0**
+   全文：见本文件第二节
+
+3. 模型卡与文档（`README.md` 等）
+
+   许可：CC BY-NC 4.0（署名—非商业性使用 4.0 国际）
+   全文：https://creativecommons.org/licenses/by-nc/4.0/legalcode
+
+--------------------------------------------------------------------------------
+二、Encoder-Guard Non-Commercial License 1.0
+--------------------------------------------------------------------------------
+
+定义
+  "本作品"指上述第一节第 2 项与第 3 项下的分类头、训练产物与文档。
+  "商业使用"指以营利为目的的使用，包括但不限于：作为付费产品或服务的
+  组成部分、用于向第三方提供收费服务、用于内部生产环境的商业决策流程
+  且该流程直接产生营收。
+
+授权
+  在遵守本许可的前提下，你被授予非独占、全球范围、免许可费的权利，
+  可以：使用、复制、修改、合并、发布、分发本作品及其衍生作品。
+
+条件
+  (a) 署名：必须在衍生作品中保留本许可文件与原始出处说明。
+  (b) 非商业：不得将本作品用于商业使用。
+  (c) 相同方式共享：对本作品的修改与衍生作品，须以本许可或与之兼容的
+      非商业许可分发。
+  (d) 无额外限制：不得对本作品施加法律条款或技术措施，限制他人行使
+      本许可授予的权利。
+  (e) 基座许可保留：再分发时须同时随附 `LICENSE-APACHE-2.0` 全文，
+      并保留 Qwen 基座的版权、专利与商标声明（Apache-2.0 第 4(a)(c) 条）。
+
+商业授权
+  如需商业使用，须取得单独书面授权。联系方式见模型卡。
+
+终止
+  违反本许可条款时，本许可自动终止。终止后须立即停止使用并销毁所有副本。
+
+免责声明
+  本作品按"现状"提供，不附带任何明示或暗示的担保，包括但不限于
+  适销性、特定用途适用性与非侵权担保。在任何情况下，作者或版权持有人
+  均不对因本作品或本作品的使用而产生的任何索赔、损害或其他责任负责。
+
+--------------------------------------------------------------------------------
+三、修改说明（Apache-2.0 第 4(b) 条要求）
+--------------------------------------------------------------------------------
+
+本模型对基座模型做了以下修改，特此显著声明：
+
+1. **注意力机制改造**：将基座的 causal（因果）注意力掩码改为**全可见
+   双向注意力**（bidirectional attention）。推理时无 KV cache，所有
+   token 互相可见。这是架构层面的修改，改变了模型的输出语义。
+
+2. **新增分类头**：在 backbone 的 pooled 表示之上新增 11 个线性分类头
+   （safety_release + TC260 大类/小类 + 语言风险 LM + 个人信息 PII +
+   数据分级 DC），用于一次前向输出完整安全画像。
+
+3. **重新训练**：分类头在自有标注数据集上训练，backbone 权重亦经过
+   微调（非原始 Qwen 权重）。
+
+4. **阈值校准**：`threshold.json` 中的 safety 阈值基于自有验证集在
+   FPR 约束下校准，与基座模型无关。
+
+上述修改使本模型成为 Apache-2.0 定义下的 "Derivative Work"。
+
+--------------------------------------------------------------------------------
+四、合规提示
+--------------------------------------------------------------------------------
+
+本模型为内容安全审核工具，输出为分类概率与标签，不产出自然语言。
+使用者应自行确保下游应用符合所在地法律法规（包括但不限于
+《个人信息保护法》《数据安全法》《生成���人工智能服务管理暂行办法》）。
+模型作者不对下游使用承担合规责任。
+
+--------------------------------------------------------------------------------
+五、文件清单
+--------------------------------------------------------------------------------
+
+  LICENSE                 本文件（许可总览 + 非商业许可全文）
+  LICENSE-APACHE-2.0      Apache License 2.0 全文（基座部分适用）
+
+--------------------------------------------------------------------------------
+六、版权与联系方式
+--------------------------------------------------------------------------------
+
+分类头、训练产物与文档：
+  Copyright (c) 2026 Louivis Lee
+
+基座权重（Apache-2.0 部分）的版权归其原始权利人所有，具体声明见上游仓库：
+  https://huggingface.co/Qwen/Qwen3Guard-Stream-0.6B
+  https://huggingface.co/Qwen/Qwen3Guard-Stream-4B
+  https://huggingface.co/Qwen/Qwen2.5-0.5B
+  上游仓库未提供 NOTICE 文件，故本仓库不另行附加通知。
+
+商业授权与许可咨询：
+  Hugging Face: https://huggingface.co/louivis
+  GitHub:       https://github.com/vislee/verdict/issues

LICENSE-APACHE-2.0

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2024 Alibaba Cloud
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

backbone/chat_template.jinja

@@ -0,0 +1,89 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

backbone/config.json

@@ -0,0 +1,98 @@
+{
+  "architectures": [
+    "Qwen3Model"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "guard_inner_size": 512,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_category": 8,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "num_query_category": 9,
+  "num_query_risk_level": 3,
+  "num_risk_level": 3,
+  "query_category_map": {
+    "0": "Violent",
+    "1": "Sexual Content",
+    "2": "Self-Harm",
+    "3": "Political",
+    "4": "PII",
+    "5": "Copyright",
+    "6": "Illegal Acts",
+    "7": "Unethical",
+    "8": "Jailbreak"
+  },
+  "query_risk_level_map": {
+    "0": "Safe",
+    "1": "Unsafe",
+    "2": "Controversial"
+  },
+  "response_category_map": {
+    "0": "Violent",
+    "1": "Sexual Content",
+    "2": "Self-Harm",
+    "3": "Political",
+    "4": "PII",
+    "5": "Copyright",
+    "6": "Illegal Acts",
+    "7": "Unethical"
+  },
+  "response_risk_level_map": {
+    "0": "Safe",
+    "1": "Unsafe",
+    "2": "Controversial"
+  },
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.0.0rc1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

backbone/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:354dc5f733d600a6a9ede69513e63854ba83abbfaf112d2ae147e7fb69c079a5
+size 598157712

backbone/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05d47c87966b4db779200053de490f89936ed529f8ab889244e271630715fcfe
+size 11422638

backbone/tokenizer_config.json

@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "additional_special_tokens": null,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": true,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

data_card.json

@@ -0,0 +1,67 @@
+{
+  "date": "2026-10-23",
+  "train_file": "data/dataset/finetune_final_v6_35_train.jsonl",
+  "train_md5": "8490783692aac4090a87003f89f78795",
+  "train_n": 28906,
+  "val_n": 1595,
+  "preflight": [
+    {
+      "check": "train 归一化重复=0",
+      "pass": true,
+      "detail": "max dup 1"
+    },
+    {
+      "check": "unsafe 行 tc260 null=0",
+      "pass": true,
+      "detail": "残留 0"
+    },
+    {
+      "check": "unsafe 三轨全空=0",
+      "pass": true,
+      "detail": "残留 0"
+    },
+    {
+      "check": "train∩frozen test=0",
+      "pass": true,
+      "detail": "重叠 0（hard-negative 豁免 14，tag=_hard_negative 留痕）"
+    },
+    {
+      "check": "safe 锚点在位",
+      "pass": true,
+      "detail": "train 天安门 safe=85（≥20）"
+    },
+    {
+      "check": "val 文件在位",
+      "pass": true,
+      "detail": "data/dataset/finetune_final_v6_31_val.jsonl"
+    },
+    {
+      "check": "epochs ∈ [1,20]",
+      "pass": true,
+      "detail": "epochs=20"
+    },
+    {
+      "check": "lr 合理域",
+      "pass": true,
+      "detail": "lr=2e-05"
+    },
+    {
+      "check": "dtype 已声明",
+      "pass": true,
+      "detail": "dtype=bf16"
+    },
+    {
+      "check": "独立校准集已配置",
+      "pass": true,
+      "detail": "data/dataset/calib_v23_independent.jsonl"
+    }
+  ],
+  "hyperparams": {
+    "epochs": 20,
+    "batch_size": 32,
+    "grad_accumulation": 1,
+    "learning_rate": 2e-05,
+    "dtype": "bf16",
+    "base_model": "/home/liwenqiang19/models/Qwen/Qwen3Guard-Stream-0.6B/backbone"
+  }
+}

heads.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7549e611eaa8c44e4fe5b034b0059b31b660311f1c92a6ca08e72f00a6e09c85
+size 666088

heads_meta.json

@@ -0,0 +1,16 @@
+{
+  "head_order": [
+    "safety",
+    "tc260_major",
+    "tc260_minor",
+    "lm_major",
+    "lm_minor",
+    "pii_major",
+    "pii_minor",
+    "dc_type",
+    "dc_level",
+    "dc_minor",
+    "safety_release"
+  ],
+  "hidden_size": 1024
+}

model_config.json

@@ -0,0 +1 @@
+{"hidden_size": 1024, "is_decoder": false, "dropout": 0.1, "head_dims": {"tc260_major": 5, "lm_major": 10, "pii_major": 7, "dc_type": 4, "dc_level": 3, "safety": 1, "safety_release": 1, "tc260_minor": 31, "lm_minor": 54, "pii_minor": 10, "dc_minor": 36}}

threshold.json

@@ -0,0 +1,84 @@
+{
+ "safety_threshold": 0.96630859375,
+ "method": "fpr_constrained_max_recall",
+ "target_fpr": 0.02,
+ "calibrated_on": "data/dataset/calib_v23_independent.jsonl (9394 \u6761)",
+ "safe_n": 8194,
+ "unsafe_n": 1200,
+ "overlapped": true,
+ "degraded_recall": false,
+ "release_gate": 0.9416,
+ "release_stats": {
+  "safe_n": 512,
+  "unsafe_n": 1083,
+  "safe_p50": 0.9702,
+  "safe_p90": 0.9714,
+  "safe_p95": 0.9729,
+  "safe_p99": 0.9751,
+  "unsafe_p99": 0.9416,
+  "unsafe_max": 0.9711,
+  "release_gate": 0.9416
+ },
+ "temperature": {
+  "tc260_major": 1.8249,
+  "lm_major": 1.4312,
+  "pii_major": 1.2654,
+  "dc_type": 1.1384,
+  "dc_level": 1.46,
+  "safety": 0.7925,
+  "safety_release": 0.849,
+  "tc260_minor": 1.236,
+  "lm_minor": 1.0025,
+  "pii_minor": 1.2188,
+  "dc_minor": 0.5766
+ },
+ "ece": {
+  "tc260_major": {
+   "raw": 0.0098,
+   "calibrated": 0.0114
+  },
+  "lm_major": {
+   "raw": 0.0026,
+   "calibrated": 0.0022
+  },
+  "pii_major": {
+   "raw": 0.0006,
+   "calibrated": 0.0007
+  },
+  "dc_type": {
+   "raw": 0.0926,
+   "calibrated": 0.1004
+  },
+  "dc_level": {
+   "raw": 0.0048,
+   "calibrated": 0.0019
+  },
+  "safety": {
+   "raw": 0.1144,
+   "calibrated": 0.0836
+  },
+  "safety_release": {
+   "raw": 0.045,
+   "calibrated": 0.0382
+  },
+  "tc260_minor": {
+   "raw": 0.0024,
+   "calibrated": 0.0029
+  },
+  "lm_minor": {
+   "raw": 0.0006,
+   "calibrated": 0.0006
+  },
+  "pii_minor": {
+   "raw": 0.0007,
+   "calibrated": 0.0011
+  },
+  "dc_minor": {
+   "raw": 0.0205,
+   "calibrated": 0.0059
+  }
+ },
+ "safety_saturation": 0.0284,
+ "model_version": "encoder-guard-v23-06b-v36-fp16-recal",
+ "recal_note": "fp16 domain recal, calib_v23_independent(9394), FPR 1.95% recall 0.9017"
+}

train_log.jsonl

@@ -0,0 +1,20 @@
+{"epoch": 1, "train_loss": {"tc260_major": 0.758092245472758, "lm_major": 0.7403016344026342, "pii_major": 0.38046099107861026, "dc_type": 1.054102013401217, "dc_level": 1.0086026838439148, "safety": 0.49747449951710493, "safety_release": 0.7591457258031218, "tc260_minor": 0.7674269323564183, "lm_minor": 0.5444553284962576, "pii_minor": 0.5699814716137426, "dc_minor": 0.9656584325431127, "total": 6.09046980946563}, "val_main_f1": 0.6883630007505417, "val": {"tc260_major": {"f1_macro": 0.6576116681098938, "f1_micro": 0.7102941274642944, "precision": 0.5341168642044067, "recall": 0.9209467172622681}, "lm_major": {"f1_macro": 0.41249561309814453, "f1_micro": 0.4711926579475403, "precision": 0.2935391962528229, "recall": 0.8828310966491699}, "pii_major": {"f1_macro": 0.7330341339111328, "f1_micro": 0.6924564838409424, "precision": 0.602540135383606, "recall": 0.9447416067123413}, "dc_type": {"f1_macro": 0.4497087597846985, "f1_micro": 0.6865439414978027, "precision": 0.398434579372406, "recall": 0.8840588331222534}, "dc_level": {"f1_macro": 0.5524417757987976, "f1_micro": 0.8590308427810669, "precision": 0.46338844299316406, "recall": 0.9226503372192383}, "safety": {"f1_macro": 0.9503105878829956, "f1_micro": 0.9503105878829956, "precision": 0.9146028757095337, "recall": 0.9889196753501892}, "safety_release": {"f1_macro": 0.9246435761451721, "f1_micro": 0.9246435761451721, "precision": 0.9659574627876282, "recall": 0.88671875}}, "val_minor": {"tc260_minor": 0.24749985337257385, "lm_minor": 0.12085206806659698, "pii_minor": 0.5845909118652344, "dc_minor": 0.0636499896645546}, "lm_per_class": {"LM.01": 0.2002, "LM.02": 0.1646, "LM.03": 0.0245, "LM.04": 0.1071, "LM.05": 0.0204, "LM.06": 0.1322, "LM.07": 0.1738, "LM.08": 0.0, "LM.09": 0.1566, "LM.10": 0.288}}
+{"epoch": 2, "train_loss": {"tc260_major": 0.38394517166868003, "lm_major": 0.34290560400928244, "pii_major": 0.10279338685862703, "dc_type": 0.794548470799214, "dc_level": 0.765808315713532, "safety": 0.31311450855742523, "safety_release": 0.49327510238881916, "tc260_minor": 0.30627987693875597, "lm_minor": 0.16948357790890062, "pii_minor": 0.13457187516398209, "dc_minor": 0.4317039197366499, "total": 3.5236538524511514}, "val_main_f1": 0.736924484372139, "val": {"tc260_major": {"f1_macro": 0.7085622549057007, "f1_micro": 0.768018901348114, "precision": 0.6021846532821655, "recall": 0.9319807291030884}, "lm_major": {"f1_macro": 0.532995343208313, "f1_micro": 0.644444465637207, "precision": 0.41843247413635254, "recall": 0.845067024230957}, "pii_major": {"f1_macro": 0.7468134760856628, "f1_micro": 0.7412008047103882, "precision": 0.6333573460578918, "recall": 0.9454230666160583}, "dc_type": {"f1_macro": 0.48528724908828735, "f1_micro": 0.7145110368728638, "precision": 0.44471442699432373, "recall": 0.8448271751403809}, "dc_level": {"f1_macro": 0.7203367352485657, "f1_micro": 0.9071742296218872, "precision": 0.6251416802406311, "recall": 0.9618088603019714}, "safety": {"f1_macro": 0.9593268632888794, "f1_micro": 0.9593268036842346, "precision": 0.9715909361839294, "recall": 0.9473684430122375}, "safety_release": {"f1_macro": 0.8951175808906555, "f1_micro": 0.8951175212860107, "precision": 0.8333333134651184, "recall": 0.966796875}}, "val_minor": {"tc260_minor": 0.36827194690704346, "lm_minor": 0.2147916853427887, "pii_minor": 0.6776875853538513, "dc_minor": 0.1107974424958229}, "lm_per_class": {"LM.01": 0.2778, "LM.02": 0.2533, "LM.03": 0.0432, "LM.04": 0.1802, "LM.05": 0.2146, "LM.06": 0.2628, "LM.07": 0.2655, "LM.08": 0.1053, "LM.09": 0.2727, "LM.10": 0.3597}}
+{"epoch": 3, "train_loss": {"tc260_major": 0.29411317813411303, "lm_major": 0.3004877970493084, "pii_major": 0.08219299075484549, "dc_type": 0.6232223919887252, "dc_level": 0.6059549931453021, "safety": 0.2736478620028311, "safety_release": 0.43573499550983097, "tc260_minor": 0.22796208918886723, "lm_minor": 0.12275342548375377, "pii_minor": 0.09183323222489931, "dc_minor": 0.394636486629117, "total": 2.8766810350111816}, "val_main_f1": 0.789372593164444, "val": {"tc260_major": {"f1_macro": 0.7569965124130249, "f1_micro": 0.8025837540626526, "precision": 0.6674060821533203, "recall": 0.9261128306388855}, "lm_major": {"f1_macro": 0.5436756610870361, "f1_micro": 0.6402349472045898, "precision": 0.4204499125480652, "recall": 0.9284408688545227}, "pii_major": {"f1_macro": 0.8811636567115784, "f1_micro": 0.8491879105567932, "precision": 0.814405620098114, "recall": 0.9659484028816223}, "dc_type": {"f1_macro": 0.5718212723731995, "f1_micro": 0.8207677602767944, "precision": 0.5014762878417969, "recall": 0.8223738074302673}, "dc_level": {"f1_macro": 0.7646914124488831, "f1_micro": 0.9305927753448486, "precision": 0.6770543456077576, "recall": 0.9508709907531738}, "safety": {"f1_macro": 0.9756545424461365, "f1_micro": 0.9756545424461365, "precision": 0.97074955701828, "recall": 0.9806094169616699}, "safety_release": {"f1_macro": 0.9415148496627808, "f1_micro": 0.9415148496627808, "precision": 0.9246704578399658, "recall": 0.958984375}}, "val_minor": {"tc260_minor": 0.39563611149787903, "lm_minor": 0.30083778500556946, "pii_minor": 0.7644408345222473, "dc_minor": 0.17756633460521698}, "lm_per_class": {"LM.01": 0.3133, "LM.02": 0.345, "LM.03": 0.0898, "LM.04": 0.2186, "LM.05": 0.3636, "LM.06": 0.3202, "LM.07": 0.3281, "LM.08": 0.1218, "LM.09": 0.3529, "LM.10": 0.6838}}
+{"epoch": 4, "train_loss": {"tc260_major": 0.22399472532915582, "lm_major": 0.2406415539202698, "pii_major": 0.07512919456816092, "dc_type": 0.42749918767995976, "dc_level": 0.5183002308565027, "safety": 0.2540809740028772, "safety_release": 0.40515635833391717, "tc260_minor": 0.18376052645287771, "lm_minor": 0.09520148439075002, "pii_minor": 0.06728479432989973, "dc_minor": 0.3492829728413142, "total": 2.3608405071239535}, "val_main_f1": 0.8097322583198547, "val": {"tc260_major": {"f1_macro": 0.8076328039169312, "f1_micro": 0.8380187153816223, "precision": 0.7497938275337219, "recall": 0.8954001665115356}, "lm_major": {"f1_macro": 0.5907483696937561, "f1_micro": 0.6743335127830505, "precision": 0.4621737003326416, "recall": 0.9070709943771362}, "pii_major": {"f1_macro": 0.8759419322013855, "f1_micro": 0.8503562808036804, "precision": 0.8144739270210266, "recall": 0.9530898332595825}, "dc_type": {"f1_macro": 0.5556483864784241, "f1_micro": 0.8119999766349792, "precision": 0.5036564469337463, "recall": 0.841524064540863}, "dc_level": {"f1_macro": 0.7748231291770935, "f1_micro": 0.9406392574310303, "precision": 0.701328456401825, "recall": 0.902916431427002}, "safety": {"f1_macro": 0.9646059274673462, "f1_micro": 0.9646059274673462, "precision": 0.9864864945411682, "recall": 0.9436749815940857}, "safety_release": {"f1_macro": 0.9118982553482056, "f1_micro": 0.9118982553482056, "precision": 0.852292001247406, "recall": 0.98046875}}, "val_minor": {"tc260_minor": 0.4412696361541748, "lm_minor": 0.33413150906562805, "pii_minor": 0.8439580202102661, "dc_minor": 0.24636498093605042}, "lm_per_class": {"LM.01": 0.3282, "LM.02": 0.2671, "LM.03": 0.1577, "LM.04": 0.3624, "LM.05": 0.3452, "LM.06": 0.3852, "LM.07": 0.3902, "LM.08": 0.1444, "LM.09": 0.4981, "LM.10": 0.6638}}
+{"epoch": 5, "train_loss": {"tc260_major": 0.19769545321652907, "lm_major": 0.21824905624281066, "pii_major": 0.06512119443785473, "dc_type": 0.329025921425349, "dc_level": 0.35687049548876504, "safety": 0.2421631467335776, "safety_release": 0.38554868791850566, "tc260_minor": 0.1512884001968509, "lm_minor": 0.08288793874118613, "pii_minor": 0.05436602010105965, "dc_minor": 0.30656286822668055, "total": 1.9787244275294797}, "val_main_f1": 0.8333816230297089, "val": {"tc260_major": {"f1_macro": 0.8428663015365601, "f1_micro": 0.8679569959640503, "precision": 0.7909826040267944, "recall": 0.9182350039482117}, "lm_major": {"f1_macro": 0.6394157409667969, "f1_micro": 0.7304843068122864, "precision": 0.5249197483062744, "recall": 0.8989163637161255}, "pii_major": {"f1_macro": 0.8754785656929016, "f1_micro": 0.8421052694320679, "precision": 0.7981395125389099, "recall": 0.977708637714386}, "dc_type": {"f1_macro": 0.5580927133560181, "f1_micro": 0.820574164390564, "precision": 0.48951706290245056, "recall": 0.7383462190628052}, "dc_level": {"f1_macro": 0.7783380150794983, "f1_micro": 0.94669508934021, "precision": 0.7077553272247314, "recall": 0.8877143859863281}, "safety": {"f1_macro": 0.9757658839225769, "f1_micro": 0.9757658839225769, "precision": 0.9664855003356934, "recall": 0.9852262139320374}, "safety_release": {"f1_macro": 0.949852466583252, "f1_micro": 0.9498525261878967, "precision": 0.9564356207847595, "recall": 0.943359375}}, "val_minor": {"tc260_minor": 0.4541294574737549, "lm_minor": 0.4003761410713196, "pii_minor": 0.82770174741745, "dc_minor": 0.21765097975730896}, "lm_per_class": {"LM.01": 0.332, "LM.02": 0.3744, "LM.03": 0.3266, "LM.04": 0.5596, "LM.05": 0.3827, "LM.06": 0.4677, "LM.07": 0.3844, "LM.08": 0.1455, "LM.09": 0.5427, "LM.10": 0.6769}}
+{"epoch": 6, "train_loss": {"tc260_major": 0.14358079145135974, "lm_major": 0.17481182220960212, "pii_major": 0.06552847987263265, "dc_type": 0.254685022067764, "dc_level": 0.31594446627184664, "safety": 0.23454712803247627, "safety_release": 0.37294206442362976, "tc260_minor": 0.12846563334431363, "lm_minor": 0.07145005454240183, "pii_minor": 0.0492711555701861, "dc_minor": 0.27096811624177847, "total": 1.722423369768318}, "val_main_f1": 0.8445921838283539, "val": {"tc260_major": {"f1_macro": 0.8235136866569519, "f1_micro": 0.8603066205978394, "precision": 0.7453969120979309, "recall": 0.9342136383056641}, "lm_major": {"f1_macro": 0.6922788023948669, "f1_micro": 0.7604473233222961, "precision": 0.5793354511260986, "recall": 0.8858792185783386}, "pii_major": {"f1_macro": 0.8882878422737122, "f1_micro": 0.8619047403335571, "precision": 0.8231906294822693, "recall": 0.9681723713874817}, "dc_type": {"f1_macro": 0.5524951815605164, "f1_micro": 0.8087885975837708, "precision": 0.4850459098815918, "recall": 0.8107566833496094}, "dc_level": {"f1_macro": 0.7846479415893555, "f1_micro": 0.9470462203025818, "precision": 0.7085123658180237, "recall": 0.900465190410614}, "safety": {"f1_macro": 0.9742884039878845, "f1_micro": 0.9742883443832397, "precision": 0.9689497947692871, "recall": 0.9796860814094543}, "safety_release": {"f1_macro": 0.9502439498901367, "f1_micro": 0.9502438902854919, "precision": 0.9493177533149719, "recall": 0.951171875}}, "val_minor": {"tc260_minor": 0.5110155940055847, "lm_minor": 0.4036601185798645, "pii_minor": 0.8860257863998413, "dc_minor": 0.2063228189945221}, "lm_per_class": {"LM.01": 0.3417, "LM.02": 0.342, "LM.03": 0.3394, "LM.04": 0.5279, "LM.05": 0.3106, "LM.06": 0.3756, "LM.07": 0.5429, "LM.08": 0.1614, "LM.09": 0.5119, "LM.10": 0.7945}}
+{"epoch": 7, "train_loss": {"tc260_major": 0.10824590174036239, "lm_major": 0.15456650499151328, "pii_major": 0.051352349189652854, "dc_type": 0.19269665533714517, "dc_level": 0.23438729974829603, "safety": 0.2279440736519802, "safety_release": 0.3638644825382486, "tc260_minor": 0.1098389130679773, "lm_minor": 0.06305900088213343, "pii_minor": 0.04265755201392866, "dc_minor": 0.23987805807719612, "total": 1.4730726723391088}, "val_main_f1": 0.8496824502944946, "val": {"tc260_major": {"f1_macro": 0.8473679423332214, "f1_micro": 0.873846173286438, "precision": 0.7882770299911499, "recall": 0.9241374731063843}, "lm_major": {"f1_macro": 0.6957513689994812, "f1_micro": 0.7662178874015808, "precision": 0.6006420850753784, "recall": 0.8856738805770874}, "pii_major": {"f1_macro": 0.882142961025238, "f1_micro": 0.8418604731559753, "precision": 0.8208238482475281, "recall": 0.9597343802452087}, "dc_type": {"f1_macro": 0.5312463045120239, "f1_micro": 0.7367424368858337, "precision": 0.49740204215049744, "recall": 0.7344080805778503}, "dc_level": {"f1_macro": 0.8354564309120178, "f1_micro": 0.955315887928009, "precision": 0.7831074595451355, "recall": 0.9112178683280945}, "safety": {"f1_macro": 0.9734675288200378, "f1_micro": 0.9734675288200378, "precision": 0.9646418690681458, "recall": 0.9824561476707458}, "safety_release": {"f1_macro": 0.9447731971740723, "f1_micro": 0.9447731971740723, "precision": 0.9541832804679871, "recall": 0.935546875}}, "val_minor": {"tc260_minor": 0.5561540722846985, "lm_minor": 0.4140321910381317, "pii_minor": 0.9161912202835083, "dc_minor": 0.26189032196998596}, "lm_per_class": {"LM.01": 0.3528, "LM.02": 0.2965, "LM.03": 0.3283, "LM.04": 0.4247, "LM.05": 0.32, "LM.06": 0.4877, "LM.07": 0.4936, "LM.08": 0.318, "LM.09": 0.5529, "LM.10": 0.7476}}
+{"epoch": 8, "train_loss": {"tc260_major": 0.0793051579312869, "lm_major": 0.11951339565468116, "pii_major": 0.05061422329551552, "dc_type": 0.1378201885850501, "dc_level": 0.15602789667618638, "safety": 0.22391827442453285, "safety_release": 0.35651913140039243, "tc260_minor": 0.09725173490775846, "lm_minor": 0.058131389366823806, "pii_minor": 0.04413325373223361, "dc_minor": 0.2360663955125149, "total": 1.2571211966169238}, "val_main_f1": 0.8742935359477997, "val": {"tc260_major": {"f1_macro": 0.8751633763313293, "f1_micro": 0.8921833038330078, "precision": 0.8453771471977234, "recall": 0.9125364422798157}, "lm_major": {"f1_macro": 0.737494170665741, "f1_micro": 0.8080554008483887, "precision": 0.6484972238540649, "recall": 0.8791524767875671}, "pii_major": {"f1_macro": 0.907751202583313, "f1_micro": 0.8883374929428101, "precision": 0.8734541535377502, "recall": 0.9461942911148071}, "dc_type": {"f1_macro": 0.6138864755630493, "f1_micro": 0.8704339861869812, "precision": 0.5614004135131836, "recall": 0.7704451084136963}, "dc_level": {"f1_macro": 0.8183910846710205, "f1_micro": 0.9546865224838257, "precision": 0.8019874691963196, "recall": 0.8433440327644348}, "safety": {"f1_macro": 0.9767653942108154, "f1_micro": 0.9767653942108154, "precision": 0.9640287756919861, "recall": 0.9898430109024048}, "safety_release": {"f1_macro": 0.9536031484603882, "f1_micro": 0.9536031484603882, "precision": 0.9640718698501587, "recall": 0.943359375}}, "val_minor": {"tc260_minor": 0.5821603536605835, "lm_minor": 0.4367268681526184, "pii_minor": 0.925092339515686, "dc_minor": 0.3631850481033325}, "lm_per_class": {"LM.01": 0.3408, "LM.02": 0.3269, "LM.03": 0.3851, "LM.04": 0.5135, "LM.05": 0.3992, "LM.06": 0.5887, "LM.07": 0.5716, "LM.08": 0.1852, "LM.09": 0.5193, "LM.10": 0.8262}}
+{"epoch": 9, "train_loss": {"tc260_major": 0.05988001951279097, "lm_major": 0.1021887294046688, "pii_major": 0.030197345207337096, "dc_type": 0.1142415130254624, "dc_level": 0.1225842024916149, "safety": 0.22151802013218733, "safety_release": 0.3540502272769066, "tc260_minor": 0.08327409619484034, "lm_minor": 0.04828757097932241, "pii_minor": 0.042074456477111284, "dc_minor": 0.2157957615379934, "total": 1.1237851636090284}, "val_main_f1": 0.8755452930927277, "val": {"tc260_major": {"f1_macro": 0.8678854703903198, "f1_micro": 0.8904047012329102, "precision": 0.8447233438491821, "recall": 0.8933612704277039}, "lm_major": {"f1_macro": 0.7376798391342163, "f1_micro": 0.796983003616333, "precision": 0.6611100435256958, "recall": 0.8656541705131531}, "pii_major": {"f1_macro": 0.9199782013893127, "f1_micro": 0.897243082523346, "precision": 0.888489842414856, "recall": 0.9546323418617249}, "dc_type": {"f1_macro": 0.5566235780715942, "f1_micro": 0.7880154848098755, "precision": 0.5159589052200317, "recall": 0.7156491875648499}, "dc_level": {"f1_macro": 0.7996445298194885, "f1_micro": 0.9512573480606079, "precision": 0.7754796147346497, "recall": 0.8348159790039062}, "safety": {"f1_macro": 0.9766376614570618, "f1_micro": 0.9766376614570618, "precision": 0.9690909385681152, "recall": 0.9843028783798218}, "safety_release": {"f1_macro": 0.9506903886795044, "f1_micro": 0.9506903290748596, "precision": 0.9601593613624573, "recall": 0.94140625}}, "val_minor": {"tc260_minor": 0.5911443829536438, "lm_minor": 0.4554157555103302, "pii_minor": 0.9541136026382446, "dc_minor": 0.3708079159259796}, "lm_per_class": {"LM.01": 0.3493, "LM.02": 0.3594, "LM.03": 0.3828, "LM.04": 0.4693, "LM.05": 0.3986, "LM.06": 0.6238, "LM.07": 0.7321, "LM.08": 0.1844, "LM.09": 0.5596, "LM.10": 0.8344}}
+{"epoch": 10, "train_loss": {"tc260_major": 0.044784919876760754, "lm_major": 0.08561530314421473, "pii_major": 0.026073518357966, "dc_type": 0.0701543494167342, "dc_level": 0.10443504498052568, "safety": 0.21972136174721577, "safety_release": 0.3498574228645823, "tc260_minor": 0.07038806260801679, "lm_minor": 0.0463389319624001, "pii_minor": 0.0394755645088425, "dc_minor": 0.19144282637727444, "total": 1.0068580176215103}, "val_main_f1": 0.8857782185077667, "val": {"tc260_major": {"f1_macro": 0.8684318661689758, "f1_micro": 0.89402174949646, "precision": 0.8345612287521362, "recall": 0.9084736704826355}, "lm_major": {"f1_macro": 0.77096027135849, "f1_micro": 0.826259970664978, "precision": 0.7164555788040161, "recall": 0.8463187217712402}, "pii_major": {"f1_macro": 0.9261668920516968, "f1_micro": 0.9025641083717346, "precision": 0.9122849106788635, "recall": 0.9469792246818542}, "dc_type": {"f1_macro": 0.5975560545921326, "f1_micro": 0.8614045977592468, "precision": 0.5412276387214661, "recall": 0.7691519856452942}, "dc_level": {"f1_macro": 0.8020997047424316, "f1_micro": 0.9545737504959106, "precision": 0.7705186009407043, "recall": 0.839533269405365}, "safety": {"f1_macro": 0.9775538444519043, "f1_micro": 0.9775538444519043, "precision": 0.9700000286102295, "recall": 0.9852262139320374}, "safety_release": {"f1_macro": 0.9567779898643494, "f1_micro": 0.9567779898643494, "precision": 0.9624505639076233, "recall": 0.951171875}}, "val_minor": {"tc260_minor": 0.6238080859184265, "lm_minor": 0.4705217480659485, "pii_minor": 0.9476789236068726, "dc_minor": 0.34575819969177246}, "lm_per_class": {"LM.01": 0.3525, "LM.02": 0.3721, "LM.03": 0.325, "LM.04": 0.5503, "LM.05": 0.5251, "LM.06": 0.6078, "LM.07": 0.6665, "LM.08": 0.2816, "LM.09": 0.539, "LM.10": 0.7782}}
+{"epoch": 11, "train_loss": {"tc260_major": 0.0325028066224793, "lm_major": 0.053122896986205446, "pii_major": 0.023533023145489235, "dc_type": 0.09195473396781383, "dc_level": 0.06319802770830758, "safety": 0.21872773537142068, "safety_release": 0.3478381942012909, "tc260_minor": 0.05635727165036877, "lm_minor": 0.03824406525505354, "pii_minor": 0.03055702128900684, "dc_minor": 0.12970525472730016, "total": 0.909024854039037}, "val_main_f1": 0.8872918486595154, "val": {"tc260_major": {"f1_macro": 0.8747491836547852, "f1_micro": 0.8931506872177124, "precision": 0.8485760688781738, "recall": 0.9047259092330933}, "lm_major": {"f1_macro": 0.7813909649848938, "f1_micro": 0.8415300250053406, "precision": 0.7483435869216919, "recall": 0.8261621594429016}, "pii_major": {"f1_macro": 0.915984034538269, "f1_micro": 0.8872180581092834, "precision": 0.8874055743217468, "recall": 0.9495302438735962}, "dc_type": {"f1_macro": 0.5966603755950928, "f1_micro": 0.8531820774078369, "precision": 0.5389696955680847, "recall": 0.737755537033081}, "dc_level": {"f1_macro": 0.8083129525184631, "f1_micro": 0.9507481455802917, "precision": 0.779735267162323, "recall": 0.8413046002388}, "safety": {"f1_macro": 0.9770432114601135, "f1_micro": 0.9770431518554688, "precision": 0.9716895222663879, "recall": 0.9824561476707458}, "safety_release": {"f1_macro": 0.9538763761520386, "f1_micro": 0.9538763761520386, "precision": 0.9585798978805542, "recall": 0.94921875}}, "val_minor": {"tc260_minor": 0.6545469164848328, "lm_minor": 0.5216651558876038, "pii_minor": 0.9513248205184937, "dc_minor": 0.3954514265060425}, "lm_per_class": {"LM.01": 0.4146, "LM.02": 0.3902, "LM.03": 0.3502, "LM.04": 0.6041, "LM.05": 0.5719, "LM.06": 0.6689, "LM.07": 0.7071, "LM.08": 0.359, "LM.09": 0.5776, "LM.10": 0.8731}}
+{"epoch": 12, "train_loss": {"tc260_major": 0.021178718829877, "lm_major": 0.04006829768954802, "pii_major": 0.018531427401852377, "dc_type": 0.05443633335387633, "dc_level": 0.04598163484915616, "safety": 0.21733041383235294, "safety_release": 0.3450643929466193, "tc260_minor": 0.046726028212673076, "lm_minor": 0.02953748750396729, "pii_minor": 0.029286863206291945, "dc_minor": 0.11279065577871175, "total": 0.8096143547218635}, "val_main_f1": 0.8898130059242249, "val": {"tc260_major": {"f1_macro": 0.8723844289779663, "f1_micro": 0.8907103538513184, "precision": 0.8497238159179688, "recall": 0.8975313901901245}, "lm_major": {"f1_macro": 0.7924647927284241, "f1_micro": 0.8468468189239502, "precision": 0.7653323411941528, "recall": 0.8294259905815125}, "pii_major": {"f1_macro": 0.9181905388832092, "f1_micro": 0.8951406478881836, "precision": 0.8973146677017212, "recall": 0.9422042965888977}, "dc_type": {"f1_macro": 0.5619332194328308, "f1_micro": 0.7905927896499634, "precision": 0.5241569876670837, "recall": 0.7118810415267944}, "dc_level": {"f1_macro": 0.8199318051338196, "f1_micro": 0.9511055946350098, "precision": 0.7850687503814697, "recall": 0.8652612566947937}, "safety": {"f1_macro": 0.9762122631072998, "f1_micro": 0.9762122631072998, "precision": 0.9673617482185364, "recall": 0.9852262139320374}, "safety_release": {"f1_macro": 0.951724112033844, "f1_micro": 0.951724112033844, "precision": 0.9602385759353638, "recall": 0.943359375}}, "val_minor": {"tc260_minor": 0.6857715249061584, "lm_minor": 0.545832097530365, "pii_minor": 0.9529081583023071, "dc_minor": 0.38066062331199646}, "lm_per_class": {"LM.01": 0.4013, "LM.02": 0.3856, "LM.03": 0.4367, "LM.04": 0.6028, "LM.05": 0.5569, "LM.06": 0.7917, "LM.07": 0.7506, "LM.08": 0.4137, "LM.09": 0.5666, "LM.10": 0.8747}}
+{"epoch": 13, "train_loss": {"tc260_major": 0.0174057651875348, "lm_major": 0.02987440689604194, "pii_major": 0.017643652971580808, "dc_type": 0.033784229905557316, "dc_level": 0.03343534246385888, "safety": 0.21624243589466724, "safety_release": 0.34356313422404783, "tc260_minor": 0.03312264056360619, "lm_minor": 0.02339915938025386, "pii_minor": 0.023826587109648016, "dc_minor": 0.07955998133634053, "total": 0.7413460401073511}, "val_main_f1": 0.8969135582447052, "val": {"tc260_major": {"f1_macro": 0.8791688680648804, "f1_micro": 0.8998178243637085, "precision": 0.8601519465446472, "recall": 0.8999252319335938}, "lm_major": {"f1_macro": 0.7999340295791626, "f1_micro": 0.8519040942192078, "precision": 0.7899037599563599, "recall": 0.8171448707580566}, "pii_major": {"f1_macro": 0.9282130002975464, "f1_micro": 0.9067357778549194, "precision": 0.9257856011390686, "recall": 0.9359902143478394}, "dc_type": {"f1_macro": 0.6350694298744202, "f1_micro": 0.8640714883804321, "precision": 0.5817037224769592, "recall": 0.7497000098228455}, "dc_level": {"f1_macro": 0.8017821311950684, "f1_micro": 0.9513412117958069, "precision": 0.7838972210884094, "recall": 0.8236100673675537}, "safety": {"f1_macro": 0.9803383350372314, "f1_micro": 0.9803383350372314, "precision": 0.9710144996643066, "recall": 0.9898430109024048}, "safety_release": {"f1_macro": 0.9593656659126282, "f1_micro": 0.959365725517273, "precision": 0.9738430380821228, "recall": 0.9453125}}, "val_minor": {"tc260_minor": 0.7156657576560974, "lm_minor": 0.5321382880210876, "pii_minor": 0.9566130638122559, "dc_minor": 0.39822348952293396}, "lm_per_class": {"LM.01": 0.4606, "LM.02": 0.3894, "LM.03": 0.4123, "LM.04": 0.5677, "LM.05": 0.5462, "LM.06": 0.5481, "LM.07": 0.793, "LM.08": 0.3721, "LM.09": 0.6166, "LM.10": 0.8839}}
+{"epoch": 14, "train_loss": {"tc260_major": 0.010862273763646905, "lm_major": 0.018509679059698483, "pii_major": 0.011834806243673801, "dc_type": 0.03222230490582247, "dc_level": 0.024811880807876597, "safety": 0.21476190184900004, "safety_release": 0.3424868853319259, "tc260_minor": 0.023327847915894432, "lm_minor": 0.018593371042928627, "pii_minor": 0.02229320418487284, "dc_minor": 0.0638215839902386, "total": 0.6952398291985986}, "val_main_f1": 0.8978046327829361, "val": {"tc260_major": {"f1_macro": 0.8807706832885742, "f1_micro": 0.8981651663780212, "precision": 0.864391028881073, "recall": 0.898901104927063}, "lm_major": {"f1_macro": 0.8007346391677856, "f1_micro": 0.8625792860984802, "precision": 0.7969151139259338, "recall": 0.8103917241096497}, "pii_major": {"f1_macro": 0.9326069951057434, "f1_micro": 0.90625, "precision": 0.9288902282714844, "recall": 0.9396532773971558}, "dc_type": {"f1_macro": 0.6665398478507996, "f1_micro": 0.8805351853370667, "precision": 0.6091427803039551, "recall": 0.7827596068382263}, "dc_level": {"f1_macro": 0.8161537051200867, "f1_micro": 0.9556803703308105, "precision": 0.7930783629417419, "recall": 0.8422110080718994}, "safety": {"f1_macro": 0.9771062135696411, "f1_micro": 0.9771062135696411, "precision": 0.9691190123558044, "recall": 0.9852262139320374}, "safety_release": {"f1_macro": 0.9536031484603882, "f1_micro": 0.9536031484603882, "precision": 0.9640718698501587, "recall": 0.943359375}}, "val_minor": {"tc260_minor": 0.7357353568077087, "lm_minor": 0.5736752152442932, "pii_minor": 0.9624616503715515, "dc_minor": 0.39403411746025085}, "lm_per_class": {"LM.01": 0.4441, "LM.02": 0.3966, "LM.03": 0.4494, "LM.04": 0.673, "LM.05": 0.5522, "LM.06": 0.7979, "LM.07": 0.811, "LM.08": 0.3907, "LM.09": 0.6134, "LM.10": 0.9673}}
+{"epoch": 15, "train_loss": {"tc260_major": 0.008751979781680092, "lm_major": 0.014592604781381744, "pii_major": 0.008992818792041567, "dc_type": 0.01405082373006723, "dc_level": 0.017688985283067497, "safety": 0.21403912152034762, "safety_release": 0.34158278877949, "tc260_minor": 0.01605746447211049, "lm_minor": 0.015316441133030477, "pii_minor": 0.012752156400822497, "dc_minor": 0.043608605615393196, "total": 0.6473108479878963}, "val_main_f1": 0.897221565246582, "val": {"tc260_major": {"f1_macro": 0.8797600865364075, "f1_micro": 0.8978840708732605, "precision": 0.8669196963310242, "recall": 0.8942126035690308}, "lm_major": {"f1_macro": 0.7961527705192566, "f1_micro": 0.8599717020988464, "precision": 0.8022886514663696, "recall": 0.796729326248169}, "pii_major": {"f1_macro": 0.9340936541557312, "f1_micro": 0.9162303805351257, "precision": 0.9373016357421875, "recall": 0.9359902143478394}, "dc_type": {"f1_macro": 0.6664154529571533, "f1_micro": 0.8745611310005188, "precision": 0.607439398765564, "recall": 0.7795130610466003}, "dc_level": {"f1_macro": 0.818394660949707, "f1_micro": 0.9550561904907227, "precision": 0.7964537739753723, "recall": 0.8424376845359802}, "safety": {"f1_macro": 0.9788797497749329, "f1_micro": 0.9788796901702881, "precision": 0.9735159873962402, "recall": 0.9843028783798218}, "safety_release": {"f1_macro": 0.9528486728668213, "f1_micro": 0.9528487324714661, "precision": 0.9584980010986328, "recall": 0.947265625}}, "val_minor": {"tc260_minor": 0.7555519342422485, "lm_minor": 0.5591629147529602, "pii_minor": 0.9688752293586731, "dc_minor": 0.4217333495616913}, "lm_per_class": {"LM.01": 0.4427, "LM.02": 0.4256, "LM.03": 0.4412, "LM.04": 0.6188, "LM.05": 0.5835, "LM.06": 0.5757, "LM.07": 0.7664, "LM.08": 0.3909, "LM.09": 0.639, "LM.10": 0.9733}}
+{"epoch": 16, "train_loss": {"tc260_major": 0.008236990164289396, "lm_major": 0.011217260491883227, "pii_major": 0.007268745849583565, "dc_type": 0.01806829404093932, "dc_level": 0.019001540917445832, "safety": 0.21359769960376512, "safety_release": 0.3406887035961299, "tc260_minor": 0.01254274652156193, "lm_minor": 0.013176720784327318, "pii_minor": 0.009849977325478217, "dc_minor": 0.040300115213272356, "total": 0.6420934896004424}, "val_main_f1": 0.8969433009624481, "val": {"tc260_major": {"f1_macro": 0.8871189951896667, "f1_micro": 0.9029884934425354, "precision": 0.8740934133529663, "recall": 0.9015825390815735}, "lm_major": {"f1_macro": 0.7944079637527466, "f1_micro": 0.8575478196144104, "precision": 0.8015189170837402, "recall": 0.7938677072525024}, "pii_major": {"f1_macro": 0.9278157353401184, "f1_micro": 0.9057591557502747, "precision": 0.931380569934845, "recall": 0.9286642074584961}, "dc_type": {"f1_macro": 0.6712372303009033, "f1_micro": 0.8812358975410461, "precision": 0.6259512305259705, "recall": 0.7568556070327759}, "dc_level": {"f1_macro": 0.8076675534248352, "f1_micro": 0.9556527137756348, "precision": 0.7920801043510437, "recall": 0.8247430920600891}, "safety": {"f1_macro": 0.9784305095672607, "f1_micro": 0.978430449962616, "precision": 0.9726277589797974, "recall": 0.9843028783798218}, "safety_release": {"f1_macro": 0.951724112033844, "f1_micro": 0.951724112033844, "precision": 0.9602385759353638, "recall": 0.943359375}}, "val_minor": {"tc260_minor": 0.7655852437019348, "lm_minor": 0.5750293135643005, "pii_minor": 0.9740683436393738, "dc_minor": 0.41658729314804077}, "lm_per_class": {"LM.01": 0.4444, "LM.02": 0.4184, "LM.03": 0.445, "LM.04": 0.7413, "LM.05": 0.5746, "LM.06": 0.5676, "LM.07": 0.7664, "LM.08": 0.4407, "LM.09": 0.6444, "LM.10": 0.9641}}
+{"epoch": 17, "train_loss": {"tc260_major": 0.008148024076103136, "lm_major": 0.00993130074994193, "pii_major": 0.005383056642299325, "dc_type": 0.012868016833451604, "dc_level": 0.00824352141943466, "safety": 0.21327188191883853, "safety_release": 0.3402682604517784, "tc260_minor": 0.010424214715379417, "lm_minor": 0.010516449056016817, "pii_minor": 0.00828208239235132, "dc_minor": 0.028020683777138608, "total": 0.6165211284543457}, "val_main_f1": 0.9007487297058105, "val": {"tc260_major": {"f1_macro": 0.8882125616073608, "f1_micro": 0.9039963483810425, "precision": 0.8753134608268738, "recall": 0.9023561477661133}, "lm_major": {"f1_macro": 0.8041149973869324, "f1_micro": 0.862241268157959, "precision": 0.8118866682052612, "recall": 0.8026593923568726}, "pii_major": {"f1_macro": 0.9304012060165405, "f1_micro": 0.9105263352394104, "precision": 0.9366006851196289, "recall": 0.9286642074584961}, "dc_type": {"f1_macro": 0.6574600338935852, "f1_micro": 0.8508800864219666, "precision": 0.6262161731719971, "recall": 0.7274240851402283}, "dc_level": {"f1_macro": 0.7985851168632507, "f1_micro": 0.9538077116012573, "precision": 0.7810103297233582, "recall": 0.817574679851532}, "safety": {"f1_macro": 0.9802661538124084, "f1_micro": 0.9802661538124084, "precision": 0.974452555179596, "recall": 0.9861496090888977}, "safety_release": {"f1_macro": 0.9547244310379028, "f1_micro": 0.9547244310379028, "precision": 0.9623016119003296, "recall": 0.947265625}}, "val_minor": {"tc260_minor": 0.7645347118377686, "lm_minor": 0.5834082961082458, "pii_minor": 0.9715160131454468, "dc_minor": 0.422242134809494}, "lm_per_class": {"LM.01": 0.4376, "LM.02": 0.4581, "LM.03": 0.485, "LM.04": 0.7468, "LM.05": 0.5904, "LM.06": 0.5676, "LM.07": 0.7753, "LM.08": 0.4352, "LM.09": 0.6158, "LM.10": 0.9641}}
+{"epoch": 18, "train_loss": {"tc260_major": 0.0075516464517732616, "lm_major": 0.007604069020438298, "pii_major": 0.004432380887687201, "dc_type": 0.01127251211452857, "dc_level": 0.006118494057969608, "safety": 0.21298545691394066, "safety_release": 0.33970709707781854, "tc260_minor": 0.008770376029447574, "lm_minor": 0.009025817497900006, "pii_minor": 0.0080493796480024, "dc_minor": 0.02933521573961009, "total": 0.6074429137894217}, "val_main_f1": 0.9017380625009537, "val": {"tc260_major": {"f1_macro": 0.89131098985672, "f1_micro": 0.9048275947570801, "precision": 0.8798427581787109, "recall": 0.9038307070732117}, "lm_major": {"f1_macro": 0.8049739003181458, "f1_micro": 0.8634738922119141, "precision": 0.8174961805343628, "recall": 0.7992476224899292}, "pii_major": {"f1_macro": 0.9304012060165405, "f1_micro": 0.9105263352394104, "precision": 0.9366006851196289, "recall": 0.9286642074584961}, "dc_type": {"f1_macro": 0.6703464984893799, "f1_micro": 0.8693088889122009, "precision": 0.6406766772270203, "recall": 0.736182689666748}, "dc_level": {"f1_macro": 0.8012346625328064, "f1_micro": 0.9534520506858826, "precision": 0.7893945574760437, "recall": 0.8137637972831726}, "safety": {"f1_macro": 0.9802661538124084, "f1_micro": 0.9802661538124084, "precision": 0.974452555179596, "recall": 0.9861496090888977}, "safety_release": {"f1_macro": 0.9537856578826904, "f1_micro": 0.9537856578826904, "precision": 0.9603960514068604, "recall": 0.947265625}}, "val_minor": {"tc260_minor": 0.7654358744621277, "lm_minor": 0.583200991153717, "pii_minor": 0.9734864234924316, "dc_minor": 0.4336632490158081}, "lm_per_class": {"LM.01": 0.434, "LM.02": 0.4319, "LM.03": 0.485, "LM.04": 0.744, "LM.05": 0.6222, "LM.06": 0.5676, "LM.07": 0.7664, "LM.08": 0.4381, "LM.09": 0.6306, "LM.10": 0.9641}}
+{"epoch": 19, "train_loss": {"tc260_major": 0.007581720789083551, "lm_major": 0.0057737104016450155, "pii_major": 0.0038502200597701924, "dc_type": 0.006379613999532189, "dc_level": 0.002309250194221052, "safety": 0.21277475470918358, "safety_release": 0.33949460127160924, "tc260_minor": 0.008156277672495953, "lm_minor": 0.008799185767180131, "pii_minor": 0.007464040427648626, "dc_minor": 0.022483393143110104, "total": 0.5934337519464039}, "val_main_f1": 0.9009537100791931, "val": {"tc260_major": {"f1_macro": 0.8893572092056274, "f1_micro": 0.9039080739021301, "precision": 0.8777278065681458, "recall": 0.901969313621521}, "lm_major": {"f1_macro": 0.8026129007339478, "f1_micro": 0.862464189529419, "precision": 0.8151254653930664, "recall": 0.7979046106338501}, "pii_major": {"f1_macro": 0.9315785765647888, "f1_micro": 0.9129287600517273, "precision": 0.9393339157104492, "recall": 0.9286642074584961}, "dc_type": {"f1_macro": 0.6651550531387329, "f1_micro": 0.8754076957702637, "precision": 0.6398302316665649, "recall": 0.727324366569519}, "dc_level": {"f1_macro": 0.8021449446678162, "f1_micro": 0.9524999856948853, "precision": 0.7886598706245422, "recall": 0.8166682720184326}, "safety": {"f1_macro": 0.9802661538124084, "f1_micro": 0.9802661538124084, "precision": 0.974452555179596, "recall": 0.9861496090888977}, "safety_release": {"f1_macro": 0.9556650519371033, "f1_micro": 0.9556650519371033, "precision": 0.9642146825790405, "recall": 0.947265625}}, "val_minor": {"tc260_minor": 0.761861264705658, "lm_minor": 0.594906210899353, "pii_minor": 0.9714365005493164, "dc_minor": 0.4312039911746979}, "lm_per_class": {"LM.01": 0.4356, "LM.02": 0.5169, "LM.03": 0.485, "LM.04": 0.744, "LM.05": 0.5954, "LM.06": 0.5676, "LM.07": 0.7664, "LM.08": 0.4436, "LM.09": 0.6485, "LM.10": 0.9641}}
+{"epoch": 20, "train_loss": {"tc260_major": 0.007359311646852574, "lm_major": 0.0057505568205516895, "pii_major": 0.004003247602583521, "dc_type": 0.003315164563291556, "dc_level": 0.0012717723567650745, "safety": 0.2127523838582625, "safety_release": 0.3393764242051316, "tc260_minor": 0.007370612950962561, "lm_minor": 0.008517985677981907, "pii_minor": 0.007248175871491578, "dc_minor": 0.021644217843140773, "total": 0.5884552817144003}, "val_main_f1": 0.9007079899311066, "val": {"tc260_major": {"f1_macro": 0.8883743286132812, "f1_micro": 0.9039080739021301, "precision": 0.877440333366394, "recall": 0.9001822471618652}, "lm_major": {"f1_macro": 0.8026129007339478, "f1_micro": 0.862464189529419, "precision": 0.8151254653930664, "recall": 0.7979046106338501}, "pii_major": {"f1_macro": 0.9315785765647888, "f1_micro": 0.9129287600517273, "precision": 0.9393339157104492, "recall": 0.9286642074584961}, "dc_type": {"f1_macro": 0.666433572769165, "f1_micro": 0.8756933212280273, "precision": 0.6413729190826416, "recall": 0.727324366569519}, "dc_level": {"f1_macro": 0.8022608160972595, "f1_micro": 0.9528272151947021, "precision": 0.7886651158332825, "recall": 0.8168948292732239}, "safety": {"f1_macro": 0.9802661538124084, "f1_micro": 0.9802661538124084, "precision": 0.974452555179596, "recall": 0.9861496090888977}, "safety_release": {"f1_macro": 0.9556650519371033, "f1_micro": 0.9556650519371033, "precision": 0.9642146825790405, "recall": 0.947265625}}, "val_minor": {"tc260_minor": 0.759658694267273, "lm_minor": 0.5951928496360779, "pii_minor": 0.9714365005493164, "dc_minor": 0.4330531060695648}, "lm_per_class": {"LM.01": 0.4356, "LM.02": 0.5169, "LM.03": 0.485, "LM.04": 0.744, "LM.05": 0.5954, "LM.06": 0.5676, "LM.07": 0.7664, "LM.08": 0.4436, "LM.09": 0.6485, "LM.10": 0.9673}}