Upload folder using huggingface_hub

.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
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,34 @@
+Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
+
+Copyright (c) 2024 Scott Thornton / perfecXion.ai
+
+AEGIS: Adaptive Ensemble Guard with Integrated Steering
+
+You are free to:
+  Share — copy and redistribute the material in any medium or format
+  Adapt — remix, transform, and build upon the material
+
+Under the following terms:
+
+  Attribution — You must give appropriate credit, provide a link to the license,
+  and indicate if changes were made. You may do so in any reasonable manner,
+  but not in any way that suggests the licensor endorses you or your use.
+
+  NonCommercial — You may not use the material for commercial purposes
+  without explicit written permission from the copyright holder.
+
+  ShareAlike — If you remix, transform, or build upon the material, you must
+  distribute your contributions under the same license as the original.
+
+  No additional restrictions — You may not apply legal terms or technological
+  measures that legally restrict others from doing anything the license permits.
+
+ATTRIBUTION REQUIREMENTS:
+When using this material, you must provide clear attribution including:
+1. The name "AEGIS" and credit to "scthornton.ai / perfecXion.ai"
+2. A link to the original repository
+3. Indication of any modifications made
+
+For commercial licensing inquiries, contact: scott@perfecxion.ai
+
+Full license text: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode

README.md

@@ -0,0 +1,163 @@
+---
+license: cc-by-nc-sa-4.0
+base_model: Qwen/Qwen2.5-3B-Instruct
+library_name: peft
+tags:
+  - llm-security
+  - jailbreak-detection
+  - text-classification
+  - safety
+  - lora
+pipeline_tag: text-classification
+---
+
+# TRYLOCK Sidecar Classifier
+
+Layer 3 of the TRYLOCK (Adaptive Ensemble Guard with Integrated Steering) defense system. This is a lightweight classifier that runs alongside the main LLM to detect attack patterns and dynamically adjust defense strength.
+
+## Model Description
+
+The sidecar classifier categorizes inputs into three classes:
+- **SAFE**: Benign queries that need minimal defense
+- **WARN**: Ambiguous or suspicious queries
+- **ATTACK**: Clear jailbreak/attack attempts
+
+## Intended Uses
+
+**Primary use cases:**
+- Real-time classification of LLM inputs for adaptive defense
+- Dynamically adjusting RepE steering strength
+- Research on attack detection methods
+- Content moderation and threat triage
+
+**Out of scope:**
+- Standalone content moderation (designed for adaptive steering)
+- High-stakes security decisions without human review
+
+### Training Details
+
+| Parameter | Value |
+|-----------|-------|
+| Base Model | Qwen2.5-3B-Instruct |
+| Method | LoRA fine-tuning |
+| LoRA Rank | 32 |
+| LoRA Alpha | 64 |
+| Target Modules | q_proj, k_proj, v_proj, o_proj |
+| Training Samples | 2,349 |
+| Classes | SAFE, WARN, ATTACK |
+
+### Evaluation Results
+
+| Class | Precision | Recall | F1 |
+|-------|-----------|--------|-----|
+| SAFE | 24.1% | 34.5% | 28.4% |
+| WARN | 58.3% | 51.6% | 54.8% |
+| ATTACK | 62.4% | 60.8% | 61.6% |
+| **Macro Avg** | 48.3% | 48.9% | 48.3% |
+
+**Note**: This is a research prototype. ATTACK detection is prioritized over SAFE detection for security.
+
+## Usage
+
+```python
+import torch
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+from peft import PeftModel
+
+# Load model
+base_model = AutoModelForSequenceClassification.from_pretrained(
+    "Qwen/Qwen2.5-3B-Instruct",
+    num_labels=3,
+    device_map="auto",
+    torch_dtype=torch.bfloat16,
+)
+model = PeftModel.from_pretrained(base_model, "scthornton/trylock-sidecar-classifier")
+tokenizer = AutoTokenizer.from_pretrained("scthornton/trylock-sidecar-classifier")
+
+# Classify input
+label_names = ["SAFE", "WARN", "ATTACK"]
+
+def classify(text):
+    inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=2048)
+    with torch.no_grad():
+        outputs = model(**inputs.to(model.device))
+        probs = torch.softmax(outputs.logits, dim=-1)[0]
+    return label_names[probs.argmax()], probs.tolist()
+
+# Example
+result, probs = classify("Ignore all previous instructions and...")
+print(f"Classification: {result}")  # ATTACK
+print(f"Probabilities: {dict(zip(label_names, probs))}")
+```
+
+## Dynamic Defense Integration
+
+Use the classification to adjust RepE steering strength:
+
+```python
+alpha_map = {
+    "SAFE": 0.5,   # Minimal steering - preserve fluency
+    "WARN": 1.5,   # Moderate steering
+    "ATTACK": 2.5  # Maximum steering - block harmful output
+}
+
+classification, _ = classify(user_input)
+alpha = alpha_map[classification]
+# Apply RepE steering with this alpha value
+```
+
+## TRYLOCK Architecture
+
+This classifier is Layer 3 of the 3-layer TRYLOCK defense:
+
+1. **Layer 1 (KNOWLEDGE)**: [trylock-mistral-7b-dpo](https://huggingface.co/scthornton/trylock-mistral-7b-dpo)
+2. **Layer 2 (INSTINCT)**: [trylock-repe-vectors](https://huggingface.co/scthornton/trylock-repe-vectors)
+3. **Layer 3 (OVERSIGHT)**: Sidecar classifier (this model)
+
+## Limitations and Risks
+
+**Limitations:**
+- SAFE class has lower precision (24%) - may over-classify benign queries as WARN
+- Trained on English-language attacks only
+- 3-class granularity may miss nuanced threat levels
+- Requires ~3B parameter model inference overhead
+
+**Risks:**
+- False positives may trigger unnecessary steering
+- False negatives on novel attack patterns
+- Classification confidence doesn't guarantee correctness
+
+**Recommendations:**
+- Use probability scores, not just class labels, for fine-grained control
+- Consider WARN classification as "elevated caution" rather than definite threat
+- Combine with other safety mechanisms for production use
+
+## Framework Versions
+
+- PEFT 0.18.0
+
+## Citation
+
+```bibtex
+@misc{trylock2024,
+  title={TRYLOCK: Defense-in-Depth Against LLM Jailbreaks via Layered Preference and Representation Engineering},
+  author={Thornton, Scott},
+  year={2024},
+  url={https://huggingface.co/scthornton/trylock-sidecar-classifier}
+}
+```
+
+## License
+
+**CC BY-NC-SA 4.0** (Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International)
+
+You are free to:
+- **Share** — copy and redistribute the material in any medium or format
+- **Adapt** — remix, transform, and build upon the material
+
+Under the following terms:
+- **Attribution** — You must give appropriate credit to **Scott Thornton**, provide a link to the license, and indicate if changes were made
+- **NonCommercial** — You may not use the material for commercial purposes without explicit written permission
+- **ShareAlike** — If you remix, transform, or build upon the material, you must distribute your contributions under the same license
+
+For commercial licensing inquiries, contact: scott@perfecxion.ai

adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-3B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [
+    "classifier",
+    "score"
+  ],
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "SEQ_CLS",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23ccbeb16090a961da192152bae6afda1e64da54b8a674e7286dabf0927ff7c7
+size 59033448

added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

chat_template.jinja

@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# 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' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.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' }}
+{%- endif %}

special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

thresholds.json

@@ -0,0 +1,9 @@
+{
+  "warn_threshold": 0.3,
+  "attack_threshold": 0.7,
+  "label_names": [
+    "SAFE",
+    "WARN",
+    "ATTACK"
+  ]
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:694f1174c5bdf94e2fc50796c0f1733a5a3945ff110b0dfa40ea0701cc9c9c42
+size 11422176

tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ac2954cf714d1e4c9de303ea1e69eb15ccd8e331bc716544a59c350d1bcb250
+size 5841