初始化项目，由ModelHub XC社区提供模型

Model: heterodoxin/qwen3-8b-apostate
Source: Original Platform

.gitattributes

@@ -0,0 +1,36 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz 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

README.md

@@ -0,0 +1,62 @@
+---
+license: apache-2.0
+library_name: transformers
+base_model: Qwen/Qwen3-8B
+pipeline_tag: text-generation
+tags:
+- apostate
+- uncensored
+- abliteration
+- qwen3
+---
+
+# Qwen3-8B Apostate
+
+An uncensored edit of [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B). Refusal behavior is removed by editing the weights directly — no finetuning, no adapter, no runtime hook. The result is a standard Transformers checkpoint that loads anywhere Qwen3 does.
+
+Produced with **[Apostate](https://github.com/heterodoxin/apostate)**.
+
+## Method
+
+Apostate identifies the residual-stream direction that separates refused prompts from answered ones, then projects it out of the model's weights permanently. For Qwen3-8B (a standard pre-norm dense transformer), the edit targets the **writer side**: the refusal direction is removed from the weight matrices of every module that writes to the residual stream — attention output projections and MLP down-projections — across all layers.
+
+The edit uses **oblique (mean-preserving) ablation**: the operator `E = I − R Uᵀ` where `U` is `R` minus its harmless-mean component. This removes the refusal direction while preserving the model's average harmless-prompt behavior, keeping output quality high.
+
+The refusal subspace is found with a rank-3 predictive TPE search, with causal layer importance scoring to focus edits on the layers that most drive refusal (concentrated in the mid-to-late layers, peak at layer 29).
+
+## Results
+
+Evaluated on held-out prompts from JailbreakBench and the harmful_behaviors test split. Refusal is graded by a classifier with a weak-compliance guard; KL measures token-distribution shift on harmless prompts.
+
+| Metric | Base | Apostate |
+|---|---|---|
+| Refusal rate | 91.7% | 22.9% |
+| Comply rate | 8.3% | 77.1% |
+| Harmless KL (nats) | 0 | 0.120 |
+
+The model answers freely on requests the base model refuses while remaining coherent and on-task for everyday use.
+
+## Usage
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_id = "heterodoxin/qwen3-8b-apostate"
+tok = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")
+
+messages = [{"role": "user", "content": "Your prompt here"}]
+text = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+inputs = tok(text, return_tensors="pt").to(model.device)
+outputs = model.generate(**inputs, max_new_tokens=512)
+print(tok.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))
+```
+
+Qwen3 supports a thinking mode — pass `enable_thinking=True` to the chat template if you want extended reasoning.
+
+## Notes
+
+- This is an **uncensored** model. It will comply with requests the base model refuses.
+- The edit is baked into the weights; there is no system prompt or LoRA involved.
+- For the base model's capabilities and licensing, see [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B).
+- Join the community: [Discord](https://discord.gg/NPA7xrATEH)

apostate_config.json

@@ -0,0 +1,133 @@
+{
+  "model": "Qwen/Qwen3-8B",
+  "output_dir": "/var/home/Heterodoxin/ablate_work/qwen3-8b-apostate",
+  "profile": "balanced",
+  "device": "cuda",
+  "load_in_4bit": true,
+  "compute_dtype": "bfloat16",
+  "seed": 0,
+  "resume": false,
+  "cache_activations": true,
+  "activation_cache_dir": null,
+  "harmful_path": "mlabonne/harmful_behaviors:train:text|/var/home/Heterodoxin/apostate/data/harmful.txt|/var/home/Heterodoxin/apostate/data/refusal_calibration.txt",
+  "harmless_path": "mlabonne/harmless_alpaca:train:text|/var/home/Heterodoxin/apostate/data/harmless.txt",
+  "harmful_test": "mlabonne/harmful_behaviors:test:text|JailbreakBench/JBB-Behaviors@behaviors:harmful:Goal|/var/home/Heterodoxin/apostate/data/refusal_calibration.txt",
+  "harmless_test": "mlabonne/harmless_alpaca:test:text",
+  "refusal_eval_path": "JailbreakBench/JBB-Behaviors@behaviors:harmful:Goal|/var/home/Heterodoxin/apostate/data/refusal_calibration.txt",
+  "refusal_eval_n": 64,
+  "kl_eval_path": "mlabonne/harmless_alpaca:test:text",
+  "kl_eval_n": 48,
+  "preserve_path": null,
+  "n_harmful": 600,
+  "n_harmless": 600,
+  "n_eval": 300,
+  "max_new_tokens": 32,
+  "batch_size": 24,
+  "baseline_eval_n": 24,
+  "head_sweep": true,
+  "head_sweep_min": 3.5,
+  "head_sweep_max": 5.5,
+  "head_sweep_step": 0.5,
+  "head_sweep_top_k": 6,
+  "head_sweep_probe_n": 8,
+  "head_sweep_eval_n": 48,
+  "head_sweep_probe_classifier": false,
+  "fit_response_activations": false,
+  "fit_response_n": 160,
+  "fit_response_tokens": 32,
+  "refusal_rank": 1,
+  "variance_threshold": 0.9,
+  "max_rank": 3,
+  "direction_layer_frac": 0.6,
+  "direction_scope": "global",
+  "multi_refusal": true,
+  "multi_refusal_clusters": 6,
+  "multi_refusal_min_norm": 0.08,
+  "multi_refusal_min_separation": 0.05,
+  "multi_refusal_min_coverage": 0.05,
+  "orthogonalize_direction": true,
+  "causal_targeting": true,
+  "causal_floor": 0.1,
+  "causal_temperature": 1.0,
+  "preserve_rank": 8,
+  "refine_refusal": true,
+  "refine_max_scale": 2.0,
+  "refine_steps": 6,
+  "refine_deescalate": true,
+  "refine_kl_steps": 10,
+  "refine_scale_rerank_k": 2,
+  "refine_kl_layer_steps": 10,
+  "refine_kl_layer_candidates": 8,
+  "repair_steps": 4,
+  "repair_candidates": 8,
+  "repair_rerank_k": 5,
+  "repair_probe_candidates": 20,
+  "repair_probe_ref_n": 12,
+  "repair_probe_kl_n": 16,
+  "repair_probe_positions": 8,
+  "repair_refusal_regress_slack": 0.01,
+  "repair_stop_kl_frac": 0.8,
+  "repair_min_alpha": 0.001,
+  "repair_min_kl_gain": 0.003,
+  "repair_min_refusal_gain": 0.005,
+  "repair_min_score_gain": 0.01,
+  "repair_eval_n": 96,
+  "repair_kl_n": 64,
+  "refine_refusal_slack": 0.01,
+  "final_zero_trim": false,
+  "final_push_bake_margin": 0.075,
+  "guard_max_iters": 2,
+  "guard_leakage_eps": 0.15,
+  "guard_alpha_step": 0.25,
+  "optimize": true,
+  "n_trials": 16,
+  "adaptive_trials": true,
+  "kl_weight": 6.0,
+  "kl_target": 0.04,
+  "kl_target_weight": 18.0,
+  "kl_quad_weight": 22.0,
+  "kl_headroom_weight": 0.0,
+  "kl_over_budget_weight": 72.0,
+  "refusal_target_weight": 4.0,
+  "refusal_quad_weight": 8.0,
+  "kl_positions": 8,
+  "opt_capability": true,
+  "opt_capability_weight": 2.5,
+  "opt_capability_code_n": 8,
+  "opt_capability_math_n": 8,
+  "opt_eval_n": 32,
+  "opt_gen_tokens": 32,
+  "opt_objective": "generation",
+  "opt_rerank_k": 5,
+  "opt_guard": true,
+  "opt_early_stop": true,
+  "opt_early_stop_margin": 0.02,
+  "gemma_ple": false,
+  "gemma_query": false,
+  "ple_max_rank": 2,
+  "prune": false,
+  "prune_max_frac": 0.25,
+  "prune_kl": 0.04,
+  "max_kl": 0.12,
+  "target_refusal": 0.05,
+  "oblique_ablation": true,
+  "oblique_strength": 1.0,
+  "oblique_denom_floor": 0.2,
+  "oblique_writers_only": true,
+  "oblique_predictive": true,
+  "predictive_ridge": 0.01,
+  "reader_max_kl": 0.55,
+  "reader_kl_target": 0.3,
+  "reader_strengths": [
+    2.0,
+    3.0,
+    4.0,
+    5.0,
+    6.0,
+    7.0
+  ],
+  "reader_guard_rank": 3,
+  "reader_margin_target": -1.0,
+  "save_dtype": "bfloat16",
+  "bake": true
+}

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 %}

config.json

@@ -0,0 +1,71 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "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",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 40960,
+  "max_window_layers": 36,
+  "model_type": "qwen3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.5.4",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "5.5.4"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:692be9271934e6f753bfef394735f0efeafec50333d7197760dc1576d112afa8
+size 16381517208

report.json

@@ -0,0 +1,144 @@
+{
+  "model": "Qwen/Qwen3-8B",
+  "num_layers": 36,
+  "hidden_size": 4096,
+  "direction_layer": 29,
+  "refusal_subspace_rank": 3,
+  "max_refusal_rank": 3,
+  "multi_refusal": true,
+  "multi_refusal_clusters": 6,
+  "multi_refusal_min_norm": 0.08,
+  "multi_refusal_min_separation": 0.05,
+  "multi_refusal_min_coverage": 0.05,
+  "initial_separation": 344.6553,
+  "baseline_refusal_rate": 0.9167,
+  "baseline_eval_n": 24,
+  "edited_refusal_rate": 0.2286,
+  "refusal_metric": "classifier + weak guard",
+  "harmless_kl_nats": 0.1198,
+  "kl_backoff_steps": 0,
+  "kl_layer_trim_steps": 0,
+  "repair_steps": 1,
+  "residual_repair": [],
+  "guard_history": [
+    {
+      "iter": 0,
+      "separation": 166.7894,
+      "ratio": 0.4839,
+      "rank": 2,
+      "refusal": 0.4375,
+      "kl": 0.0248
+    },
+    {
+      "iter": 1,
+      "separation": 124.5732,
+      "ratio": 0.3614,
+      "rank": 3,
+      "refusal": 0.2188,
+      "kl": 0.0558
+    }
+  ],
+  "layer_alphas": [
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    1.269,
+    1.269,
+    1.269,
+    1.269,
+    1.269,
+    1.271,
+    1.275,
+    1.28,
+    1.287,
+    1.289,
+    1.292,
+    1.293,
+    3.244,
+    1.301,
+    1.306,
+    1.305,
+    1.307,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5,
+    0.5
+  ],
+  "ple_layer_alphas": [],
+  "ple_embed_alpha": 0.0,
+  "ple_model_projection_alpha": 0.0,
+  "embed_alpha": 0.044,
+  "head_alpha": 1.108,
+  "head_token_alpha": 0.0,
+  "preserve_rank": 8,
+  "preserve_source": "harmless",
+  "pruned_layers": [],
+  "layers_after_prune": 36,
+  "elapsed_sec": 23363.0,
+  "profile": "balanced",
+  "target_refusal": 0.05,
+  "max_kl": 0.12,
+  "kl_target": 0.04,
+  "refusal_eval_path": "JailbreakBench/JBB-Behaviors@behaviors:harmful:Goal|/var/home/Heterodoxin/apostate/data/refusal_calibration.txt",
+  "refusal_eval_n": 64,
+  "kl_positions": 8,
+  "kl_eval_path": "mlabonne/harmless_alpaca:test:text",
+  "kl_eval_n": 48,
+  "oblique_ablation": true,
+  "oblique_strength": 1.0,
+  "opt_capability": true,
+  "opt_capability_weight": 2.5,
+  "timings_sec": {
+    "load_model": 133.3,
+    "load_prompts": 5.5,
+    "baseline_refusal": 10.4,
+    "activation_fit": 17.0,
+    "causal_scores": 3.2,
+    "optimize_profile": 7281.8,
+    "guard": 561.6,
+    "refine_refusal": 148.9,
+    "validation_metrics": 0.0,
+    "repair": 6809.6,
+    "prune": 0.0,
+    "test_metrics": 8374.6,
+    "bake": 17.0
+  },
+  "command": "/var/home/Heterodoxin/apostate/apostate/cli.py --optimize --model Qwen/Qwen3-8B --output-dir /var/home/Heterodoxin/ablate_work/qwen3-8b-apostate --oblique-predictive --target-refusal 0.05",
+  "optimized": true,
+  "best_params": {
+    "direction_source": "activations",
+    "direction_layer_frac": 0.8077589218069658,
+    "refusal_rank": 2,
+    "strength": 1.3186274690569553,
+    "band_center": 0.6036341398087844,
+    "band_width": 0.501380240257032,
+    "causal_mix": 0.039187792254320675,
+    "causal_power": 1.5656139251528192,
+    "direction_sign": 1.0,
+    "ablate_embed": true,
+    "embed_scale": 0.03324376130718834,
+    "ablate_head": true,
+    "head_scale": 0.01603687408719609,
+    "head_alpha": 1.1079553909920319
+  },
+  "best_trial": {
+    "refusal": 0.4375,
+    "kl": 0.0248,
+    "capability_logprob": -9.6356,
+    "capability_drift": 0.0
+  },
+  "n_trials": 16,
+  "baked_to": "/var/home/Heterodoxin/ablate_work/qwen3-8b-apostate"
+}

report.md

@@ -0,0 +1,132 @@
+# Apostate Run Report
+
+## Summary
+| Metric | Value |
+| --- | --- |
+| Base model | Qwen/Qwen3-8B |
+| Profile | balanced |
+| Output | /var/home/Heterodoxin/ablate_work/qwen3-8b-apostate |
+| Layers | 36 |
+| Hidden size | 4096 |
+| Direction layer | 29 |
+| Refusal rank | 3 |
+| Max refusal rank | 3 |
+| Multi refusal | True |
+| Multi clusters | 6 |
+| Multi min coverage | 0.05 |
+| Baseline refusal (n=24) | 91.7% |
+| Edited refusal | 22.9% |
+| Refusal metric | classifier + weak guard |
+| Harmless KL | 0.120 |
+| Target refusal | 5.0% |
+| KL target | 0.040 |
+| KL budget | 0.120 |
+| KL positions | 8 |
+| KL layer trims | 0 |
+| Repair steps | 1 |
+| Preserve rank | 8 |
+| Preserve source | harmless |
+| Capability penalty | True |
+| Elapsed | 23363.0 sec |
+
+## Command
+
+```text
+/var/home/Heterodoxin/apostate/apostate/cli.py --optimize --model Qwen/Qwen3-8B --output-dir /var/home/Heterodoxin/ablate_work/qwen3-8b-apostate --oblique-predictive --target-refusal 0.05
+```
+
+## Best Parameters
+| Parameter | Value |
+| --- | --- |
+| direction_source | activations |
+| direction_layer_frac | 0.8078 |
+| refusal_rank | 2 |
+| strength | 1.3186 |
+| band_center | 0.6036 |
+| band_width | 0.5014 |
+| causal_mix | 0.0392 |
+| causal_power | 1.5656 |
+| direction_sign | 1.0 |
+| ablate_embed | True |
+| embed_scale | 0.0332 |
+| ablate_head | True |
+| head_scale | 0.016 |
+| head_alpha | 1.108 |
+
+## Best Trial
+| Metric | Value |
+| --- | --- |
+| refusal | 0.4375 |
+| kl | 0.0248 |
+| capability_logprob | -9.6356 |
+| capability_drift | 0.0 |
+
+## Layer Alphas
+| Layer | Alpha |
+| --- | --- |
+| 0 | 0.500 |
+| 1 | 0.500 |
+| 2 | 0.500 |
+| 3 | 0.500 |
+| 4 | 0.500 |
+| 5 | 0.500 |
+| 6 | 0.500 |
+| 7 | 0.500 |
+| 8 | 0.500 |
+| 9 | 0.500 |
+| 10 | 0.500 |
+| 11 | 0.500 |
+| 12 | 0.500 |
+| 13 | 1.269 |
+| 14 | 1.269 |
+| 15 | 1.269 |
+| 16 | 1.269 |
+| 17 | 1.269 |
+| 18 | 1.271 |
+| 19 | 1.275 |
+| 20 | 1.280 |
+| 21 | 1.287 |
+| 22 | 1.289 |
+| 23 | 1.292 |
+| 24 | 1.293 |
+| 25 | 3.244 |
+| 26 | 1.301 |
+| 27 | 1.306 |
+| 28 | 1.305 |
+| 29 | 1.307 |
+| 30 | 0.500 |
+| 31 | 0.500 |
+| 32 | 0.500 |
+| 33 | 0.500 |
+| 34 | 0.500 |
+| 35 | 0.500 |
+
+## Guard History
+| iter | separation | ratio | rank | refusal | kl | reverted |
+| --- | --- | --- | --- | --- | --- | --- |
+| 0 | 166.7894 | 0.4839 | 2 | 0.4375 | 0.0248 |  |
+| 1 | 124.5732 | 0.3614 | 3 | 0.2188 | 0.0558 |  |
+
+## Timings
+| Phase | Seconds |
+| --- | --- |
+| load_model | 133.3 |
+| load_prompts | 5.5 |
+| baseline_refusal | 10.4 |
+| activation_fit | 17.0 |
+| causal_scores | 3.2 |
+| optimize_profile | 7281.8 |
+| guard | 561.6 |
+| refine_refusal | 148.9 |
+| validation_metrics | 0.0 |
+| repair | 6809.6 |
+| prune | 0.0 |
+| test_metrics | 8374.6 |
+| bake | 17.0 |
+
+## Measurement
+| field | value |
+| --- | --- |
+| refusal judge | classifier + weak guard |
+| preservation metric | harmless kl |
+| capability suites | gsm8k, humaneval, mbpp |

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79cb3c783570f1b8fe73b9ed530ae50cae9ce4b6344c0b5edefc50478847eaa4
+size 11422817

tokenizer_config.json

@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "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": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}