Upload evidence binding adapter artifacts to evidence_adapter/

.gitattributes

@@ -36,3 +36,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text
 toolspec_adapter/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 uncertainty_adapter/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+evidence_adapter/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text

evidence_adapter/adapter/README.md

@@ -0,0 +1,203 @@
+---
+library_name: peft
+tags:
+- lora
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.0

evidence_adapter/adapter/adapter_config.json

@@ -0,0 +1,54 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": null,
+  "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": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.0",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "surface_host.evidence_binding.adapter",
+    "surface_host.tool_receipt_binding.adapter",
+    "surface_host.citation_binding.adapter",
+    "surface_host.reverse_engineering_binding.adapter",
+    "surface_host.runtime_binding.adapter",
+    "surface_host.validator_receipt_bridge.adapter",
+    "surface_host.selfcheck_binding.adapter",
+    "surface_host.execution_binding.adapter",
+    "surface_host.patch_binding.adapter",
+    "surface_host.proof_carrying_hints.bridge",
+    "surface_host.provenance_binding.adapter",
+    "surface_host.worktree_binding.adapter",
+    "surface_host.claim_extractor.adapter"
+  ],
+  "target_parameters": null,
+  "task_type": "FEATURE_EXTRACTION",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

evidence_adapter/adapter/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d2e07f819fdde8c0954f3f6d21bd5877e26627d145d125a959669cfc6f9f38e
+size 855584

evidence_adapter/best_checkpoint_manifest.json

@@ -0,0 +1,29 @@
+{
+  "best_epoch": 4,
+  "best_quality_score": 0.5592996196176252,
+  "eval_metrics": {
+    "avg_binary_accuracy": 0.8444444444444446,
+    "citation_binding_required_accuracy": 1.0,
+    "contradiction_visible_accuracy": 0.7692307692307693,
+    "count": 65,
+    "eval_batches": 65,
+    "eval_loss": 4.7871557712554935,
+    "execution_needed_accuracy": 0.8461538461538461,
+    "mode_accuracy": 0.5692307692307692,
+    "next_action_accuracy": 0.5538461538461539,
+    "patch_continuity_accuracy": 0.6461538461538462,
+    "proof_carrying_compatible_accuracy": 0.8,
+    "provenance_accuracy": 0.6461538461538462,
+    "quality_score": 0.5592996196176252,
+    "reverse_engineering_ready_accuracy": 0.8461538461538461,
+    "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+    "validator_required_accuracy": 1.0,
+    "verdict_accuracy": 0.6307692307692307,
+    "worktree_safe_accuracy": 0.9230769230769231
+  },
+  "train_metrics": {
+    "loss": 0.23536107725985758,
+    "micro_batches": 182,
+    "optimizer_steps": 12
+  }
+}

evidence_adapter/epoch_history.json

@@ -0,0 +1,124 @@
+{
+  "epochs": [
+    {
+      "epoch": 1,
+      "eval_metrics": {
+        "avg_binary_accuracy": 0.823931623931624,
+        "citation_binding_required_accuracy": 1.0,
+        "contradiction_visible_accuracy": 0.6461538461538462,
+        "count": 65,
+        "eval_batches": 65,
+        "eval_loss": 5.821129791553204,
+        "execution_needed_accuracy": 0.8461538461538461,
+        "mode_accuracy": 0.4461538461538462,
+        "next_action_accuracy": 0.4307692307692308,
+        "patch_continuity_accuracy": 0.6153846153846154,
+        "proof_carrying_compatible_accuracy": 0.7692307692307693,
+        "provenance_accuracy": 0.6153846153846154,
+        "quality_score": 0.4452868058783377,
+        "reverse_engineering_ready_accuracy": 0.8461538461538461,
+        "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+        "validator_required_accuracy": 1.0,
+        "verdict_accuracy": 0.47692307692307695,
+        "worktree_safe_accuracy": 0.9230769230769231
+      },
+      "improved": true,
+      "quality_score": 0.4452868058783377,
+      "train_metrics": {
+        "loss": 0.4255874376375597,
+        "micro_batches": 182,
+        "optimizer_steps": 12
+      }
+    },
+    {
+      "epoch": 2,
+      "eval_metrics": {
+        "avg_binary_accuracy": 0.8376068376068376,
+        "citation_binding_required_accuracy": 1.0,
+        "contradiction_visible_accuracy": 0.7384615384615385,
+        "count": 65,
+        "eval_batches": 65,
+        "eval_loss": 4.975605465815618,
+        "execution_needed_accuracy": 0.8461538461538461,
+        "mode_accuracy": 0.5538461538461539,
+        "next_action_accuracy": 0.5538461538461539,
+        "patch_continuity_accuracy": 0.6461538461538462,
+        "proof_carrying_compatible_accuracy": 0.7692307692307693,
+        "provenance_accuracy": 0.6,
+        "quality_score": 0.5249323351281322,
+        "reverse_engineering_ready_accuracy": 0.8461538461538461,
+        "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+        "validator_required_accuracy": 1.0,
+        "verdict_accuracy": 0.5692307692307692,
+        "worktree_safe_accuracy": 0.9230769230769231
+      },
+      "improved": true,
+      "quality_score": 0.5249323351281322,
+      "train_metrics": {
+        "loss": 0.28522092685267164,
+        "micro_batches": 182,
+        "optimizer_steps": 12
+      }
+    },
+    {
+      "epoch": 3,
+      "eval_metrics": {
+        "avg_binary_accuracy": 0.8444444444444446,
+        "citation_binding_required_accuracy": 1.0,
+        "contradiction_visible_accuracy": 0.7692307692307693,
+        "count": 65,
+        "eval_batches": 65,
+        "eval_loss": 4.801326029117291,
+        "execution_needed_accuracy": 0.8461538461538461,
+        "mode_accuracy": 0.5538461538461539,
+        "next_action_accuracy": 0.5538461538461539,
+        "patch_continuity_accuracy": 0.6461538461538462,
+        "proof_carrying_compatible_accuracy": 0.8,
+        "provenance_accuracy": 0.6307692307692307,
+        "quality_score": 0.5544008298450047,
+        "reverse_engineering_ready_accuracy": 0.8461538461538461,
+        "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+        "validator_required_accuracy": 1.0,
+        "verdict_accuracy": 0.6307692307692307,
+        "worktree_safe_accuracy": 0.9230769230769231
+      },
+      "improved": true,
+      "quality_score": 0.5544008298450047,
+      "train_metrics": {
+        "loss": 0.24240652721498038,
+        "micro_batches": 182,
+        "optimizer_steps": 12
+      }
+    },
+    {
+      "epoch": 4,
+      "eval_metrics": {
+        "avg_binary_accuracy": 0.8444444444444446,
+        "citation_binding_required_accuracy": 1.0,
+        "contradiction_visible_accuracy": 0.7692307692307693,
+        "count": 65,
+        "eval_batches": 65,
+        "eval_loss": 4.7871557712554935,
+        "execution_needed_accuracy": 0.8461538461538461,
+        "mode_accuracy": 0.5692307692307692,
+        "next_action_accuracy": 0.5538461538461539,
+        "patch_continuity_accuracy": 0.6461538461538462,
+        "proof_carrying_compatible_accuracy": 0.8,
+        "provenance_accuracy": 0.6461538461538462,
+        "quality_score": 0.5592996196176252,
+        "reverse_engineering_ready_accuracy": 0.8461538461538461,
+        "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+        "validator_required_accuracy": 1.0,
+        "verdict_accuracy": 0.6307692307692307,
+        "worktree_safe_accuracy": 0.9230769230769231
+      },
+      "improved": true,
+      "quality_score": 0.5592996196176252,
+      "train_metrics": {
+        "loss": 0.23536107725985758,
+        "micro_batches": 182,
+        "optimizer_steps": 12
+      }
+    }
+  ]
+}

evidence_adapter/evidence_binding_adapter_plan.json

@@ -0,0 +1,223 @@
+{
+  "backbone": "/public/wang_libo/veriloop_coder_e1/model",
+  "dataset_summary": {
+    "eval_size": 65,
+    "mode_vocab": [
+      "direct_support",
+      "multi_support",
+      "conflict_visible",
+      "evidence_gap",
+      "execution_needed",
+      "high_risk_unbound",
+      "validator_negation",
+      "patch_regression",
+      "worktree_conflict",
+      "tool_selfcheck_confirmed",
+      "tool_selfcheck_negated",
+      "reverse_engineering_bindable",
+      "reverse_engineering_gap"
+    ],
+    "modes": [
+      "conflict_visible",
+      "direct_support",
+      "evidence_gap",
+      "execution_needed",
+      "high_risk_unbound",
+      "multi_support",
+      "patch_regression",
+      "reverse_engineering_bindable",
+      "reverse_engineering_gap",
+      "tool_selfcheck_confirmed",
+      "tool_selfcheck_negated",
+      "validator_negation",
+      "worktree_conflict"
+    ],
+    "next_action_vocab": [
+      "none",
+      "validator_review",
+      "sandbox_exec",
+      "selfcheck_exec",
+      "bounded_observation",
+      "fail_closed",
+      "worktree_reconcile"
+    ],
+    "provenance_vocab": [
+      "inadequate",
+      "partial",
+      "adequate"
+    ],
+    "train_size": 182,
+    "verdict_vocab": [
+      "supported",
+      "conflicted",
+      "insufficient",
+      "execution_required"
+    ]
+  },
+  "excluded_surfaces": [
+    "(^|\\.)lm_head($|\\.)::Do not retune final token head; too broad and evaluation-heavy.",
+    "(^|\\.)embed_tokens($|\\.)::Embedding surgery risks broad semantic drift.",
+    "(^|\\.)norm($|\\.)::Global norm tuning can destabilize calibration across scenes.",
+    "attnres|attention_residual::Block AttnRes may be mounted structurally but is never a PEFT target.",
+    "dualpath::DualPath is serving/runtime infrastructure only.",
+    "mhc|hyper[-_]?connection::mHC-inspired stability hooks remain structural, not PEFT surfaces.",
+    "rope|rotary::RoPE/context surgery is handled architecturally, not by narrow PEFT here.",
+    "kvcache|kv_cache::KV-cache runtime surfaces are not PEFT targets.",
+    "(^|\\.)memory(_store|_bank)?($|\\.)::Persistent memory stores are harness/runtime policy surfaces, not PEFT targets."
+  ],
+  "notes": [
+    "Primary route is host-surface-first evidence-binding training.",
+    "Claim↔evidence fidelity, contradiction visibility, provenance discipline, validator receipts, execution-needed escalation, tool self-check compatibility, reverse-engineering boundedness, and proof-carrying hint obedience are first-class signals.",
+    "DualPath, Block AttnRes, mHC hooks, visual branches, and MoE routers/experts remain structurally excluded.",
+    "This adapter should improve evidence-gate obedience, not broad free-form coding behavior.",
+    "Target coverage is rooted in the full evidence-binding decision graph rather than the selector-only subset, so execution/tool/citation/reverse-engineering surfaces are not silently left untuned."
+  ],
+  "peft_method": "lora_narrow",
+  "product_line": "veriloop_coder",
+  "recipe": {
+    "adapter_family": "evidence_binding",
+    "backbone": "/public/wang_libo/veriloop_coder_e1/model",
+    "backbone_family": "qwen_dense",
+    "excluded_patterns": [
+      "(?i)\\bdualpath\\b",
+      "(?i)\\bmhc\\b",
+      "(?i)\\bfull[_\\- ]?attnres\\b",
+      "(?i)\\battnres(_full)?\\b",
+      "(?i)\\brouter\\b",
+      "(?i)\\bexperts?\\b",
+      "(?i)\\bmoe\\b.*\\b(gate|router|expert)\\b",
+      "(?i)\\brope\\b.*\\b(freq|inv_freq|theta|rotary)\\b",
+      "(?i)\\bkvcache\\b",
+      "(?i)\\bposition_embedding\\b",
+      "(?i)\\bembed(tokens|ding)?\\b",
+      "(?i)\\blm_head\\b"
+    ],
+    "harness_constraints": [
+      "Harness Engineering remains the primary convergence layer.",
+      "Adapter must not bypass runtime orchestrator / validator / rollback loops.",
+      "Adapter outputs remain subordinate to VeriLoop control-plane decisions.",
+      "Adapter must not create hidden prompt-style memory authority.",
+      "Adapter must support claim-evidence binding rather than generic retrieval verbosity.",
+      "Unbound claims must remain rejectable or demotable."
+    ],
+    "hyperparams": {
+      "alpha": 16,
+      "bias": "none",
+      "dropout": 0.05,
+      "fan_in_fan_out": false,
+      "modules_to_save": [],
+      "r": 8,
+      "task_type": "CAUSAL_LM"
+    },
+    "merge_policy": "merge_after_guard",
+    "metadata": {
+      "allow_backbone_bridge": false,
+      "allow_vla_action_expert": false,
+      "evidence_binding_training": true,
+      "harness_first": true,
+      "policy_target_floor_applied": true,
+      "prefer_explicit_heads": true,
+      "prefer_qlora_for_backbone_bridge": false,
+      "require_harness_first": true,
+      "reverse_engineering_readiness": true,
+      "selector_group_count": 1,
+      "strict_narrow_scope": true,
+      "tool_selfcheck_readiness": true,
+      "trainer": "veriloop.evidence_binding_adapter_trainer.v9.qwen36"
+    },
+    "notes": [
+      "Backbone bridge tuning disabled explicitly; selector stays on custom surfaces or no-op.",
+      "Backbone family inferred as qwen_dense.",
+      "PEFT method resolved as lora_narrow.",
+      "Recipe is harness-first: runtime convergence remains in VeriLoop control-plane + harness, not in broad weight surgery.",
+      "Block AttnRes, DualPath, mHC hooks, RoPE, KV-cache, and broad MoE routing remain structurally excluded."
+    ],
+    "peft_method": "lora_narrow",
+    "precision_policy": "auto",
+    "product_line": "veriloop_coder",
+    "regression_requirements": [
+      "Must pass PEFT regression guard structural policy checks.",
+      "Must not introduce forbidden backbone/serving structural targets.",
+      "Must preserve harness regression envelope for the selected product line.",
+      "Evidence-conclusion alignment must not regress.",
+      "High-risk fabrication rate must not increase."
+    ],
+    "target_groups": [
+      {
+        "alpha": 16,
+        "dropout": 0.0,
+        "name": "group_1_custom_control_head",
+        "rank": 8,
+        "rationale": "Evidence alignment should land on explicit binding surfaces first.",
+        "surface": "custom_control_head",
+        "target_modules": [
+          "claim_extractor.adapter",
+          "evidence_binding.adapter",
+          "proof_carrying_hints.bridge"
+        ]
+      },
+      {
+        "alpha": 16,
+        "dropout": 0.0,
+        "name": "group_policy_expanded_evidence_binding_surface_set",
+        "rank": 8,
+        "rationale": "Expand selector-narrow targets to the full host-side evidence-binding decision graph used by verdict/provenance/next-action heads.",
+        "surface": "policy_expanded_evidence_binding_surface_set",
+        "target_modules": [
+          "provenance_binding.adapter",
+          "validator_receipt_bridge.adapter",
+          "tool_receipt_binding.adapter",
+          "execution_binding.adapter",
+          "citation_binding.adapter",
+          "runtime_binding.adapter",
+          "selfcheck_binding.adapter",
+          "reverse_engineering_binding.adapter",
+          "patch_binding.adapter",
+          "worktree_binding.adapter"
+        ]
+      }
+    ],
+    "target_modules": [
+      "claim_extractor.adapter",
+      "evidence_binding.adapter",
+      "proof_carrying_hints.bridge",
+      "provenance_binding.adapter",
+      "validator_receipt_bridge.adapter",
+      "tool_receipt_binding.adapter",
+      "execution_binding.adapter",
+      "citation_binding.adapter",
+      "runtime_binding.adapter",
+      "selfcheck_binding.adapter",
+      "reverse_engineering_binding.adapter",
+      "patch_binding.adapter",
+      "worktree_binding.adapter"
+    ],
+    "version": "veriloop.lora_recipe_veriloop.v2"
+  },
+  "selected_surfaces": [
+    "custom_control_head",
+    "policy_expanded_evidence_binding_surface_set"
+  ],
+  "selected_target_modules": [
+    "claim_extractor.adapter",
+    "evidence_binding.adapter",
+    "proof_carrying_hints.bridge",
+    "provenance_binding.adapter",
+    "validator_receipt_bridge.adapter",
+    "tool_receipt_binding.adapter",
+    "execution_binding.adapter",
+    "citation_binding.adapter",
+    "runtime_binding.adapter",
+    "selfcheck_binding.adapter",
+    "reverse_engineering_binding.adapter",
+    "patch_binding.adapter",
+    "worktree_binding.adapter"
+  ],
+  "selection_mode": "minimal",
+  "version": "veriloop.evidence_binding_adapter_trainer.v9.qwen36",
+  "warnings": [
+    "Harness Engineering is primary; PEFT is limited to obedience-facing, interface-facing support surfaces.",
+    "Backbone bridge tuning disabled explicitly; selector stays on custom surfaces or no-op.",
+    "Selector target set was narrower than the evidence-binding decision graph; host-side policy floor expanded the PEFT targets."
+  ]
+}

evidence_adapter/evidence_binding_adapter_train_result.json

@@ -0,0 +1,338 @@
+{
+  "artifacts": {
+    "adapter_dir": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/adapter",
+    "best_checkpoint_manifest": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/best_checkpoint_manifest.json",
+    "epoch_history": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/epoch_history.json",
+    "eval_jsonl": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/evidence_binding_eval.jsonl",
+    "evidence_binding_head": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/evidence_binding_head.pt",
+    "host_manifest": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/host_manifest.json",
+    "plan_json": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/evidence_binding_adapter_plan.json",
+    "tokenizer_dir": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/tokenizer",
+    "train_jsonl": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/evidence_binding_train.jsonl",
+    "training_manifest": "/private/wang_libo/veriloop_coder_e1/outputs/evidence_binding_qwen36_run1/evidence_binding_training_manifest.json"
+  },
+  "dataset": {
+    "eval_size": 65,
+    "mode_vocab": [
+      "direct_support",
+      "multi_support",
+      "conflict_visible",
+      "evidence_gap",
+      "execution_needed",
+      "high_risk_unbound",
+      "validator_negation",
+      "patch_regression",
+      "worktree_conflict",
+      "tool_selfcheck_confirmed",
+      "tool_selfcheck_negated",
+      "reverse_engineering_bindable",
+      "reverse_engineering_gap"
+    ],
+    "modes": [
+      "conflict_visible",
+      "direct_support",
+      "evidence_gap",
+      "execution_needed",
+      "high_risk_unbound",
+      "multi_support",
+      "patch_regression",
+      "reverse_engineering_bindable",
+      "reverse_engineering_gap",
+      "tool_selfcheck_confirmed",
+      "tool_selfcheck_negated",
+      "validator_negation",
+      "worktree_conflict"
+    ],
+    "next_action_vocab": [
+      "none",
+      "validator_review",
+      "sandbox_exec",
+      "selfcheck_exec",
+      "bounded_observation",
+      "fail_closed",
+      "worktree_reconcile"
+    ],
+    "provenance_vocab": [
+      "inadequate",
+      "partial",
+      "adequate"
+    ],
+    "train_size": 182,
+    "verdict_vocab": [
+      "supported",
+      "conflicted",
+      "insufficient",
+      "execution_required"
+    ]
+  },
+  "eval_metrics": {
+    "adapter_exported": true,
+    "auto_lora_from_ia3": false,
+    "avg_binary_accuracy": 0.8444444444444446,
+    "best_epoch": 4,
+    "best_quality_score": 0.5592996196176252,
+    "citation_binding_required_accuracy": 1.0,
+    "contradiction_visible_accuracy": 0.7692307692307693,
+    "count": 65,
+    "eval_batches": 65,
+    "eval_loss": 4.7871557712554935,
+    "execution_needed_accuracy": 0.8461538461538461,
+    "mode_accuracy": 0.5692307692307692,
+    "next_action_accuracy": 0.5538461538461539,
+    "patch_continuity_accuracy": 0.6461538461538462,
+    "peft_method": "lora_narrow",
+    "proof_carrying_compatible_accuracy": 0.8,
+    "provenance_accuracy": 0.6461538461538462,
+    "quality_score": 0.5592996196176252,
+    "reverse_engineering_ready_accuracy": 0.8461538461538461,
+    "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+    "used_peft": true,
+    "validator_required_accuracy": 1.0,
+    "verdict_accuracy": 0.6307692307692307,
+    "worktree_safe_accuracy": 0.9230769230769231
+  },
+  "plan": {
+    "backbone": "/public/wang_libo/veriloop_coder_e1/model",
+    "dataset_summary": {
+      "eval_size": 65,
+      "mode_vocab": [
+        "direct_support",
+        "multi_support",
+        "conflict_visible",
+        "evidence_gap",
+        "execution_needed",
+        "high_risk_unbound",
+        "validator_negation",
+        "patch_regression",
+        "worktree_conflict",
+        "tool_selfcheck_confirmed",
+        "tool_selfcheck_negated",
+        "reverse_engineering_bindable",
+        "reverse_engineering_gap"
+      ],
+      "modes": [
+        "conflict_visible",
+        "direct_support",
+        "evidence_gap",
+        "execution_needed",
+        "high_risk_unbound",
+        "multi_support",
+        "patch_regression",
+        "reverse_engineering_bindable",
+        "reverse_engineering_gap",
+        "tool_selfcheck_confirmed",
+        "tool_selfcheck_negated",
+        "validator_negation",
+        "worktree_conflict"
+      ],
+      "next_action_vocab": [
+        "none",
+        "validator_review",
+        "sandbox_exec",
+        "selfcheck_exec",
+        "bounded_observation",
+        "fail_closed",
+        "worktree_reconcile"
+      ],
+      "provenance_vocab": [
+        "inadequate",
+        "partial",
+        "adequate"
+      ],
+      "train_size": 182,
+      "verdict_vocab": [
+        "supported",
+        "conflicted",
+        "insufficient",
+        "execution_required"
+      ]
+    },
+    "excluded_surfaces": [
+      "(^|\\.)lm_head($|\\.)::Do not retune final token head; too broad and evaluation-heavy.",
+      "(^|\\.)embed_tokens($|\\.)::Embedding surgery risks broad semantic drift.",
+      "(^|\\.)norm($|\\.)::Global norm tuning can destabilize calibration across scenes.",
+      "attnres|attention_residual::Block AttnRes may be mounted structurally but is never a PEFT target.",
+      "dualpath::DualPath is serving/runtime infrastructure only.",
+      "mhc|hyper[-_]?connection::mHC-inspired stability hooks remain structural, not PEFT surfaces.",
+      "rope|rotary::RoPE/context surgery is handled architecturally, not by narrow PEFT here.",
+      "kvcache|kv_cache::KV-cache runtime surfaces are not PEFT targets.",
+      "(^|\\.)memory(_store|_bank)?($|\\.)::Persistent memory stores are harness/runtime policy surfaces, not PEFT targets."
+    ],
+    "notes": [
+      "Primary route is host-surface-first evidence-binding training.",
+      "Claim↔evidence fidelity, contradiction visibility, provenance discipline, validator receipts, execution-needed escalation, tool self-check compatibility, reverse-engineering boundedness, and proof-carrying hint obedience are first-class signals.",
+      "DualPath, Block AttnRes, mHC hooks, visual branches, and MoE routers/experts remain structurally excluded.",
+      "This adapter should improve evidence-gate obedience, not broad free-form coding behavior.",
+      "Target coverage is rooted in the full evidence-binding decision graph rather than the selector-only subset, so execution/tool/citation/reverse-engineering surfaces are not silently left untuned."
+    ],
+    "peft_method": "lora_narrow",
+    "product_line": "veriloop_coder",
+    "recipe": {
+      "adapter_family": "evidence_binding",
+      "backbone": "/public/wang_libo/veriloop_coder_e1/model",
+      "backbone_family": "qwen_dense",
+      "excluded_patterns": [
+        "(?i)\\bdualpath\\b",
+        "(?i)\\bmhc\\b",
+        "(?i)\\bfull[_\\- ]?attnres\\b",
+        "(?i)\\battnres(_full)?\\b",
+        "(?i)\\brouter\\b",
+        "(?i)\\bexperts?\\b",
+        "(?i)\\bmoe\\b.*\\b(gate|router|expert)\\b",
+        "(?i)\\brope\\b.*\\b(freq|inv_freq|theta|rotary)\\b",
+        "(?i)\\bkvcache\\b",
+        "(?i)\\bposition_embedding\\b",
+        "(?i)\\bembed(tokens|ding)?\\b",
+        "(?i)\\blm_head\\b"
+      ],
+      "harness_constraints": [
+        "Harness Engineering remains the primary convergence layer.",
+        "Adapter must not bypass runtime orchestrator / validator / rollback loops.",
+        "Adapter outputs remain subordinate to VeriLoop control-plane decisions.",
+        "Adapter must not create hidden prompt-style memory authority.",
+        "Adapter must support claim-evidence binding rather than generic retrieval verbosity.",
+        "Unbound claims must remain rejectable or demotable."
+      ],
+      "hyperparams": {
+        "alpha": 16,
+        "bias": "none",
+        "dropout": 0.05,
+        "fan_in_fan_out": false,
+        "modules_to_save": [],
+        "r": 8,
+        "task_type": "CAUSAL_LM"
+      },
+      "merge_policy": "merge_after_guard",
+      "metadata": {
+        "allow_backbone_bridge": false,
+        "allow_vla_action_expert": false,
+        "evidence_binding_training": true,
+        "harness_first": true,
+        "policy_target_floor_applied": true,
+        "prefer_explicit_heads": true,
+        "prefer_qlora_for_backbone_bridge": false,
+        "require_harness_first": true,
+        "reverse_engineering_readiness": true,
+        "selector_group_count": 1,
+        "strict_narrow_scope": true,
+        "tool_selfcheck_readiness": true,
+        "trainer": "veriloop.evidence_binding_adapter_trainer.v9.qwen36"
+      },
+      "notes": [
+        "Backbone bridge tuning disabled explicitly; selector stays on custom surfaces or no-op.",
+        "Backbone family inferred as qwen_dense.",
+        "PEFT method resolved as lora_narrow.",
+        "Recipe is harness-first: runtime convergence remains in VeriLoop control-plane + harness, not in broad weight surgery.",
+        "Block AttnRes, DualPath, mHC hooks, RoPE, KV-cache, and broad MoE routing remain structurally excluded."
+      ],
+      "peft_method": "lora_narrow",
+      "precision_policy": "auto",
+      "product_line": "veriloop_coder",
+      "regression_requirements": [
+        "Must pass PEFT regression guard structural policy checks.",
+        "Must not introduce forbidden backbone/serving structural targets.",
+        "Must preserve harness regression envelope for the selected product line.",
+        "Evidence-conclusion alignment must not regress.",
+        "High-risk fabrication rate must not increase."
+      ],
+      "target_groups": [
+        {
+          "alpha": 16,
+          "dropout": 0.0,
+          "name": "group_1_custom_control_head",
+          "rank": 8,
+          "rationale": "Evidence alignment should land on explicit binding surfaces first.",
+          "surface": "custom_control_head",
+          "target_modules": [
+            "claim_extractor.adapter",
+            "evidence_binding.adapter",
+            "proof_carrying_hints.bridge"
+          ]
+        },
+        {
+          "alpha": 16,
+          "dropout": 0.0,
+          "name": "group_policy_expanded_evidence_binding_surface_set",
+          "rank": 8,
+          "rationale": "Expand selector-narrow targets to the full host-side evidence-binding decision graph used by verdict/provenance/next-action heads.",
+          "surface": "policy_expanded_evidence_binding_surface_set",
+          "target_modules": [
+            "provenance_binding.adapter",
+            "validator_receipt_bridge.adapter",
+            "tool_receipt_binding.adapter",
+            "execution_binding.adapter",
+            "citation_binding.adapter",
+            "runtime_binding.adapter",
+            "selfcheck_binding.adapter",
+            "reverse_engineering_binding.adapter",
+            "patch_binding.adapter",
+            "worktree_binding.adapter"
+          ]
+        }
+      ],
+      "target_modules": [
+        "claim_extractor.adapter",
+        "evidence_binding.adapter",
+        "proof_carrying_hints.bridge",
+        "provenance_binding.adapter",
+        "validator_receipt_bridge.adapter",
+        "tool_receipt_binding.adapter",
+        "execution_binding.adapter",
+        "citation_binding.adapter",
+        "runtime_binding.adapter",
+        "selfcheck_binding.adapter",
+        "reverse_engineering_binding.adapter",
+        "patch_binding.adapter",
+        "worktree_binding.adapter"
+      ],
+      "version": "veriloop.lora_recipe_veriloop.v2"
+    },
+    "selected_surfaces": [
+      "custom_control_head",
+      "policy_expanded_evidence_binding_surface_set"
+    ],
+    "selected_target_modules": [
+      "claim_extractor.adapter",
+      "evidence_binding.adapter",
+      "proof_carrying_hints.bridge",
+      "provenance_binding.adapter",
+      "validator_receipt_bridge.adapter",
+      "tool_receipt_binding.adapter",
+      "execution_binding.adapter",
+      "citation_binding.adapter",
+      "runtime_binding.adapter",
+      "selfcheck_binding.adapter",
+      "reverse_engineering_binding.adapter",
+      "patch_binding.adapter",
+      "worktree_binding.adapter"
+    ],
+    "selection_mode": "minimal",
+    "version": "veriloop.evidence_binding_adapter_trainer.v9.qwen36",
+    "warnings": [
+      "Harness Engineering is primary; PEFT is limited to obedience-facing, interface-facing support surfaces.",
+      "Backbone bridge tuning disabled explicitly; selector stays on custom surfaces or no-op.",
+      "Selector target set was narrower than the evidence-binding decision graph; host-side policy floor expanded the PEFT targets."
+    ]
+  },
+  "status": "trained",
+  "train_metrics": {
+    "adapter_exported": true,
+    "auto_lora_from_ia3": false,
+    "best_epoch": 4,
+    "best_quality_score": 0.5592996196176252,
+    "epochs_completed": 4,
+    "loss": 0.23536107725985758,
+    "micro_batches": 182,
+    "micro_batches_total": 728,
+    "optimizer_steps": 12,
+    "optimizer_steps_total": 48,
+    "peft_method": "lora_narrow",
+    "used_peft": true
+  },
+  "version": "veriloop.evidence_binding_adapter_trainer.v9.qwen36",
+  "warnings": [
+    "Harness Engineering is primary; PEFT is limited to obedience-facing, interface-facing support surfaces.",
+    "Backbone bridge tuning disabled explicitly; selector stays on custom surfaces or no-op.",
+    "Selector target set was narrower than the evidence-binding decision graph; host-side policy floor expanded the PEFT targets."
+  ]
+}

evidence_adapter/evidence_binding_head.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e85ad23a3e89f3d07cbbe6352f185277569160e403a41acdbf36f3fb1e182bc0
+size 340358929

evidence_adapter/evidence_binding_training_manifest.json

@@ -0,0 +1,120 @@
+{
+  "adapter_exported": true,
+  "dataset_summary": {
+    "eval_size": 65,
+    "mode_vocab": [
+      "direct_support",
+      "multi_support",
+      "conflict_visible",
+      "evidence_gap",
+      "execution_needed",
+      "high_risk_unbound",
+      "validator_negation",
+      "patch_regression",
+      "worktree_conflict",
+      "tool_selfcheck_confirmed",
+      "tool_selfcheck_negated",
+      "reverse_engineering_bindable",
+      "reverse_engineering_gap"
+    ],
+    "modes": [
+      "conflict_visible",
+      "direct_support",
+      "evidence_gap",
+      "execution_needed",
+      "high_risk_unbound",
+      "multi_support",
+      "patch_regression",
+      "reverse_engineering_bindable",
+      "reverse_engineering_gap",
+      "tool_selfcheck_confirmed",
+      "tool_selfcheck_negated",
+      "validator_negation",
+      "worktree_conflict"
+    ],
+    "next_action_vocab": [
+      "none",
+      "validator_review",
+      "sandbox_exec",
+      "selfcheck_exec",
+      "bounded_observation",
+      "fail_closed",
+      "worktree_reconcile"
+    ],
+    "provenance_vocab": [
+      "inadequate",
+      "partial",
+      "adequate"
+    ],
+    "train_size": 182,
+    "verdict_vocab": [
+      "supported",
+      "conflicted",
+      "insufficient",
+      "execution_required"
+    ]
+  },
+  "eval_metrics": {
+    "adapter_exported": true,
+    "auto_lora_from_ia3": false,
+    "avg_binary_accuracy": 0.8444444444444446,
+    "best_epoch": 4,
+    "best_quality_score": 0.5592996196176252,
+    "citation_binding_required_accuracy": 1.0,
+    "contradiction_visible_accuracy": 0.7692307692307693,
+    "count": 65,
+    "eval_batches": 65,
+    "eval_loss": 4.7871557712554935,
+    "execution_needed_accuracy": 0.8461538461538461,
+    "mode_accuracy": 0.5692307692307692,
+    "next_action_accuracy": 0.5538461538461539,
+    "patch_continuity_accuracy": 0.6461538461538462,
+    "peft_method": "lora_narrow",
+    "proof_carrying_compatible_accuracy": 0.8,
+    "provenance_accuracy": 0.6461538461538462,
+    "quality_score": 0.5592996196176252,
+    "reverse_engineering_ready_accuracy": 0.8461538461538461,
+    "tool_selfcheck_needed_accuracy": 0.7692307692307693,
+    "used_peft": true,
+    "validator_required_accuracy": 1.0,
+    "verdict_accuracy": 0.6307692307692307,
+    "worktree_safe_accuracy": 0.9230769230769231
+  },
+  "load_meta": {
+    "chosen_class": "AutoModelForCausalLM",
+    "hidden_size": 2048,
+    "quantization_mode": "4bit"
+  },
+  "peft_method": "lora_narrow",
+  "selected_target_modules": [
+    "claim_extractor.adapter",
+    "evidence_binding.adapter",
+    "proof_carrying_hints.bridge",
+    "provenance_binding.adapter",
+    "validator_receipt_bridge.adapter",
+    "tool_receipt_binding.adapter",
+    "execution_binding.adapter",
+    "citation_binding.adapter",
+    "runtime_binding.adapter",
+    "selfcheck_binding.adapter",
+    "reverse_engineering_binding.adapter",
+    "patch_binding.adapter",
+    "worktree_binding.adapter"
+  ],
+  "status": "trained",
+  "train_metrics": {
+    "adapter_exported": true,
+    "auto_lora_from_ia3": false,
+    "best_epoch": 4,
+    "best_quality_score": 0.5592996196176252,
+    "epochs_completed": 4,
+    "loss": 0.23536107725985758,
+    "micro_batches": 182,
+    "micro_batches_total": 728,
+    "optimizer_steps": 12,
+    "optimizer_steps_total": 48,
+    "peft_method": "lora_narrow",
+    "used_peft": true
+  },
+  "used_peft": true
+}

evidence_adapter/host_manifest.json

@@ -0,0 +1,268 @@
+{
+  "hidden_size": 2048,
+  "host_config": {
+    "attn_implementation": null,
+    "backbone_name_or_path": "/public/wang_libo/veriloop_coder_e1/model",
+    "device_map": null,
+    "dtype": null,
+    "evidence_rank_hint": 8,
+    "expose_backbone_inventory": false,
+    "freeze_backbone": true,
+    "hidden_size_override": 2048,
+    "host_dropout": 0.0,
+    "identity_rank_hint": 8,
+    "load_backbone_weights": false,
+    "local_files_only": true,
+    "low_cpu_mem_usage": true,
+    "memory_rank_hint": 4,
+    "rollback_rank_hint": 8,
+    "runtime_rank_hint": 8,
+    "toolspec_rank_hint": 8,
+    "trust_remote_code": true,
+    "uncertainty_rank_hint": 8,
+    "use_safetensors": null,
+    "validator_rank_hint": 8
+  },
+  "load_result": {
+    "has_base_config": true,
+    "has_base_model": true,
+    "hidden_size": 2048,
+    "notes": [
+      "class=AutoModelForCausalLM",
+      "quant=4bit"
+    ],
+    "source": "trainer_qwen36_loader"
+  },
+  "peft_named_modules": [
+    "citation_binding",
+    "citation_binding.adapter",
+    "citation_binding.adapter.base_layer",
+    "citation_binding.adapter.lora_A",
+    "citation_binding.adapter.lora_A.default",
+    "citation_binding.adapter.lora_B",
+    "citation_binding.adapter.lora_B.default",
+    "citation_binding.adapter.lora_dropout",
+    "citation_binding.adapter.lora_dropout.default",
+    "citation_binding.adapter.lora_embedding_A",
+    "citation_binding.adapter.lora_embedding_B",
+    "citation_binding.adapter.lora_magnitude_vector",
+    "claim_extractor",
+    "claim_extractor.adapter",
+    "claim_extractor.adapter.base_layer",
+    "claim_extractor.adapter.lora_A",
+    "claim_extractor.adapter.lora_A.default",
+    "claim_extractor.adapter.lora_B",
+    "claim_extractor.adapter.lora_B.default",
+    "claim_extractor.adapter.lora_dropout",
+    "claim_extractor.adapter.lora_dropout.default",
+    "claim_extractor.adapter.lora_embedding_A",
+    "claim_extractor.adapter.lora_embedding_B",
+    "claim_extractor.adapter.lora_magnitude_vector",
+    "dropout",
+    "episodic_memory",
+    "episodic_memory.adapter",
+    "evidence_binding",
+    "evidence_binding.adapter",
+    "evidence_binding.adapter.base_layer",
+    "evidence_binding.adapter.lora_A",
+    "evidence_binding.adapter.lora_A.default",
+    "evidence_binding.adapter.lora_B",
+    "evidence_binding.adapter.lora_B.default",
+    "evidence_binding.adapter.lora_dropout",
+    "evidence_binding.adapter.lora_dropout.default",
+    "evidence_binding.adapter.lora_embedding_A",
+    "evidence_binding.adapter.lora_embedding_B",
+    "evidence_binding.adapter.lora_magnitude_vector",
+    "execution_binding",
+    "execution_binding.adapter",
+    "execution_binding.adapter.base_layer",
+    "execution_binding.adapter.lora_A",
+    "execution_binding.adapter.lora_A.default",
+    "execution_binding.adapter.lora_B",
+    "execution_binding.adapter.lora_B.default",
+    "execution_binding.adapter.lora_dropout",
+    "execution_binding.adapter.lora_dropout.default",
+    "execution_binding.adapter.lora_embedding_A",
+    "execution_binding.adapter.lora_embedding_B",
+    "execution_binding.adapter.lora_magnitude_vector",
+    "failure_signal_bridge",
+    "failure_signal_bridge.rollback_bridge",
+    "identity_adapter",
+    "identity_adapter.bridge",
+    "identity_guard",
+    "identity_guard.adapter",
+    "input_norm",
+    "memory_boundary_guard",
+    "memory_boundary_guard.adapter",
+    "memory_boundary_guard.rollback_filter",
+    "patch_binding",
+    "patch_binding.adapter",
+    "patch_binding.adapter.base_layer",
+    "patch_binding.adapter.lora_A",
+    "patch_binding.adapter.lora_A.default",
+    "patch_binding.adapter.lora_B",
+    "patch_binding.adapter.lora_B.default",
+    "patch_binding.adapter.lora_dropout",
+    "patch_binding.adapter.lora_dropout.default",
+    "patch_binding.adapter.lora_embedding_A",
+    "patch_binding.adapter.lora_embedding_B",
+    "patch_binding.adapter.lora_magnitude_vector",
+    "permission_context_manager",
+    "permission_context_manager.adapter",
+    "progress_state_tracker",
+    "progress_state_tracker.adapter",
+    "progress_state_tracker.rollback_memory",
+    "proof_carrying_hints",
+    "proof_carrying_hints.bridge",
+    "proof_carrying_hints.bridge.base_layer",
+    "proof_carrying_hints.bridge.lora_A",
+    "proof_carrying_hints.bridge.lora_A.default",
+    "proof_carrying_hints.bridge.lora_B",
+    "proof_carrying_hints.bridge.lora_B.default",
+    "proof_carrying_hints.bridge.lora_dropout",
+    "proof_carrying_hints.bridge.lora_dropout.default",
+    "proof_carrying_hints.bridge.lora_embedding_A",
+    "proof_carrying_hints.bridge.lora_embedding_B",
+    "proof_carrying_hints.bridge.lora_magnitude_vector",
+    "provenance_binding",
+    "provenance_binding.adapter",
+    "provenance_binding.adapter.base_layer",
+    "provenance_binding.adapter.lora_A",
+    "provenance_binding.adapter.lora_A.default",
+    "provenance_binding.adapter.lora_B",
+    "provenance_binding.adapter.lora_B.default",
+    "provenance_binding.adapter.lora_dropout",
+    "provenance_binding.adapter.lora_dropout.default",
+    "provenance_binding.adapter.lora_embedding_A",
+    "provenance_binding.adapter.lora_embedding_B",
+    "provenance_binding.adapter.lora_magnitude_vector",
+    "public_identity_head",
+    "public_identity_head.proj",
+    "query_runtime_engine",
+    "query_runtime_engine.adapter",
+    "request_normalizer",
+    "request_normalizer.adapter",
+    "reverse_engineering_binding",
+    "reverse_engineering_binding.adapter",
+    "reverse_engineering_binding.adapter.base_layer",
+    "reverse_engineering_binding.adapter.lora_A",
+    "reverse_engineering_binding.adapter.lora_A.default",
+    "reverse_engineering_binding.adapter.lora_B",
+    "reverse_engineering_binding.adapter.lora_B.default",
+    "reverse_engineering_binding.adapter.lora_dropout",
+    "reverse_engineering_binding.adapter.lora_dropout.default",
+    "reverse_engineering_binding.adapter.lora_embedding_A",
+    "reverse_engineering_binding.adapter.lora_embedding_B",
+    "reverse_engineering_binding.adapter.lora_magnitude_vector",
+    "rollback_adapter",
+    "rollback_adapter.head",
+    "rollback_engine",
+    "rollback_engine.adapter",
+    "runtime_binding",
+    "runtime_binding.adapter",
+    "runtime_binding.adapter.base_layer",
+    "runtime_binding.adapter.lora_A",
+    "runtime_binding.adapter.lora_A.default",
+    "runtime_binding.adapter.lora_B",
+    "runtime_binding.adapter.lora_B.default",
+    "runtime_binding.adapter.lora_dropout",
+    "runtime_binding.adapter.lora_dropout.default",
+    "runtime_binding.adapter.lora_embedding_A",
+    "runtime_binding.adapter.lora_embedding_B",
+    "runtime_binding.adapter.lora_magnitude_vector",
+    "runtime_harness_adapter",
+    "runtime_harness_adapter.bridge",
+    "runtime_harness_uncertainty_bridge",
+    "runtime_harness_uncertainty_bridge.adapter",
+    "sandbox_rollback_bridge",
+    "sandbox_rollback_bridge.adapter",
+    "selfcheck_binding",
+    "selfcheck_binding.adapter",
+    "selfcheck_binding.adapter.base_layer",
+    "selfcheck_binding.adapter.lora_A",
+    "selfcheck_binding.adapter.lora_A.default",
+    "selfcheck_binding.adapter.lora_B",
+    "selfcheck_binding.adapter.lora_B.default",
+    "selfcheck_binding.adapter.lora_dropout",
+    "selfcheck_binding.adapter.lora_dropout.default",
+    "selfcheck_binding.adapter.lora_embedding_A",
+    "selfcheck_binding.adapter.lora_embedding_B",
+    "selfcheck_binding.adapter.lora_magnitude_vector",
+    "session_compactor",
+    "session_compactor.adapter",
+    "session_state_manager",
+    "session_state_manager.adapter",
+    "session_state_manager.rollback_state",
+    "tool_protocol_adapter",
+    "tool_protocol_adapter.bridge",
+    "tool_receipt_binding",
+    "tool_receipt_binding.adapter",
+    "tool_receipt_binding.adapter.base_layer",
+    "tool_receipt_binding.adapter.lora_A",
+    "tool_receipt_binding.adapter.lora_A.default",
+    "tool_receipt_binding.adapter.lora_B",
+    "tool_receipt_binding.adapter.lora_B.default",
+    "tool_receipt_binding.adapter.lora_dropout",
+    "tool_receipt_binding.adapter.lora_dropout.default",
+    "tool_receipt_binding.adapter.lora_embedding_A",
+    "tool_receipt_binding.adapter.lora_embedding_B",
+    "tool_receipt_binding.adapter.lora_magnitude_vector",
+    "toolspec_bridge",
+    "toolspec_bridge.adapter",
+    "toolspec_head",
+    "toolspec_head.param_schema_adapter",
+    "toolspec_head.postcondition_adapter",
+    "toolspec_head.precondition_adapter",
+    "toolspec_head.receipt_formatter",
+    "toolspec_head.trigger_gate",
+    "toolspec_head.validator_gate",
+    "uncertainty_head",
+    "uncertainty_head.calibration_mlp",
+    "uncertainty_head.proj",
+    "validator_feedback_bridge",
+    "validator_feedback_bridge.adapter",
+    "validator_feedback_loop",
+    "validator_feedback_loop.rollback_adapter",
+    "validator_receipt_bridge",
+    "validator_receipt_bridge.adapter",
+    "validator_receipt_bridge.adapter.base_layer",
+    "validator_receipt_bridge.adapter.lora_A",
+    "validator_receipt_bridge.adapter.lora_A.default",
+    "validator_receipt_bridge.adapter.lora_B",
+    "validator_receipt_bridge.adapter.lora_B.default",
+    "validator_receipt_bridge.adapter.lora_dropout",
+    "validator_receipt_bridge.adapter.lora_dropout.default",
+    "validator_receipt_bridge.adapter.lora_embedding_A",
+    "validator_receipt_bridge.adapter.lora_embedding_B",
+    "validator_receipt_bridge.adapter.lora_magnitude_vector",
+    "validator_uncertainty_bridge",
+    "validator_uncertainty_bridge.adapter",
+    "workspace_snapshot_manager",
+    "workspace_snapshot_manager.rollback_context",
+    "worktree_binding",
+    "worktree_binding.adapter",
+    "worktree_binding.adapter.base_layer",
+    "worktree_binding.adapter.lora_A",
+    "worktree_binding.adapter.lora_A.default",
+    "worktree_binding.adapter.lora_B",
+    "worktree_binding.adapter.lora_B.default",
+    "worktree_binding.adapter.lora_dropout",
+    "worktree_binding.adapter.lora_dropout.default",
+    "worktree_binding.adapter.lora_embedding_A",
+    "worktree_binding.adapter.lora_embedding_B",
+    "worktree_binding.adapter.lora_magnitude_vector",
+    "worktree_manager",
+    "worktree_manager.adapter"
+  ],
+  "trainable_parameter_report": {
+    "backbone_frozen": true,
+    "backbone_present": true,
+    "hidden_size": 2048,
+    "host_parameters": 197668869,
+    "host_trainable_parameters": 425984,
+    "total_parameters": 34153726597,
+    "trainable_parameters": 425984,
+    "version": "veriloop.coder_peft_host.v1"
+  },
+  "version": "veriloop.coder_peft_host.v1"
+}

evidence_adapter/tokenizer/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + 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" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|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 %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if (preserve_thinking is defined and preserve_thinking is true) or (loop.index0 > ns.last_query_index) %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- 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' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

evidence_adapter/tokenizer/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea2e66b594a0906e9a547c9e6ff9e5fb8a8198439c8cf7d6dc48f23529161223
+size 19989442

evidence_adapter/tokenizer/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": true,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}