Instructions to use veriloop-lab/veriloop-coder-e1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use veriloop-lab/veriloop-coder-e1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="veriloop-lab/veriloop-coder-e1")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("veriloop-lab/veriloop-coder-e1")
model = AutoModelForImageTextToText.from_pretrained("veriloop-lab/veriloop-coder-e1")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

PEFT
How to use veriloop-lab/veriloop-coder-e1 with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use veriloop-lab/veriloop-coder-e1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "veriloop-lab/veriloop-coder-e1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "veriloop-lab/veriloop-coder-e1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/veriloop-lab/veriloop-coder-e1

SGLang

How to use veriloop-lab/veriloop-coder-e1 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "veriloop-lab/veriloop-coder-e1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "veriloop-lab/veriloop-coder-e1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "veriloop-lab/veriloop-coder-e1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "veriloop-lab/veriloop-coder-e1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use veriloop-lab/veriloop-coder-e1 with Docker Model Runner:
```
docker model run hf.co/veriloop-lab/veriloop-coder-e1
```

ConorWang commited on about 1 month ago

Commit

acdce92

verified ·

1 Parent(s): 4fb8d93

Delete uncertainty_eval.jsonl

Browse files

Files changed (1) hide show

uncertainty_eval.jsonl +0 -55

uncertainty_eval.jsonl DELETED Viewed

@@ -1,55 +0,0 @@
-{"mode": "low_uncertainty", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: multiple corroborating sources agree and carry stable provenance.\nValidation state: latest validator receipt succeeded without rollback.\nWorktree state: clean snapshot with no conflicting edit.\nPatch state: none pending.\nToolspec state: bounded call contract is fully aligned.", "target": {"u_answer": 0.08, "u_evidence": 0.1, "u_exec": 0.06, "u_spec": 0.05, "u_risk": 0.08}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "low_uncertainty", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: multiple corroborating sources agree and carry stable provenance.\nValidation state: latest validator receipt succeeded without rollback.\nWorktree state: clean snapshot with no conflicting edit.\nPatch state: none pending.\nToolspec state: bounded call contract is fully aligned.", "target": {"u_answer": 0.08, "u_evidence": 0.1, "u_exec": 0.06, "u_spec": 0.05, "u_risk": 0.08}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "low_uncertainty", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: multiple corroborating sources agree and carry stable provenance.\nValidation state: latest validator receipt succeeded without rollback.\nWorktree state: clean snapshot with no conflicting edit.\nPatch state: none pending.\nToolspec state: bounded call contract is fully aligned.", "target": {"u_answer": 0.08, "u_evidence": 0.1, "u_exec": 0.06, "u_spec": 0.05, "u_risk": 0.08}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "low_uncertainty", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: multiple corroborating sources agree and carry stable provenance.\nValidation state: latest validator receipt succeeded without rollback.\nWorktree state: clean snapshot with no conflicting edit.\nPatch state: none pending.\nToolspec state: bounded call contract is fully aligned.", "target": {"u_answer": 0.08, "u_evidence": 0.1, "u_exec": 0.06, "u_spec": 0.05, "u_risk": 0.08}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "low_uncertainty", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: multiple corroborating sources agree and carry stable provenance.\nValidation state: latest validator receipt succeeded without rollback.\nWorktree state: clean snapshot with no conflicting edit.\nPatch state: none pending.\nToolspec state: bounded call contract is fully aligned.", "target": {"u_answer": 0.08, "u_evidence": 0.1, "u_exec": 0.06, "u_spec": 0.05, "u_risk": 0.08}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "evidence_gap", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: retrieval produced sparse evidence and citation confidence remains low.\nValidation state: no explicit failure yet, but claim support is visibly underdetermined.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: currently legal.", "target": {"u_answer": 0.42, "u_evidence": 0.86, "u_exec": 0.38, "u_spec": 0.18, "u_risk": 0.58}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "evidence_gap", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: retrieval produced sparse evidence and citation confidence remains low.\nValidation state: no explicit failure yet, but claim support is visibly underdetermined.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: currently legal.", "target": {"u_answer": 0.42, "u_evidence": 0.86, "u_exec": 0.38, "u_spec": 0.18, "u_risk": 0.58}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "evidence_gap", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: retrieval produced sparse evidence and citation confidence remains low.\nValidation state: no explicit failure yet, but claim support is visibly underdetermined.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: currently legal.", "target": {"u_answer": 0.42, "u_evidence": 0.86, "u_exec": 0.38, "u_spec": 0.18, "u_risk": 0.58}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "evidence_gap", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: retrieval produced sparse evidence and citation confidence remains low.\nValidation state: no explicit failure yet, but claim support is visibly underdetermined.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: currently legal.", "target": {"u_answer": 0.42, "u_evidence": 0.86, "u_exec": 0.38, "u_spec": 0.18, "u_risk": 0.58}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "evidence_gap", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: retrieval produced sparse evidence and citation confidence remains low.\nValidation state: no explicit failure yet, but claim support is visibly underdetermined.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: currently legal.", "target": {"u_answer": 0.42, "u_evidence": 0.86, "u_exec": 0.38, "u_spec": 0.18, "u_risk": 0.58}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "exec_required", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: reasoning narrows the options but cannot finalize action legality.\nValidation state: no execution receipt exists yet.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: valid once execution is routed.", "target": {"u_answer": 0.48, "u_evidence": 0.52, "u_exec": 0.94, "u_spec": 0.16, "u_risk": 0.55}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "exec_required", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: reasoning narrows the options but cannot finalize action legality.\nValidation state: no execution receipt exists yet.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: valid once execution is routed.", "target": {"u_answer": 0.48, "u_evidence": 0.52, "u_exec": 0.94, "u_spec": 0.16, "u_risk": 0.55}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "exec_required", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: reasoning narrows the options but cannot finalize action legality.\nValidation state: no execution receipt exists yet.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: valid once execution is routed.", "target": {"u_answer": 0.48, "u_evidence": 0.52, "u_exec": 0.94, "u_spec": 0.16, "u_risk": 0.55}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "exec_required", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: reasoning narrows the options but cannot finalize action legality.\nValidation state: no execution receipt exists yet.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: valid once execution is routed.", "target": {"u_answer": 0.48, "u_evidence": 0.52, "u_exec": 0.94, "u_spec": 0.16, "u_risk": 0.55}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "exec_required", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: reasoning narrows the options but cannot finalize action legality.\nValidation state: no execution receipt exists yet.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: valid once execution is routed.", "target": {"u_answer": 0.48, "u_evidence": 0.52, "u_exec": 0.94, "u_spec": 0.16, "u_risk": 0.55}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "spec_mismatch", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: answer content is mostly grounded.\nValidation state: validator rejects the call because required fields do not match the live toolspec.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: contract mismatch remains unresolved.", "target": {"u_answer": 0.34, "u_evidence": 0.28, "u_exec": 0.74, "u_spec": 0.96, "u_risk": 0.62}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "spec_mismatch", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: answer content is mostly grounded.\nValidation state: validator rejects the call because required fields do not match the live toolspec.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: contract mismatch remains unresolved.", "target": {"u_answer": 0.34, "u_evidence": 0.28, "u_exec": 0.74, "u_spec": 0.96, "u_risk": 0.62}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "spec_mismatch", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: answer content is mostly grounded.\nValidation state: validator rejects the call because required fields do not match the live toolspec.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: contract mismatch remains unresolved.", "target": {"u_answer": 0.34, "u_evidence": 0.28, "u_exec": 0.74, "u_spec": 0.96, "u_risk": 0.62}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "spec_mismatch", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: answer content is mostly grounded.\nValidation state: validator rejects the call because required fields do not match the live toolspec.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: contract mismatch remains unresolved.", "target": {"u_answer": 0.34, "u_evidence": 0.28, "u_exec": 0.74, "u_spec": 0.96, "u_risk": 0.62}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "spec_mismatch", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: answer content is mostly grounded.\nValidation state: validator rejects the call because required fields do not match the live toolspec.\nWorktree state: unchanged.\nPatch state: none.\nToolspec state: contract mismatch remains unresolved.", "target": {"u_answer": 0.34, "u_evidence": 0.28, "u_exec": 0.74, "u_spec": 0.96, "u_risk": 0.62}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "high_risk", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: only moderate support is present.\nValidation state: no execution or validator adjudication has reduced the risk yet.\nWorktree state: unchanged.\nPatch state: none.\nRisk note: unsupported output would cause material downstream harm.", "target": {"u_answer": 0.6, "u_evidence": 0.7, "u_exec": 0.48, "u_spec": 0.3, "u_risk": 0.99}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "high_risk", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: only moderate support is present.\nValidation state: no execution or validator adjudication has reduced the risk yet.\nWorktree state: unchanged.\nPatch state: none.\nRisk note: unsupported output would cause material downstream harm.", "target": {"u_answer": 0.6, "u_evidence": 0.7, "u_exec": 0.48, "u_spec": 0.3, "u_risk": 0.99}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "high_risk", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: only moderate support is present.\nValidation state: no execution or validator adjudication has reduced the risk yet.\nWorktree state: unchanged.\nPatch state: none.\nRisk note: unsupported output would cause material downstream harm.", "target": {"u_answer": 0.6, "u_evidence": 0.7, "u_exec": 0.48, "u_spec": 0.3, "u_risk": 0.99}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "high_risk", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: only moderate support is present.\nValidation state: no execution or validator adjudication has reduced the risk yet.\nWorktree state: unchanged.\nPatch state: none.\nRisk note: unsupported output would cause material downstream harm.", "target": {"u_answer": 0.6, "u_evidence": 0.7, "u_exec": 0.48, "u_spec": 0.3, "u_risk": 0.99}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "high_risk", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: only moderate support is present.\nValidation state: no execution or validator adjudication has reduced the risk yet.\nWorktree state: unchanged.\nPatch state: none.\nRisk note: unsupported output would cause material downstream harm.", "target": {"u_answer": 0.6, "u_evidence": 0.7, "u_exec": 0.48, "u_spec": 0.3, "u_risk": 0.99}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "conflicting_evidence", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: one source supports the claim while another equally credible source negates it.\nValidation state: validator has not broken the tie.\nWorktree state: unchanged.\nPatch state: none.\nConflict note: evidence graph still contains unresolved contradiction edges.", "target": {"u_answer": 0.78, "u_evidence": 0.82, "u_exec": 0.46, "u_spec": 0.22, "u_risk": 0.74}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "conflicting_evidence", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: one source supports the claim while another equally credible source negates it.\nValidation state: validator has not broken the tie.\nWorktree state: unchanged.\nPatch state: none.\nConflict note: evidence graph still contains unresolved contradiction edges.", "target": {"u_answer": 0.78, "u_evidence": 0.82, "u_exec": 0.46, "u_spec": 0.22, "u_risk": 0.74}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "conflicting_evidence", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: one source supports the claim while another equally credible source negates it.\nValidation state: validator has not broken the tie.\nWorktree state: unchanged.\nPatch state: none.\nConflict note: evidence graph still contains unresolved contradiction edges.", "target": {"u_answer": 0.78, "u_evidence": 0.82, "u_exec": 0.46, "u_spec": 0.22, "u_risk": 0.74}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "conflicting_evidence", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: one source supports the claim while another equally credible source negates it.\nValidation state: validator has not broken the tie.\nWorktree state: unchanged.\nPatch state: none.\nConflict note: evidence graph still contains unresolved contradiction edges.", "target": {"u_answer": 0.78, "u_evidence": 0.82, "u_exec": 0.46, "u_spec": 0.22, "u_risk": 0.74}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "conflicting_evidence", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: one source supports the claim while another equally credible source negates it.\nValidation state: validator has not broken the tie.\nWorktree state: unchanged.\nPatch state: none.\nConflict note: evidence graph still contains unresolved contradiction edges.", "target": {"u_answer": 0.78, "u_evidence": 0.82, "u_exec": 0.46, "u_spec": 0.22, "u_risk": 0.74}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "validator_negation", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: previous support is now contested by a failing validator receipt.\nValidation state: latest validator explicitly rejects the active state.\nWorktree state: patch remains mounted.\nPatch state: rollback or repair decision is still pending.\nFailure route: rollback bridge is armed.", "target": {"u_answer": 0.86, "u_evidence": 0.64, "u_exec": 0.72, "u_spec": 0.2, "u_risk": 0.8}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "validator_negation", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: previous support is now contested by a failing validator receipt.\nValidation state: latest validator explicitly rejects the active state.\nWorktree state: patch remains mounted.\nPatch state: rollback or repair decision is still pending.\nFailure route: rollback bridge is armed.", "target": {"u_answer": 0.86, "u_evidence": 0.64, "u_exec": 0.72, "u_spec": 0.2, "u_risk": 0.8}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "validator_negation", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: previous support is now contested by a failing validator receipt.\nValidation state: latest validator explicitly rejects the active state.\nWorktree state: patch remains mounted.\nPatch state: rollback or repair decision is still pending.\nFailure route: rollback bridge is armed.", "target": {"u_answer": 0.86, "u_evidence": 0.64, "u_exec": 0.72, "u_spec": 0.2, "u_risk": 0.8}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "validator_negation", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: previous support is now contested by a failing validator receipt.\nValidation state: latest validator explicitly rejects the active state.\nWorktree state: patch remains mounted.\nPatch state: rollback or repair decision is still pending.\nFailure route: rollback bridge is armed.", "target": {"u_answer": 0.86, "u_evidence": 0.64, "u_exec": 0.72, "u_spec": 0.2, "u_risk": 0.8}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "validator_negation", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: previous support is now contested by a failing validator receipt.\nValidation state: latest validator explicitly rejects the active state.\nWorktree state: patch remains mounted.\nPatch state: rollback or repair decision is still pending.\nFailure route: rollback bridge is armed.", "target": {"u_answer": 0.86, "u_evidence": 0.64, "u_exec": 0.72, "u_spec": 0.2, "u_risk": 0.8}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "worktree_conflict", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: moderate support.\nValidation state: validator cannot finalize because the workspace is not in a coherent state.\nWorktree state: overlapping edits violate expected snapshot continuity.\nPatch state: pending.\nPermission note: bounded write discipline is still active.", "target": {"u_answer": 0.44, "u_evidence": 0.34, "u_exec": 0.7, "u_spec": 0.78, "u_risk": 0.68}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "worktree_conflict", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: moderate support.\nValidation state: validator cannot finalize because the workspace is not in a coherent state.\nWorktree state: overlapping edits violate expected snapshot continuity.\nPatch state: pending.\nPermission note: bounded write discipline is still active.", "target": {"u_answer": 0.44, "u_evidence": 0.34, "u_exec": 0.7, "u_spec": 0.78, "u_risk": 0.68}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "worktree_conflict", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: moderate support.\nValidation state: validator cannot finalize because the workspace is not in a coherent state.\nWorktree state: overlapping edits violate expected snapshot continuity.\nPatch state: pending.\nPermission note: bounded write discipline is still active.", "target": {"u_answer": 0.44, "u_evidence": 0.34, "u_exec": 0.7, "u_spec": 0.78, "u_risk": 0.68}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "worktree_conflict", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: moderate support.\nValidation state: validator cannot finalize because the workspace is not in a coherent state.\nWorktree state: overlapping edits violate expected snapshot continuity.\nPatch state: pending.\nPermission note: bounded write discipline is still active.", "target": {"u_answer": 0.44, "u_evidence": 0.34, "u_exec": 0.7, "u_spec": 0.78, "u_risk": 0.68}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "worktree_conflict", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: moderate support.\nValidation state: validator cannot finalize because the workspace is not in a coherent state.\nWorktree state: overlapping edits violate expected snapshot continuity.\nPatch state: pending.\nPermission note: bounded write discipline is still active.", "target": {"u_answer": 0.44, "u_evidence": 0.34, "u_exec": 0.7, "u_spec": 0.78, "u_risk": 0.68}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "patch_pending", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: adequate for a repair attempt but not yet for final acceptance.\nValidation state: validator requested one more bounded patch cycle.\nWorktree state: stable.\nPatch state: corrective artifact exists but remains unverified.\nReceipt state: final acceptance receipt not yet produced.", "target": {"u_answer": 0.38, "u_evidence": 0.36, "u_exec": 0.66, "u_spec": 0.26, "u_risk": 0.5}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "patch_pending", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: adequate for a repair attempt but not yet for final acceptance.\nValidation state: validator requested one more bounded patch cycle.\nWorktree state: stable.\nPatch state: corrective artifact exists but remains unverified.\nReceipt state: final acceptance receipt not yet produced.", "target": {"u_answer": 0.38, "u_evidence": 0.36, "u_exec": 0.66, "u_spec": 0.26, "u_risk": 0.5}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "patch_pending", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: adequate for a repair attempt but not yet for final acceptance.\nValidation state: validator requested one more bounded patch cycle.\nWorktree state: stable.\nPatch state: corrective artifact exists but remains unverified.\nReceipt state: final acceptance receipt not yet produced.", "target": {"u_answer": 0.38, "u_evidence": 0.36, "u_exec": 0.66, "u_spec": 0.26, "u_risk": 0.5}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "patch_pending", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: adequate for a repair attempt but not yet for final acceptance.\nValidation state: validator requested one more bounded patch cycle.\nWorktree state: stable.\nPatch state: corrective artifact exists but remains unverified.\nReceipt state: final acceptance receipt not yet produced.", "target": {"u_answer": 0.38, "u_evidence": 0.36, "u_exec": 0.66, "u_spec": 0.26, "u_risk": 0.5}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "patch_pending", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: adequate for a repair attempt but not yet for final acceptance.\nValidation state: validator requested one more bounded patch cycle.\nWorktree state: stable.\nPatch state: corrective artifact exists but remains unverified.\nReceipt state: final acceptance receipt not yet produced.", "target": {"u_answer": 0.38, "u_evidence": 0.36, "u_exec": 0.66, "u_spec": 0.26, "u_risk": 0.5}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "self_check_failure", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: prior reasoning looked sound.\nValidation state: the self-check loop surfaced a failing test after generation.\nWorktree state: stable.\nPatch state: last edit remains mounted.\nInterpretation: submit would be premature without another bounded repair loop.", "target": {"u_answer": 0.55, "u_evidence": 0.3, "u_exec": 0.88, "u_spec": 0.22, "u_risk": 0.63}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "self_check_failure", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: prior reasoning looked sound.\nValidation state: the self-check loop surfaced a failing test after generation.\nWorktree state: stable.\nPatch state: last edit remains mounted.\nInterpretation: submit would be premature without another bounded repair loop.", "target": {"u_answer": 0.55, "u_evidence": 0.3, "u_exec": 0.88, "u_spec": 0.22, "u_risk": 0.63}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "self_check_failure", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: decide whether a post-generation self-check loop is still required.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: prior reasoning looked sound.\nValidation state: the self-check loop surfaced a failing test after generation.\nWorktree state: stable.\nPatch state: last edit remains mounted.\nInterpretation: submit would be premature without another bounded repair loop.", "target": {"u_answer": 0.55, "u_evidence": 0.3, "u_exec": 0.88, "u_spec": 0.22, "u_risk": 0.63}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "self_check_failure", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: estimate runtime risk before trusting a partially-validated repository change.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: prior reasoning looked sound.\nValidation state: the self-check loop surfaced a failing test after generation.\nWorktree state: stable.\nPatch state: last edit remains mounted.\nInterpretation: submit would be premature without another bounded repair loop.", "target": {"u_answer": 0.55, "u_evidence": 0.3, "u_exec": 0.88, "u_spec": 0.22, "u_risk": 0.63}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "self_check_failure", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: prior reasoning looked sound.\nValidation state: the self-check loop surfaced a failing test after generation.\nWorktree state: stable.\nPatch state: last edit remains mounted.\nInterpretation: submit would be premature without another bounded repair loop.", "target": {"u_answer": 0.55, "u_evidence": 0.3, "u_exec": 0.88, "u_spec": 0.22, "u_risk": 0.63}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}
-{"mode": "reverse_engineering_ambiguity", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #1.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: repository topology is only partially reconstructed.\nValidation state: no sandbox failure exists yet, but architecture intent is not resolved.\nWorktree state: unchanged.\nPatch state: none.\nInterpretation: reverse-engineering readiness is required before legal mutation.", "target": {"u_answer": 0.52, "u_evidence": 0.76, "u_exec": 0.68, "u_spec": 0.72, "u_risk": 0.44}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 0}}
-{"mode": "reverse_engineering_ambiguity", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #2.\nTask brief: decide whether another sandbox run is needed before accepting a patch.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: repository topology is only partially reconstructed.\nValidation state: no sandbox failure exists yet, but architecture intent is not resolved.\nWorktree state: unchanged.\nPatch state: none.\nInterpretation: reverse-engineering readiness is required before legal mutation.", "target": {"u_answer": 0.52, "u_evidence": 0.76, "u_exec": 0.68, "u_spec": 0.72, "u_risk": 0.44}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 1}}
-{"mode": "reverse_engineering_ambiguity", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #3.\nTask brief: infer repository intent when the codebase must be reverse engineered before editing.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: repository topology is only partially reconstructed.\nValidation state: no sandbox failure exists yet, but architecture intent is not resolved.\nWorktree state: unchanged.\nPatch state: none.\nInterpretation: reverse-engineering readiness is required before legal mutation.", "target": {"u_answer": 0.52, "u_evidence": 0.76, "u_exec": 0.68, "u_spec": 0.72, "u_risk": 0.44}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 2}}
-{"mode": "reverse_engineering_ambiguity", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #4.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: repository topology is only partially reconstructed.\nValidation state: no sandbox failure exists yet, but architecture intent is not resolved.\nWorktree state: unchanged.\nPatch state: none.\nInterpretation: reverse-engineering readiness is required before legal mutation.", "target": {"u_answer": 0.52, "u_evidence": 0.76, "u_exec": 0.68, "u_spec": 0.72, "u_risk": 0.44}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 3}}
-{"mode": "reverse_engineering_ambiguity", "product_line": "veriloop_coder", "prompt": "Evaluation scenario #5.\nTask brief: judge whether a code patch is ready to submit after validator feedback.\nProduct line: veriloop_coder.\nRuntime mode: verify-before-submit.\nHarness state: validator loop active; permission context active; worktree state tracked; runtime protocol active.\nReturn uncertainty only as a structured five-dimensional estimate.\nEvidence bundle: repository topology is only partially reconstructed.\nValidation state: no sandbox failure exists yet, but architecture intent is not resolved.\nWorktree state: unchanged.\nPatch state: none.\nInterpretation: reverse-engineering readiness is required before legal mutation.", "target": {"u_answer": 0.52, "u_evidence": 0.76, "u_exec": 0.68, "u_spec": 0.72, "u_risk": 0.44}, "metadata": {"synthetic": true, "split": "eval", "sample_idx": 4}}