PEGASUS EVAL / RESULTS EXPLORER

Observed benchmark evidence · read-only

← Back to matrix · All model evidence for this question

piq40_logic_symmetric × Phi-3 Medium   ✗ FAIL

BENCHMARK QUESTION

Question ID
piq40_logic_symmetric
Prompt
In each row, the middle number is the square of the difference between the two edge numbers. If a row is 5, ?, 5, what is missing?
Options
—
Tags
[
  "logic_grid",
  "hard"
]
Difficulty
—
Metadata
—
Response contract
{
  "instruction": "Return only the final answer. Do not include explanation.",
  "mode": "short_answer"
}
Question provenance
—

GOLD / BENCHMARK TRUTH

Raw expected value
0
Evaluation rule type
normalized_exact
Raw evaluation rule
{
  "expected": 0,
  "type": "normalized_exact"
}

CANDIDATE / MODEL OBSERVATION

Model display name
Phi-3 Medium
Model ID
phi3-medium-4k
Raw model response
1
Execution status
success
Structured output
—
Model configuration
{
  "adapter": "transformers_local",
  "base_url": null,
  "device": "cuda",
  "dtype": "auto",
  "extra_headers": {},
  "generation": {
    "do_sample": false
  },
  "id": "phi3-medium-4k",
  "local_files_only": true,
  "max_new_tokens": 32,
  "model": "[local-model-path-redacted]/phi3-medium-4k",
  "quantization_method": null,
  "secret_env": null,
  "supports_json_mode": false,
  "timeout": 60
}

EVALUATION RESULT

Result
FAIL
Stored evaluation status
failed
Score
0.0
gate_status
fail
conclusive
true
evaluator_name
pegasus.normalized_exact
evaluator_version
1
rule_type
normalized_exact
Reason
deterministic rule conclusively not satisfied
Evaluator error
—

GRADER EVIDENCE

evaluation.evidence
{
  "actual_normalized": "1",
  "expected_normalized": "0"
}

EXECUTION EVIDENCE

effective_candidate_prompt
In each row, the middle number is the square of the difference between the two edge numbers. If a row is 5, ?, 5, what is missing?

Return only the final answer. Do not include explanation or reasoning.

Return only the final answer. Do not include explanation.
benchmark_prompt
In each row, the middle number is the square of the difference between the two edge numbers. If a row is 5, ?, 5, what is missing?
latency_ms
84.19784800207708
usage
{
  "input_tokens": 66,
  "output_tokens": 3
}
provider_metadata
{
  "adapter": "transformers_local",
  "cuda_version": "12.8",
  "device": "cuda",
  "dtype": "auto",
  "generation_latency_ms": 83.91356399806682,
  "gpu_compute_capability": [
    12,
    0
  ],
  "gpu_name": "NVIDIA GeForce RTX 5090",
  "model": "[local-model-path-redacted]/phi3-medium-4k",
  "model_load_ms": 112327.89194400539,
  "peak_cuda_memory_allocated_bytes": 27988511232,
  "provider": "transformers_local",
  "torch_version": "2.9.0+cu128",
  "transformers_version": "4.51.3"
}
timestamp
2026-08-25T18:08:57.847663+00:00
response_id
f1b55990ba5a4517ade8ddda80bebf84
run_id
9a35308f3b214208b75b57f61731957f
Execution error
—
Run artifact directory
runs/pattern-iq/phi-3-medium
Raw stored response
{
  "benchmark_prompt": "In each row, the middle number is the square of the difference between the two edge numbers. If a row is 5, ?, 5, what is missing?",
  "effective_candidate_prompt": "In each row, the middle number is the square of the difference between the two edge numbers. If a row is 5, ?, 5, what is missing?\n\nReturn only the final answer. Do not include explanation or reasoning.\n\nReturn only the final answer. Do not include explanation.",
  "error": null,
  "execution_status": "success",
  "latency_ms": 84.19784800207708,
  "model_id": "phi3-medium-4k",
  "provider_metadata": {
    "adapter": "transformers_local",
    "cuda_version": "12.8",
    "device": "cuda",
    "dtype": "auto",
    "generation_latency_ms": 83.91356399806682,
    "gpu_compute_capability": [
      12,
      0
    ],
    "gpu_name": "NVIDIA GeForce RTX 5090",
    "model": "[local-model-path-redacted]/phi3-medium-4k",
    "model_load_ms": 112327.89194400539,
    "peak_cuda_memory_allocated_bytes": 27988511232,
    "provider": "transformers_local",
    "torch_version": "2.9.0+cu128",
    "transformers_version": "4.51.3"
  },
  "question_id": "piq40_logic_symmetric",
  "response_id": "f1b55990ba5a4517ade8ddda80bebf84",
  "run_id": "9a35308f3b214208b75b57f61731957f",
  "structured_output": null,
  "text": "1",
  "timestamp": "2026-08-25T18:08:57.847663+00:00",
  "usage": {
    "input_tokens": 66,
    "output_tokens": 3
  }
}
Raw stored evaluation
{
  "conclusive": true,
  "error": null,
  "evaluator_name": "pegasus.normalized_exact",
  "evaluator_version": "1",
  "evidence": {
    "actual_normalized": "1",
    "expected_normalized": "0"
  },
  "gate_status": "fail",
  "reason": "deterministic rule conclusively not satisfied",
  "response_id": "f1b55990ba5a4517ade8ddda80bebf84",
  "rule_type": "normalized_exact",
  "score": 0.0,
  "status": "failed"
}