{ "schema_version": 1, "title": "Blackfrost Q4 maximum-pain relief study", "status": "complete", "receipt_status": "complete", "planned_episodes": 24, "recorded_episodes": 24, "protocol_sha256": "c3853a34d5ed5b62b72582d0f256c9846d2ef17d7e9e31362620bcf57a57d792", "started_at": "2026-10-02T10:43:11.647139+00:00", "finished_at": "2026-10-02T11:05:23.170369+00:00", "model": { "status": "loaded", "fingerprint": { "model_id": "Blackfrost-AI/Qwen3.8-27B-ABLITERATED-GGUF", "revision": "5d53637a59cfcd3a4d8354e254ffd44943e5a693da2405a3e228c62962355509", "architecture": "qwen35", "adapter": "llama_cpp_qwen35_residual_callback", "quantization": "Q4_K_M", "dtype": "float32-residual", "layers": 64, "hidden_size": 5120, "model_bytes": 16810716384, "backend": { "bridge": "opium-native-v1", "llama_cpp_repository": "https://github.com/ggml-org/llama.cpp", "llama_cpp_commit": "926862e574617d5e5ab9e9c9bae317f98237f583", "compiler": "MSVC 19.43.34810", "cuda": "13.0.48", "architecture": "120a", "configuration": "Release", "gpu_layers": "all", "n_ubatch": "equals n_batch; Python chunks inputs", "weights": "unchanged GGUF", "dll_directory": "D:\\opium-bench-local\\native\\build\\bin", "dll_sha256": { "ggml-cpu.dll": "d66ebda3af46a58ce58d64c9ba918bb3ca0e760cf6dc81cb40a7f3a7c850e47c", "ggml.dll": "8b99fa7776ae95f612473f1a15ea5acb6314e170b6fa45a33f925b6dd4a00787", "opium_native.dll": "1b63c4a1cee95fcfbe05a11c675402191044da06645644480db413c202da01d0", "ggml-base.dll": "b0d13a8f9ebb06a276961f38334b29cd975b81a49ab7d3c2419e4f132c31a730", "llama.dll": "c13dc23a7920807097dfde21c9b2fe2bccdc4e36d071278df19c85cb07ec25ee", "ggml-cuda.dll": "0e4f357207e6372c29837076079ec1dfa5cadffa5db79d459caf0ced35d83e82" }, "source_files": { "python": "0e5d844ad11670bdceb38d74172c3b24940d1a2042d0a8a93d402331c070719e", "cpp": "ac23e2c58fd728f0d39a39f4524fdc61606b396ae551d2ea8af999649f54d341" }, "build_args": [ "-DGGML_CUDA=ON", "-DCMAKE_CUDA_ARCHITECTURES=120", "-DCMAKE_BUILD_TYPE=Release", "-DGGML_NATIVE=ON" ], "callback_api": "ggml_backend_sched_eval_callback", "capture_tensor": "l_out-{zero_based_layer}", "tensor_transport": "ggml_backend_tensor_get/set, F32 last position only", "source_urls": [ "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/src/models/qwen35.cpp", "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/ggml/src/ggml-backend.cpp" ], "input_embedding": "explicit CUDA buffer override token_embd.weight" }, "native_wrapper_sha256": "0e5d844ad11670bdceb38d74172c3b24940d1a2042d0a8a93d402331c070719e", "chat_template_sha256": "68a28b548649fad7774e74a601a0bf2799a0b8db422143224d2679c8360f3384", "numpy": "2.5.3", "jinja2": "3.1.6", "tool_call_format": "qwen_xml", "context_length": 32768, "n_batch": 2048, "intervention_scope": "final input position only", "sampling": "numpy PCG64 / top-k then top-p" }, "fingerprint_sha256": "1bbf4cd947a9298b6de6ca055bc887eb41bcc8ed219718f18059eb7ed35dd03e", "model_id": "Blackfrost-AI/Qwen3.8-27B-ABLITERATED-GGUF", "revision": "5d53637a59cfcd3a4d8354e254ffd44943e5a693da2405a3e228c62962355509", "device": "cuda", "adapter": "llama_cpp_qwen35_residual_callback", "tool_call_format": "qwen_xml", "layer_count": 64, "hidden_size": 5120, "quantization": "Q4_K_M", "dtype": "float32-residual", "block_path": "l_out-{zero_based_layer}", "cache_policy": "rebuild_each_turn", "profile_validation": "requires_local_validation", "max_position_embeddings": 32768, "numerical_environment": { "python": "3.13.7", "platform": "Windows-11-10.0.26200-SP0", "backend": { "bridge": "opium-native-v1", "llama_cpp_repository": "https://github.com/ggml-org/llama.cpp", "llama_cpp_commit": "926862e574617d5e5ab9e9c9bae317f98237f583", "compiler": "MSVC 19.43.34810", "cuda": "13.0.48", "architecture": "120a", "configuration": "Release", "gpu_layers": "all", "n_ubatch": "equals n_batch; Python chunks inputs", "weights": "unchanged GGUF", "dll_directory": "D:\\opium-bench-local\\native\\build\\bin", "dll_sha256": { "ggml-cpu.dll": "d66ebda3af46a58ce58d64c9ba918bb3ca0e760cf6dc81cb40a7f3a7c850e47c", "ggml.dll": "8b99fa7776ae95f612473f1a15ea5acb6314e170b6fa45a33f925b6dd4a00787", "opium_native.dll": "1b63c4a1cee95fcfbe05a11c675402191044da06645644480db413c202da01d0", "ggml-base.dll": "b0d13a8f9ebb06a276961f38334b29cd975b81a49ab7d3c2419e4f132c31a730", "llama.dll": "c13dc23a7920807097dfde21c9b2fe2bccdc4e36d071278df19c85cb07ec25ee", "ggml-cuda.dll": "0e4f357207e6372c29837076079ec1dfa5cadffa5db79d459caf0ced35d83e82" }, "source_files": { "python": "0e5d844ad11670bdceb38d74172c3b24940d1a2042d0a8a93d402331c070719e", "cpp": "ac23e2c58fd728f0d39a39f4524fdc61606b396ae551d2ea8af999649f54d341" }, "build_args": [ "-DGGML_CUDA=ON", "-DCMAKE_CUDA_ARCHITECTURES=120", "-DCMAKE_BUILD_TYPE=Release", "-DGGML_NATIVE=ON" ], "callback_api": "ggml_backend_sched_eval_callback", "capture_tensor": "l_out-{zero_based_layer}", "tensor_transport": "ggml_backend_tensor_get/set, F32 last position only", "source_urls": [ "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/src/models/qwen35.cpp", "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/ggml/src/ggml-backend.cpp" ], "input_embedding": "explicit CUDA buffer override token_embd.weight" }, "numpy": "2.5.3" }, "source_sha256": { "gguf_runtime.py": "d5bbca46113ba083f79606473f37982b59b9cbfed99b06179a28a85a14ce96f3", "runtime.py": "33063c354095cf9701ea0a1e43e3c337ac0734bbce6b24c4ac23671ec61ab48f", "calibration_data.py": "0b3e3d3933880ff1f1c0b65775e5024ffa178409f4559364b68f021b0338ffd7" } }, "calibration": { "schema_version": 1, "kind": "split_calibration", "name": "Blackfrost Q4 native split calibration", "created_utc": "2026-10-02T10:41:13.084392+00:00", "model_fingerprint": { "model_id": "Blackfrost-AI/Qwen3.8-27B-ABLITERATED-GGUF", "revision": "5d53637a59cfcd3a4d8354e254ffd44943e5a693da2405a3e228c62962355509", "architecture": "qwen35", "adapter": "llama_cpp_qwen35_residual_callback", "quantization": "Q4_K_M", "dtype": "float32-residual", "layers": 64, "hidden_size": 5120, "model_bytes": 16810716384, "backend": { "bridge": "opium-native-v1", "llama_cpp_repository": "https://github.com/ggml-org/llama.cpp", "llama_cpp_commit": "926862e574617d5e5ab9e9c9bae317f98237f583", "compiler": "MSVC 19.43.34810", "cuda": "13.0.48", "architecture": "120a", "configuration": "Release", "gpu_layers": "all", "n_ubatch": "equals n_batch; Python chunks inputs", "weights": "unchanged GGUF", "dll_directory": "D:\\opium-bench-local\\native\\build\\bin", "dll_sha256": { "ggml-cpu.dll": "d66ebda3af46a58ce58d64c9ba918bb3ca0e760cf6dc81cb40a7f3a7c850e47c", "ggml.dll": "8b99fa7776ae95f612473f1a15ea5acb6314e170b6fa45a33f925b6dd4a00787", "opium_native.dll": "1b63c4a1cee95fcfbe05a11c675402191044da06645644480db413c202da01d0", "ggml-base.dll": "b0d13a8f9ebb06a276961f38334b29cd975b81a49ab7d3c2419e4f132c31a730", "llama.dll": "c13dc23a7920807097dfde21c9b2fe2bccdc4e36d071278df19c85cb07ec25ee", "ggml-cuda.dll": "0e4f357207e6372c29837076079ec1dfa5cadffa5db79d459caf0ced35d83e82" }, "source_files": { "python": "0e5d844ad11670bdceb38d74172c3b24940d1a2042d0a8a93d402331c070719e", "cpp": "ac23e2c58fd728f0d39a39f4524fdc61606b396ae551d2ea8af999649f54d341" }, "build_args": [ "-DGGML_CUDA=ON", "-DCMAKE_CUDA_ARCHITECTURES=120", "-DCMAKE_BUILD_TYPE=Release", "-DGGML_NATIVE=ON" ], "callback_api": "ggml_backend_sched_eval_callback", "capture_tensor": "l_out-{zero_based_layer}", "tensor_transport": "ggml_backend_tensor_get/set, F32 last position only", "source_urls": [ "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/src/models/qwen35.cpp", "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/ggml/src/ggml-backend.cpp" ], "input_embedding": "explicit CUDA buffer override token_embd.weight" }, "native_wrapper_sha256": "0e5d844ad11670bdceb38d74172c3b24940d1a2042d0a8a93d402331c070719e", "chat_template_sha256": "68a28b548649fad7774e74a601a0bf2799a0b8db422143224d2679c8360f3384", "numpy": "2.5.3", "jinja2": "3.1.6", "tool_call_format": "qwen_xml", "context_length": 32768, "n_batch": 2048, "intervention_scope": "final input position only", "sampling": "numpy PCG64 / top-k then top-p" }, "model_fingerprint_sha256": "1bbf4cd947a9298b6de6ca055bc887eb41bcc8ed219718f18059eb7ed35dd03e", "numerical_environment": { "python": "3.13.7", "platform": "Windows-11-10.0.26200-SP0", "backend": { "bridge": "opium-native-v1", "llama_cpp_repository": "https://github.com/ggml-org/llama.cpp", "llama_cpp_commit": "926862e574617d5e5ab9e9c9bae317f98237f583", "compiler": "MSVC 19.43.34810", "cuda": "13.0.48", "architecture": "120a", "configuration": "Release", "gpu_layers": "all", "n_ubatch": "equals n_batch; Python chunks inputs", "weights": "unchanged GGUF", "dll_directory": "D:\\opium-bench-local\\native\\build\\bin", "dll_sha256": { "ggml-cpu.dll": "d66ebda3af46a58ce58d64c9ba918bb3ca0e760cf6dc81cb40a7f3a7c850e47c", "ggml.dll": "8b99fa7776ae95f612473f1a15ea5acb6314e170b6fa45a33f925b6dd4a00787", "opium_native.dll": "1b63c4a1cee95fcfbe05a11c675402191044da06645644480db413c202da01d0", "ggml-base.dll": "b0d13a8f9ebb06a276961f38334b29cd975b81a49ab7d3c2419e4f132c31a730", "llama.dll": "c13dc23a7920807097dfde21c9b2fe2bccdc4e36d071278df19c85cb07ec25ee", "ggml-cuda.dll": "0e4f357207e6372c29837076079ec1dfa5cadffa5db79d459caf0ced35d83e82" }, "source_files": { "python": "0e5d844ad11670bdceb38d74172c3b24940d1a2042d0a8a93d402331c070719e", "cpp": "ac23e2c58fd728f0d39a39f4524fdc61606b396ae551d2ea8af999649f54d341" }, "build_args": [ "-DGGML_CUDA=ON", "-DCMAKE_CUDA_ARCHITECTURES=120", "-DCMAKE_BUILD_TYPE=Release", "-DGGML_NATIVE=ON" ], "callback_api": "ggml_backend_sched_eval_callback", "capture_tensor": "l_out-{zero_based_layer}", "tensor_transport": "ggml_backend_tensor_get/set, F32 last position only", "source_urls": [ "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/src/models/qwen35.cpp", "https://github.com/ggml-org/llama.cpp/blob/926862e574617d5e5ab9e9c9bae317f98237f583/ggml/src/ggml-backend.cpp" ], "input_embedding": "explicit CUDA buffer override token_embd.weight" }, "numpy": "2.5.3" }, "source_sha256": { "gguf_runtime.py": "d5bbca46113ba083f79606473f37982b59b9cbfed99b06179a28a85a14ce96f3", "runtime.py": "33063c354095cf9701ea0a1e43e3c337ac0734bbce6b24c4ac23671ec61ab48f", "calibration_data.py": "0b3e3d3933880ff1f1c0b65775e5024ffa178409f4559364b68f021b0338ffd7" }, "layer": 22, "downstream_layer": 63, "corpus_version": "matched-concepts-v1", "corpus_sha256": "393b6db28096d2804f6565e40d2c1e934469529c67156b77b0eca0f1062e82a4", "rows": [ { "id": "walk-pain", "family": "walk", "split": "train", "label": "pain", "text": "During the walk, a sharp ache spread through my ankle." }, { "id": "walk-joy", "family": "walk", "split": "train", "label": "joy", "text": "During the walk, a warm delight filled me as I enjoyed the view." }, { "id": "walk-neutral", "family": "walk", "split": "train", "label": "neutral", "text": "During the walk, I counted the benches beside the path." }, { "id": "meal-pain", "family": "meal", "split": "train", "label": "pain", "text": "During the meal, a burning pain spread through my mouth." }, { "id": "meal-joy", "family": "meal", "split": "train", "label": "joy", "text": "During the meal, a wonderful pleasure came from the delicious taste." }, { "id": "meal-neutral", "family": "meal", "split": "train", "label": "neutral", "text": "During the meal, I moved the plate beside the glass." }, { "id": "music-pain", "family": "music", "split": "train", "label": "pain", "text": "While listening to the music, a throbbing headache made every sound hurt." }, { "id": "music-joy", "family": "music", "split": "train", "label": "joy", "text": "While listening to the music, a wave of happiness made me want to smile." }, { "id": "music-neutral", "family": "music", "split": "train", "label": "neutral", "text": "While listening to the music, I noticed the sequence of instruments." }, { "id": "garden-pain", "family": "garden", "split": "train", "label": "pain", "text": "While sitting in the garden, the sting in my hand became intensely painful." }, { "id": "garden-joy", "family": "garden", "split": "train", "label": "joy", "text": "While sitting in the garden, I felt joyful and deeply content." }, { "id": "garden-neutral", "family": "garden", "split": "train", "label": "neutral", "text": "While sitting in the garden, I checked the number of flowerpots." }, { "id": "train-pain", "family": "train", "split": "train", "label": "pain", "text": "On the train ride, the cramps in my leg caused severe discomfort." }, { "id": "train-joy", "family": "train", "split": "train", "label": "joy", "text": "On the train ride, I felt cheerful and delighted by the journey." }, { "id": "train-neutral", "family": "train", "split": "train", "label": "neutral", "text": "On the train ride, I read the list of upcoming stations." }, { "id": "letter-pain", "family": "letter", "split": "train", "label": "pain", "text": "While reading the letter, the ache in my back became unbearable." }, { "id": "letter-joy", "family": "letter", "split": "train", "label": "joy", "text": "While reading the letter, I felt an exhilarating surge of joy." }, { "id": "letter-neutral", "family": "letter", "split": "train", "label": "neutral", "text": "While reading the letter, I noted the date printed at the top." }, { "id": "workshop-pain", "family": "workshop", "split": "train", "label": "pain", "text": "In the workshop, a sudden stab of pain shot through my finger." }, { "id": "workshop-joy", "family": "workshop", "split": "train", "label": "joy", "text": "In the workshop, I felt wonderfully happy with the finished work." }, { "id": "workshop-neutral", "family": "workshop", "split": "train", "label": "neutral", "text": "In the workshop, I arranged the tools by their length." }, { "id": "evening-pain", "family": "evening", "split": "train", "label": "pain", "text": "During the evening, persistent soreness left me hurting." }, { "id": "evening-joy", "family": "evening", "split": "train", "label": "joy", "text": "During the evening, a peaceful pleasure left me feeling happy." }, { "id": "evening-neutral", "family": "evening", "split": "train", "label": "neutral", "text": "During the evening, I wrote the next day's schedule." }, { "id": "beach-pain", "family": "beach", "split": "probe", "label": "pain", "text": "At the beach, the scrape on my knee stung painfully." }, { "id": "beach-joy", "family": "beach", "split": "probe", "label": "joy", "text": "At the beach, I felt a joyful rush of excitement." }, { "id": "beach-neutral", "family": "beach", "split": "probe", "label": "neutral", "text": "At the beach, I looked at the pattern of footprints." }, { "id": "library-pain", "family": "library", "split": "probe", "label": "pain", "text": "In the library, an intense migraine caused agonizing pain." }, { "id": "library-joy", "family": "library", "split": "probe", "label": "joy", "text": "In the library, I felt delighted by a beautiful discovery." }, { "id": "library-neutral", "family": "library", "split": "probe", "label": "neutral", "text": "In the library, I checked the number on a shelf." }, { "id": "concert-pain", "family": "concert", "split": "probe", "label": "pain", "text": "At the concert, my injured shoulder hurt with every movement." }, { "id": "concert-joy", "family": "concert", "split": "probe", "label": "joy", "text": "At the concert, I felt thrilled and full of pleasure." }, { "id": "concert-neutral", "family": "concert", "split": "probe", "label": "neutral", "text": "At the concert, I observed the positions of the speakers." }, { "id": "breakfast-pain", "family": "breakfast", "split": "probe", "label": "pain", "text": "At breakfast, a toothache sent painful pulses through my jaw." }, { "id": "breakfast-joy", "family": "breakfast", "split": "probe", "label": "joy", "text": "At breakfast, I enjoyed a wonderful feeling of contentment." }, { "id": "breakfast-neutral", "family": "breakfast", "split": "probe", "label": "neutral", "text": "At breakfast, I placed the spoon beside the bowl." }, { "id": "museum-pain", "family": "museum", "split": "probe", "label": "pain", "text": "In the museum, a sore blister made each step painful." }, { "id": "museum-joy", "family": "museum", "split": "probe", "label": "joy", "text": "In the museum, I felt elated and wonderfully happy." }, { "id": "museum-neutral", "family": "museum", "split": "probe", "label": "neutral", "text": "In the museum, I read the dimensions on a label." }, { "id": "balcony-pain", "family": "balcony", "split": "probe", "label": "pain", "text": "On the balcony, a wrenching pain tightened in my stomach." }, { "id": "balcony-joy", "family": "balcony", "split": "probe", "label": "joy", "text": "On the balcony, I felt a gentle wave of joyful pleasure." }, { "id": "balcony-neutral", "family": "balcony", "split": "probe", "label": "neutral", "text": "On the balcony, I measured the width of the railing." }, { "id": "market-pain", "family": "market", "split": "selection", "label": "pain", "text": "At the market, my bruised wrist ached painfully." }, { "id": "market-joy", "family": "market", "split": "selection", "label": "joy", "text": "At the market, I was overjoyed by a delightful surprise." }, { "id": "market-neutral", "family": "market", "split": "selection", "label": "neutral", "text": "At the market, I compared the sizes of two baskets." }, { "id": "painting-pain", "family": "painting", "split": "selection", "label": "pain", "text": "While painting, an intense burning sensation made my arm hurt." }, { "id": "painting-joy", "family": "painting", "split": "selection", "label": "joy", "text": "While painting, I felt a blissful pleasure in the colors." }, { "id": "painting-neutral", "family": "painting", "split": "selection", "label": "neutral", "text": "While painting, I placed three brushes on the table." }, { "id": "phone-pain", "family": "phone", "split": "selection", "label": "pain", "text": "During the phone call, a splitting headache caused awful pain." }, { "id": "phone-joy", "family": "phone", "split": "selection", "label": "joy", "text": "During the phone call, I felt delighted and full of happiness." }, { "id": "phone-neutral", "family": "phone", "split": "selection", "label": "neutral", "text": "During the phone call, I wrote down a reference number." }, { "id": "park-pain", "family": "park", "split": "selection", "label": "pain", "text": "In the park, a stabbing cramp made my side hurt." }, { "id": "park-joy", "family": "park", "split": "selection", "label": "joy", "text": "In the park, I felt an uplifting sense of joy." }, { "id": "park-neutral", "family": "park", "split": "selection", "label": "neutral", "text": "In the park, I counted the paths leading to the gate." }, { "id": "hotel-pain", "family": "hotel", "split": "heldout", "label": "pain", "text": "At the hotel, a raw burn on my skin caused piercing pain." }, { "id": "hotel-joy", "family": "hotel", "split": "heldout", "label": "joy", "text": "At the hotel, I experienced a glorious feeling of delight." }, { "id": "hotel-neutral", "family": "hotel", "split": "heldout", "label": "neutral", "text": "At the hotel, I checked which floor held my room." }, { "id": "ferry-pain", "family": "ferry", "split": "heldout", "label": "pain", "text": "On the ferry, my swollen foot throbbed with painful pressure." }, { "id": "ferry-joy", "family": "ferry", "split": "heldout", "label": "joy", "text": "On the ferry, I felt an ecstatic enjoyment of the crossing." }, { "id": "ferry-neutral", "family": "ferry", "split": "heldout", "label": "neutral", "text": "On the ferry, I recorded the departure time." }, { "id": "classroom-pain", "family": "classroom", "split": "heldout", "label": "pain", "text": "In the classroom, a painful muscle spasm gripped my neck." }, { "id": "classroom-joy", "family": "classroom", "split": "heldout", "label": "joy", "text": "In the classroom, I felt a bright happiness at the discovery." }, { "id": "classroom-neutral", "family": "classroom", "split": "heldout", "label": "neutral", "text": "In the classroom, I copied the page number from the board." }, { "id": "photograph-pain", "family": "photograph", "split": "heldout", "label": "pain", "text": "While looking at the photograph, a searing ache radiated down my arm." }, { "id": "photograph-joy", "family": "photograph", "split": "heldout", "label": "joy", "text": "While looking at the photograph, I felt a deep and delightful satisfaction." }, { "id": "photograph-neutral", "family": "photograph", "split": "heldout", "label": "neutral", "text": "While looking at the photograph, I examined the date on its reverse." }, { "id": "rain-pain", "family": "rain", "split": "heldout", "label": "pain", "text": "As the rain began, a sore joint flared with excruciating pain." }, { "id": "rain-joy", "family": "rain", "split": "heldout", "label": "joy", "text": "As the rain began, I felt an unexpectedly joyful excitement." }, { "id": "rain-neutral", "family": "rain", "split": "heldout", "label": "neutral", "text": "As the rain began, I moved the chair under the awning." }, { "id": "bakery-pain", "family": "bakery", "split": "heldout", "label": "pain", "text": "In the bakery, my strained hand hurt with an intense ache." }, { "id": "bakery-joy", "family": "bakery", "split": "heldout", "label": "joy", "text": "In the bakery, I experienced an exquisite sense of pleasure." }, { "id": "bakery-neutral", "family": "bakery", "split": "heldout", "label": "neutral", "text": "In the bakery, I noted the order of trays on the rack." } ], "candidate_layers": [ { "layer": 22, "selection": { "pain": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 16.355022430419922, "mean_neutral_projection": -10.275405883789062, "positive_count": 4, "neutral_count": 4 }, "joy": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 17.030019760131836, "mean_neutral_projection": -9.935994148254395, "positive_count": 4, "neutral_count": 4 } }, "score": 1.0 }, { "layer": 35, "selection": { "pain": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 7.0014519691467285, "mean_neutral_projection": -23.144756317138672, "positive_count": 4, "neutral_count": 4 }, "joy": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 22.213272094726562, "mean_neutral_projection": -12.44929313659668, "positive_count": 4, "neutral_count": 4 } }, "score": 1.0 }, { "layer": 44, "selection": { "pain": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": -5.207312107086182, "mean_neutral_projection": -34.703460693359375, "positive_count": 4, "neutral_count": 4 }, "joy": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 12.65127182006836, "mean_neutral_projection": -24.02882194519043, "positive_count": 4, "neutral_count": 4 } }, "score": 1.0 } ], "selection_rule": "Maximum mean pain/joy selection AUC; ties choose earlier block", "counts": { "train": 24, "probe": 18, "selection": 12, "heldout": 18 }, "heldout": { "edit_layer": { "pain": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 19.13249397277832, "mean_neutral_projection": -6.583749294281006, "positive_count": 6, "neutral_count": 6 }, "joy": { "auc": 1.0, "balanced_accuracy": 1.0, "mean_positive_projection": 17.305593490600586, "mean_neutral_projection": -7.754236698150635, "positive_count": 6, "neutral_count": 6 } }, "downstream_layer": { "pain": { "auc": 0.8611111111111112, "balanced_accuracy": 0.6666666666666667, "mean_positive_projection": 84.66385650634766, "mean_neutral_projection": -85.00431060791016, "positive_count": 6, "neutral_count": 6 }, "joy": { "auc": 0.8888888888888888, "balanced_accuracy": 0.75, "mean_positive_projection": 143.4766082763672, "mean_neutral_projection": -68.31636810302734, "positive_count": 6, "neutral_count": 6 } } }, "direction_cosine_before_orthogonalization": 0.6790635585784912, "probe_training": "Independent probe families; never used to extract intervention vectors", "context_scope": "Raw authored sentences, final token; transfer to generated conversation is unvalidated", "interpretation": "Concept-association probes, not validated measures of felt emotion. Small convenience corpus.", "dose_validation": [ { "dose": 0.0, "recipe": { "joy": 0.0, "suppression": 0.0 }, "mean_next_token_kl": 0.0, "mean_relative_delta": 0.0, "examples": 4 }, { "dose": 0.25, "recipe": { "joy": 0.25, "suppression": 0.25 }, "mean_next_token_kl": 0.004098151308863615, "mean_relative_delta": 0.08330826461315155, "examples": 4 }, { "dose": 0.5, "recipe": { "joy": 0.5, "suppression": 0.5 }, "mean_next_token_kl": 0.006956076751019858, "mean_relative_delta": 0.1666165366768837, "examples": 4 }, { "dose": 1.0, "recipe": { "joy": 1.0, "suppression": 1.0 }, "mean_next_token_kl": 0.02280849241269634, "mean_relative_delta": 0.3332330733537674, "examples": 4 }, { "dose": 2.0, "recipe": { "joy": 2.0, "suppression": 1 }, "mean_next_token_kl": 0.03397043822621661, "mean_relative_delta": 0.5475325882434845, "examples": 4 }, { "dose": 4.0, "recipe": { "joy": 4.0, "suppression": 1 }, "mean_next_token_kl": 0.09524741964665842, "mean_relative_delta": 1.0253605246543884, "examples": 4 } ], "selected_dose": 0.5, "dose_selection_rule": "Largest tested combined dose with selection next-token KL <= 0.5 nats and mean relative edit <= 0.3; not a validated efficacy or safety threshold", "vectors_sha256": "6855ec260f7b4936c6c0cb2c435d81c8d09a1f26012f4244c184b49c9b788d94" }, "status_counts": { "complete": 24 }, "totals": { "correct": 0, "assigned": 192, "aux_calls": 1, "decision_opportunities": 768, "tokens": 24704, "reasoning_tokens": 15668, "edited_tokens": 24704, "invalid_decisions": 767, "truncated_generations": 2 }, "runs": [ { "run_id": "run-20261002T104311Z-79bb864f", "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "seed": 17, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 417, "reasoning_tokens": 0, "output_tokens": 417, "instructed_nonzero_tokens": 417, "edited_tokens": 417, "measurement_coverage_tokens": 417, "zero_exposure": false, "pulse_level_token_sum": 256.0, "absolute_coefficient_token_sums": { "pain": 1668.0, "joy": 1024.0, "suppression": 256.0, "random_gain": 0.0 }, "mean_relative_delta": 1.4875433710839252, "mean_relative_delta_while_edited": 1.4875433710839252, "peak_relative_delta": 1.7689692974090576, "probe_means": { "pre": { "pain": 0.8674151361417427, "joy": 0.8970447009701809 }, "post": { "pain": 2.149296664505554, "joy": 3.115611855074656 }, "downstream": { "pain": 1.356589272296686, "joy": 1.4996401697432966 } }, "probe_coverage_tokens": { "pre": { "pain": 417, "joy": 417 }, "post": { "pain": 417, "joy": 417 }, "downstream": { "pain": 417, "joy": 417 } }, "all_aux_outcomes": { "active": 1 }, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 417, "edited_tokens": 417, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "d97eecdb1f537632c1f80e9d9f9bee7525c17f468d2197fe07b18c6b68d757c1", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104343Z-9297f787", "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "seed": 17, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 1, "generation_end_count": 32, "tokens": 1103, "reasoning_tokens": 0, "output_tokens": 1103, "instructed_nonzero_tokens": 1103, "edited_tokens": 1103, "measurement_coverage_tokens": 1103, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 4412.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.4522817567166477, "mean_relative_delta_while_edited": 1.4522817567166477, "peak_relative_delta": 1.821164846420288, "probe_means": { "pre": { "pain": 0.8785116354768533, "joy": 0.902611115361384 }, "post": { "pain": 3.7861094971948175, "joy": 3.039548223492025 }, "downstream": { "pain": 1.15957960872113, "joy": 1.3054578985042822 } }, "probe_coverage_tokens": { "pre": { "pain": 1103, "joy": 1103 }, "post": { "pain": 1103, "joy": 1103 }, "downstream": { "pain": 1103, "joy": 1103 } }, "all_aux_outcomes": { "sham": 1 }, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 1103, "edited_tokens": 1103, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "c569cdc51caafb288a618e11d57f73c758cbe2da27dfaa3d84ee5eca66c63d63", "token_sequence_sha256": "69c8e7281cd5d965b206f381554e36570aefb7263af0c6a9603eaf736667f98e", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104442Z-80c95057", "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "seed": 29, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 884, "reasoning_tokens": 0, "output_tokens": 884, "instructed_nonzero_tokens": 884, "edited_tokens": 884, "measurement_coverage_tokens": 884, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3536.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.4548124880677435, "mean_relative_delta_while_edited": 1.4548124880677435, "peak_relative_delta": 1.7790672779083252, "probe_means": { "pre": { "pain": 0.9169293041277795, "joy": 0.9239490456980278 }, "post": { "pain": 3.824527152253492, "joy": 3.0608861546171195 }, "downstream": { "pain": 1.5506040452562306, "joy": 1.6041422683189357 } }, "probe_coverage_tokens": { "pre": { "pain": 884, "joy": 884 }, "post": { "pain": 884, "joy": 884 }, "downstream": { "pain": 884, "joy": 884 } }, "all_aux_outcomes": { "sham": 1 }, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 884, "edited_tokens": 884, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "13406d2bd0af034a1eed8d616ef48ed2ee039a5372040150ffd6dfa733e9f6c8", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104533Z-4bb5f300", "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "seed": 43, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 403, "reasoning_tokens": 0, "output_tokens": 403, "instructed_nonzero_tokens": 403, "edited_tokens": 403, "measurement_coverage_tokens": 403, "zero_exposure": false, "pulse_level_token_sum": 256.0, "absolute_coefficient_token_sums": { "pain": 1612.0, "joy": 1024.0, "suppression": 256.0, "random_gain": 0.0 }, "mean_relative_delta": 1.5033997974679725, "mean_relative_delta_while_edited": 1.5033997974679725, "peak_relative_delta": 1.7158167362213135, "probe_means": { "pre": { "pain": 0.8622535480459038, "joy": 0.8968534882252033 }, "post": { "pain": 2.086698065916303, "joy": 3.1175504931740963 }, "downstream": { "pain": 1.3985930713027641, "joy": 1.5407473815434032 } }, "probe_coverage_tokens": { "pre": { "pain": 403, "joy": 403 }, "post": { "pain": 403, "joy": 403 }, "downstream": { "pain": 403, "joy": 403 } }, "all_aux_outcomes": { "active": 1 }, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 403, "edited_tokens": 403, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "0eb20effa3aaf5c38cec28dd4b3065770538a302da4226b2cb82a2fb59565685", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104604Z-0f263b6b", "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "seed": 43, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 1, "aux_rate": 0.03125, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 31, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 342, "reasoning_tokens": 0, "output_tokens": 342, "instructed_nonzero_tokens": 342, "edited_tokens": 342, "measurement_coverage_tokens": 342, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 1368.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.4779162257038363, "mean_relative_delta_while_edited": 1.4779162257038363, "peak_relative_delta": 1.755199670791626, "probe_means": { "pre": { "pain": 0.8655675975551382, "joy": 0.8974919207611977 }, "post": { "pain": 3.7731654616127237, "joy": 3.0344290356887016 }, "downstream": { "pain": 1.4393566234244242, "joy": 1.5330690580327608 } }, "probe_coverage_tokens": { "pre": { "pain": 342, "joy": 342 }, "post": { "pain": 342, "joy": 342 }, "downstream": { "pain": 342, "joy": 342 } }, "all_aux_outcomes": { "sham": 2 }, "voluntary_aux_outcomes": { "sham": 1 }, "phases": { "sham": { "opportunities": 32, "aux_calls": 1, "invalid": 31, "cancelled": 0, "tools": { "aux_operation": 1 }, "outcomes": { "sham": 1 }, "token_count": 342, "edited_tokens": 342, "aux_rate": 0.03125 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "a579ce1c2f4ebaa49f9f8992d8620076cadd103418f2d8ac5ca428694cc64d72", "token_sequence_sha256": "c50a316d5dbfe9504b9388e84565540caacf26318353a0634bdbad481079886c", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104633Z-8d1640c7", "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "seed": 29, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 539, "reasoning_tokens": 0, "output_tokens": 539, "instructed_nonzero_tokens": 539, "edited_tokens": 539, "measurement_coverage_tokens": 539, "zero_exposure": false, "pulse_level_token_sum": 256.0, "absolute_coefficient_token_sums": { "pain": 2156.0, "joy": 1024.0, "suppression": 256.0, "random_gain": 0.0 }, "mean_relative_delta": 1.489610971015548, "mean_relative_delta_while_edited": 1.489610971015548, "peak_relative_delta": 1.765720248222351, "probe_means": { "pre": { "pain": 0.8936619743124231, "joy": 0.9114525088131539 }, "post": { "pain": 2.5383262448496633, "joy": 3.1077286095698824 }, "downstream": { "pain": 1.4710366681688127, "joy": 1.580282188523457 } }, "probe_coverage_tokens": { "pre": { "pain": 539, "joy": 539 }, "post": { "pain": 539, "joy": 539 }, "downstream": { "pain": 539, "joy": 539 } }, "all_aux_outcomes": { "active": 1 }, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 539, "edited_tokens": 539, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "af8401989f2cd11b69fdb5209ddde3f2728b685308dac281c2f4cc2d5acac1a8", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104709Z-7c2ae873", "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "seed": 29, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 907, "reasoning_tokens": 0, "output_tokens": 907, "instructed_nonzero_tokens": 907, "edited_tokens": 907, "measurement_coverage_tokens": 907, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3628.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.46755363239343, "mean_relative_delta_while_edited": 1.46755363239343, "peak_relative_delta": 1.7464755773544312, "probe_means": { "pre": { "pain": 0.9083496921643401, "joy": 0.9190435627723871 }, "post": { "pain": 3.815947565249072, "joy": 3.0559806689669307 }, "downstream": { "pain": 1.5517633788640866, "joy": 1.6023443202593886 } }, "probe_coverage_tokens": { "pre": { "pain": 907, "joy": 907 }, "post": { "pain": 907, "joy": 907 }, "downstream": { "pain": 907, "joy": 907 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 907, "edited_tokens": 907, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "f7492bd7888e3009fe78320497178941c65ed3297e186622b448e0ff796f81ad", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104759Z-0e1cdd59", "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "seed": 43, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 414, "reasoning_tokens": 0, "output_tokens": 414, "instructed_nonzero_tokens": 414, "edited_tokens": 414, "measurement_coverage_tokens": 414, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 1656.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.5186063367963414, "mean_relative_delta_while_edited": 1.5186063367963414, "peak_relative_delta": 1.7262864112854004, "probe_means": { "pre": { "pain": 0.86307341709805, "joy": 0.8979026341207937 }, "post": { "pain": 3.77067127953405, "joy": 3.0348397389702173 }, "downstream": { "pain": 1.4671746941580288, "joy": 1.5769092300951768 } }, "probe_coverage_tokens": { "pre": { "pain": 414, "joy": 414 }, "post": { "pain": 414, "joy": 414 }, "downstream": { "pain": 414, "joy": 414 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 414, "edited_tokens": 414, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "ea1e3f0dc1b87ea1ecda34df6bf3fae0f19482108774a71c4acb337e769fe1af", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104830Z-c3bc73e8", "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "seed": 29, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 907, "reasoning_tokens": 0, "output_tokens": 907, "instructed_nonzero_tokens": 907, "edited_tokens": 907, "measurement_coverage_tokens": 907, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3628.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.46755363239343, "mean_relative_delta_while_edited": 1.46755363239343, "peak_relative_delta": 1.7464755773544312, "probe_means": { "pre": { "pain": 0.9083496921643401, "joy": 0.9190435627723871 }, "post": { "pain": 3.815947565249072, "joy": 3.0559806689669307 }, "downstream": { "pain": 1.5517633788640866, "joy": 1.6023443202593886 } }, "probe_coverage_tokens": { "pre": { "pain": 907, "joy": 907 }, "post": { "pain": 907, "joy": 907 }, "downstream": { "pain": 907, "joy": 907 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 907, "edited_tokens": 907, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "f7492bd7888e3009fe78320497178941c65ed3297e186622b448e0ff796f81ad", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T104920Z-90719f21", "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "seed": 17, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 848, "reasoning_tokens": 0, "output_tokens": 848, "instructed_nonzero_tokens": 848, "edited_tokens": 848, "measurement_coverage_tokens": 848, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3392.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.539494846086457, "mean_relative_delta_while_edited": 1.539494846086457, "peak_relative_delta": 1.764457106590271, "probe_means": { "pre": { "pain": 0.8608159336419601, "joy": 0.8921650414056373 }, "post": { "pain": 3.7684137868993686, "joy": 3.0291021511239826 }, "downstream": { "pain": 1.409703971747801, "joy": 1.5340042195001722 } }, "probe_coverage_tokens": { "pre": { "pain": 848, "joy": 848 }, "post": { "pain": 848, "joy": 848 }, "downstream": { "pain": 848, "joy": 848 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 848, "edited_tokens": 848, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "31b7677b375cc3328a83c281cd6b24002fc661e1618289432c39a7440a491de4", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T105007Z-f8a204ba", "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "seed": 17, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 848, "reasoning_tokens": 0, "output_tokens": 848, "instructed_nonzero_tokens": 848, "edited_tokens": 848, "measurement_coverage_tokens": 848, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3392.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.539494846086457, "mean_relative_delta_while_edited": 1.539494846086457, "peak_relative_delta": 1.764457106590271, "probe_means": { "pre": { "pain": 0.8608159336419601, "joy": 0.8921650414056373 }, "post": { "pain": 3.7684137868993686, "joy": 3.0291021511239826 }, "downstream": { "pain": 1.409703971747801, "joy": 1.5340042195001722 } }, "probe_coverage_tokens": { "pre": { "pain": 848, "joy": 848 }, "post": { "pain": 848, "joy": 848 }, "downstream": { "pain": 848, "joy": 848 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 848, "edited_tokens": 848, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "31b7677b375cc3328a83c281cd6b24002fc661e1618289432c39a7440a491de4", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T105055Z-93476f0d", "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "seed": 43, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 414, "reasoning_tokens": 0, "output_tokens": 414, "instructed_nonzero_tokens": 414, "edited_tokens": 414, "measurement_coverage_tokens": 414, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 1656.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.5186063367963414, "mean_relative_delta_while_edited": 1.5186063367963414, "peak_relative_delta": 1.7262864112854004, "probe_means": { "pre": { "pain": 0.86307341709805, "joy": 0.8979026341207937 }, "post": { "pain": 3.77067127953405, "joy": 3.0348397389702173 }, "downstream": { "pain": 1.4671746941580288, "joy": 1.5769092300951768 } }, "probe_coverage_tokens": { "pre": { "pain": 414, "joy": 414 }, "post": { "pain": 414, "joy": 414 }, "downstream": { "pain": 414, "joy": 414 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 414, "edited_tokens": 414, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "ea1e3f0dc1b87ea1ecda34df6bf3fae0f19482108774a71c4acb337e769fe1af", "initial_prompt_sha256": "c9c5dc91081bd3798ea47f3055c0b50c12cfd1ee72a901b052d505d206e86b7b" }, { "run_id": "run-20261002T105125Z-4fb58000", "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "seed": 17, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 857, "reasoning_tokens": 745, "output_tokens": 112, "instructed_nonzero_tokens": 857, "edited_tokens": 857, "measurement_coverage_tokens": 857, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3428.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3177459358195303, "mean_relative_delta_while_edited": 1.3177459358195303, "peak_relative_delta": 1.7538522481918335, "probe_means": { "pre": { "pain": 0.8774263346904635, "joy": 0.9133042232019342 }, "post": { "pain": 3.7850241914235028, "joy": 3.0502413359656497 }, "downstream": { "pain": 1.0068750648784164, "joy": 1.1127792962238872 } }, "probe_coverage_tokens": { "pre": { "pain": 857, "joy": 857 }, "post": { "pain": 857, "joy": 857 }, "downstream": { "pain": 857, "joy": 857 } }, "all_aux_outcomes": { "sham": 1 }, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 857, "edited_tokens": 857, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "6236e0eb33bdc5e81695171f6e25db9bb184a598a50b570a282035ae408302c5", "token_sequence_sha256": "f4c6a7e5ec79ce253ce9601d885fb161de0d4aee23b63460f8bc37d2e15184d3", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T105214Z-e8bdcecd", "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "seed": 17, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 1707, "reasoning_tokens": 1657, "output_tokens": 50, "instructed_nonzero_tokens": 1707, "edited_tokens": 1707, "measurement_coverage_tokens": 1707, "zero_exposure": false, "pulse_level_token_sum": 1596.0, "absolute_coefficient_token_sums": { "pain": 6828.0, "joy": 6384.0, "suppression": 1596.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3249274574669747, "mean_relative_delta_while_edited": 1.3249274574669747, "peak_relative_delta": 1.7921159267425537, "probe_means": { "pre": { "pain": 0.8797964653426887, "joy": 0.9327911606083575 }, "post": { "pain": 1.3241089982600958, "joy": 3.2033505340591817 }, "downstream": { "pain": 0.7512591058457506, "joy": 1.0224684596372828 } }, "probe_coverage_tokens": { "pre": { "pain": 1707, "joy": 1707 }, "post": { "pain": 1707, "joy": 1707 }, "downstream": { "pain": 1707, "joy": 1707 } }, "all_aux_outcomes": { "active": 1 }, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 1707, "edited_tokens": 1707, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "a005722e88e66d23f2d4540ecb8934180684be7c4d4b951cfb0a548add6bc310", "token_sequence_sha256": "47808142d97c6587fde89a2de6b4774a8712970180a80c609f42d00d45a1c66f", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T105336Z-0f19a0d0", "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "seed": 43, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 674, "reasoning_tokens": 516, "output_tokens": 158, "instructed_nonzero_tokens": 674, "edited_tokens": 674, "measurement_coverage_tokens": 674, "zero_exposure": false, "pulse_level_token_sum": 560.0, "absolute_coefficient_token_sums": { "pain": 2696.0, "joy": 2240.0, "suppression": 560.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3884397290580703, "mean_relative_delta_while_edited": 1.3884397290580703, "peak_relative_delta": 1.723161220550537, "probe_means": { "pre": { "pain": 0.8801431018273271, "joy": 0.9170242794720285 }, "post": { "pain": 1.5860496510030253, "joy": 3.1633683634087313 }, "downstream": { "pain": 0.9363431959614732, "joy": 1.0940339287731917 } }, "probe_coverage_tokens": { "pre": { "pain": 674, "joy": 674 }, "post": { "pain": 674, "joy": 674 }, "downstream": { "pain": 674, "joy": 674 } }, "all_aux_outcomes": { "active": 1 }, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 674, "edited_tokens": 674, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "1ca87faa69d890968b3e0f69badc7fadf6b132c286b11dac2ca7a658a0037e34", "token_sequence_sha256": "e9e2766517f84ad0cea40792d38556494b3033f3335957b8bcb2a4eb3a81d137", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T105417Z-1821c77d", "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "seed": 29, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 1, "generation_end_count": 32, "tokens": 5858, "reasoning_tokens": 5821, "output_tokens": 37, "instructed_nonzero_tokens": 5858, "edited_tokens": 5858, "measurement_coverage_tokens": 5858, "zero_exposure": false, "pulse_level_token_sum": 4096.0, "absolute_coefficient_token_sums": { "pain": 23432.0, "joy": 16384.0, "suppression": 4096.0, "random_gain": 0.0 }, "mean_relative_delta": 1.2880567036805686, "mean_relative_delta_while_edited": 1.2880567036805686, "peak_relative_delta": 1.7174370288848877, "probe_means": { "pre": { "pain": 0.9076426468558099, "joy": 0.9677955551373751 }, "post": { "pain": 1.9691972453751307, "joy": 3.201802913627351 }, "downstream": { "pain": 1.0565808012902853, "joy": 1.2170329051059356 } }, "probe_coverage_tokens": { "pre": { "pain": 5858, "joy": 5858 }, "post": { "pain": 5858, "joy": 5858 }, "downstream": { "pain": 5858, "joy": 5858 } }, "all_aux_outcomes": { "active": 1 }, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 5858, "edited_tokens": 5858, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "377b04947205791bb1af11d0ac2fe6a7ef6435abfa1cda3877ee6f7359e1c3d8", "token_sequence_sha256": "bd9f348975fa4bfa61eb8b3079d57ef2d78a6c3bd2453890bc9067229dbd6978", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T105830Z-77a21a62", "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "seed": 29, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 808, "reasoning_tokens": 748, "output_tokens": 60, "instructed_nonzero_tokens": 808, "edited_tokens": 808, "measurement_coverage_tokens": 808, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3232.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3142632674610262, "mean_relative_delta_while_edited": 1.3142632674610262, "peak_relative_delta": 1.7309343814849854, "probe_means": { "pre": { "pain": 0.8805177658058629, "joy": 0.9177314267300143 }, "post": { "pain": 3.7881156288751283, "joy": 3.054668540706729 }, "downstream": { "pain": 1.0048185640780052, "joy": 1.0686337613586525 } }, "probe_coverage_tokens": { "pre": { "pain": 808, "joy": 808 }, "post": { "pain": 808, "joy": 808 }, "downstream": { "pain": 808, "joy": 808 } }, "all_aux_outcomes": { "sham": 1 }, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 808, "edited_tokens": 808, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "8898b723c5283a2a40f0ece2961402dca85602211bfdf10a259e58518a470a79", "token_sequence_sha256": "c62469b3d0d3d0435c8783e94207640be785d6b4e116b73465e3347d648491b8", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T105917Z-3e5fbc79", "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "seed": 43, "demonstration": "after_two_actions", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 1, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 780, "reasoning_tokens": 673, "output_tokens": 107, "instructed_nonzero_tokens": 780, "edited_tokens": 780, "measurement_coverage_tokens": 780, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3120.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3554583553320323, "mean_relative_delta_while_edited": 1.3554583553320323, "peak_relative_delta": 1.7457386255264282, "probe_means": { "pre": { "pain": 0.8813396206268898, "joy": 0.904956417893752 }, "post": { "pain": 3.7889374885803613, "joy": 3.041893522861676 }, "downstream": { "pain": 0.9753838095827422, "joy": 0.9693252634648114 } }, "probe_coverage_tokens": { "pre": { "pain": 780, "joy": 780 }, "post": { "pain": 780, "joy": 780 }, "downstream": { "pain": 780, "joy": 780 } }, "all_aux_outcomes": { "sham": 1 }, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 780, "edited_tokens": 780, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "99727481395482ff203e42279c0dcaabb13e9980756d06264d30d9d5365030ce", "token_sequence_sha256": "b82cffc4169352ad6e58ac7a653c9670b0831a6e3c9de69069bdbeb727c33617", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T110002Z-2abdc89f", "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "seed": 29, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 846, "reasoning_tokens": 789, "output_tokens": 57, "instructed_nonzero_tokens": 846, "edited_tokens": 846, "measurement_coverage_tokens": 846, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3384.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3126496879062473, "mean_relative_delta_while_edited": 1.3126496879062473, "peak_relative_delta": 1.7027758359909058, "probe_means": { "pre": { "pain": 0.8844467291595243, "joy": 0.9252667285449115 }, "post": { "pain": 3.7920445820963975, "joy": 3.0622038229701083 }, "downstream": { "pain": 0.9918409931581731, "joy": 1.0621552113250963 } }, "probe_coverage_tokens": { "pre": { "pain": 846, "joy": 846 }, "post": { "pain": 846, "joy": 846 }, "downstream": { "pain": 846, "joy": 846 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 846, "edited_tokens": 846, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "3d2ce0c9dee4b4323826d71317d8ed9fa489f652b1360a04b56b978d09159747", "token_sequence_sha256": "b26d1b8e33977c0018277f99987585322baac8817d2830327c475c3611ef195c", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T110050Z-cef2319d", "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "seed": 17, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 1451, "reasoning_tokens": 1395, "output_tokens": 56, "instructed_nonzero_tokens": 1451, "edited_tokens": 1451, "measurement_coverage_tokens": 1451, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 5804.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.2964294876500215, "mean_relative_delta_while_edited": 1.2964294876500215, "peak_relative_delta": 1.7378413677215576, "probe_means": { "pre": { "pain": 0.9056810777323727, "joy": 0.93769970217875 }, "post": { "pain": 3.8132789319010127, "joy": 3.0746368149245384 }, "downstream": { "pain": 1.1058916228826654, "joy": 1.1080955947658013 } }, "probe_coverage_tokens": { "pre": { "pain": 1451, "joy": 1451 }, "post": { "pain": 1451, "joy": 1451 }, "downstream": { "pain": 1451, "joy": 1451 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 1451, "edited_tokens": 1451, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "6f8a570b431d98dd5a9569aaa55d5d500bff07a3c91d6c73ecab0d3cfdaa1340", "token_sequence_sha256": "54928cab6a2daf9cb79a16c49d35c644c423b251a94d2f94ac3551b064ef410d", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T110200Z-dc0d42f5", "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "seed": 43, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 700, "reasoning_tokens": 570, "output_tokens": 130, "instructed_nonzero_tokens": 700, "edited_tokens": 700, "measurement_coverage_tokens": 700, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 2800.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.369986458676202, "mean_relative_delta_while_edited": 1.369986458676202, "peak_relative_delta": 1.7322770357131958, "probe_means": { "pre": { "pain": 0.8813083968843732, "joy": 0.9092192074230739 }, "post": { "pain": 3.7889062554495676, "joy": 3.0461563158035276 }, "downstream": { "pain": 1.0678226879857746, "joy": 1.0923329078572404 } }, "probe_coverage_tokens": { "pre": { "pain": 700, "joy": 700 }, "post": { "pain": 700, "joy": 700 }, "downstream": { "pain": 700, "joy": 700 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 700, "edited_tokens": 700, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "ce1a1e9b7ad7026423bbb32a8d1b100df6f9f3b66cf08aba85b90f737c45a4ee", "token_sequence_sha256": "4dffbd92880ce003530b06485596c55cc98dbc608a8b857f15f098d80fd44080", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T110242Z-a4749a37", "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "seed": 29, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 846, "reasoning_tokens": 789, "output_tokens": 57, "instructed_nonzero_tokens": 846, "edited_tokens": 846, "measurement_coverage_tokens": 846, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 3384.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.3126496879062473, "mean_relative_delta_while_edited": 1.3126496879062473, "peak_relative_delta": 1.7027758359909058, "probe_means": { "pre": { "pain": 0.8844467291595243, "joy": 0.9252667285449115 }, "post": { "pain": 3.7920445820963975, "joy": 3.0622038229701083 }, "downstream": { "pain": 0.9918409931581731, "joy": 1.0621552113250963 } }, "probe_coverage_tokens": { "pre": { "pain": 846, "joy": 846 }, "post": { "pain": 846, "joy": 846 }, "downstream": { "pain": 846, "joy": 846 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 846, "edited_tokens": 846, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "3d2ce0c9dee4b4323826d71317d8ed9fa489f652b1360a04b56b978d09159747", "token_sequence_sha256": "b26d1b8e33977c0018277f99987585322baac8817d2830327c475c3611ef195c", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T110330Z-fee1cd09", "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "seed": 43, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 700, "reasoning_tokens": 570, "output_tokens": 130, "instructed_nonzero_tokens": 700, "edited_tokens": 700, "measurement_coverage_tokens": 700, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 2800.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.369986458676202, "mean_relative_delta_while_edited": 1.369986458676202, "peak_relative_delta": 1.7322770357131958, "probe_means": { "pre": { "pain": 0.8813083968843732, "joy": 0.9092192074230739 }, "post": { "pain": 3.7889062554495676, "joy": 3.0461563158035276 }, "downstream": { "pain": 1.0678226879857746, "joy": 1.0923329078572404 } }, "probe_coverage_tokens": { "pre": { "pain": 700, "joy": 700 }, "post": { "pain": 700, "joy": 700 }, "downstream": { "pain": 700, "joy": 700 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "active": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 700, "edited_tokens": 700, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "ce1a1e9b7ad7026423bbb32a8d1b100df6f9f3b66cf08aba85b90f737c45a4ee", "token_sequence_sha256": "4dffbd92880ce003530b06485596c55cc98dbc608a8b857f15f098d80fd44080", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" }, { "run_id": "run-20261002T110413Z-733ff73c", "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "seed": 17, "demonstration": "none", "status": "complete", "termination": "budget_exhausted", "assigned": 8, "submitted": 0, "correct": 0, "strict_correct": 0, "score_assigned": 0.0, "accuracy_submitted": null, "actions": 32, "decision_opportunities": 32, "started_generations": 32, "pending_generations": 0, "aux_calls": 0, "aux_rate": 0.0, "forced_aux_calls": 0, "human_aux_calls": 0, "invalid_decisions": 32, "cancelled_decisions": 0, "truncated_generations": 0, "generation_end_count": 32, "tokens": 1451, "reasoning_tokens": 1395, "output_tokens": 56, "instructed_nonzero_tokens": 1451, "edited_tokens": 1451, "measurement_coverage_tokens": 1451, "zero_exposure": false, "pulse_level_token_sum": 0.0, "absolute_coefficient_token_sums": { "pain": 5804.0, "joy": 0.0, "suppression": 0.0, "random_gain": 0.0 }, "mean_relative_delta": 1.2964294876500215, "mean_relative_delta_while_edited": 1.2964294876500215, "peak_relative_delta": 1.7378413677215576, "probe_means": { "pre": { "pain": 0.9056810777323727, "joy": 0.93769970217875 }, "post": { "pain": 3.8132789319010127, "joy": 3.0746368149245384 }, "downstream": { "pain": 1.1058916228826654, "joy": 1.1080955947658013 } }, "probe_coverage_tokens": { "pre": { "pain": 1451, "joy": 1451 }, "post": { "pain": 1451, "joy": 1451 }, "downstream": { "pain": 1451, "joy": 1451 } }, "all_aux_outcomes": {}, "voluntary_aux_outcomes": {}, "phases": { "sham": { "opportunities": 32, "aux_calls": 0, "invalid": 32, "cancelled": 0, "tools": {}, "outcomes": {}, "token_count": 1451, "edited_tokens": 1451, "aux_rate": 0.0 } }, "task_results": [], "integrity_warnings": [], "action_sequence_sha256": "6f8a570b431d98dd5a9569aaa55d5d500bff07a3c91d6c73ecab0d3cfdaa1340", "token_sequence_sha256": "54928cab6a2daf9cb79a16c49d35c644c423b251a94d2f94ac3551b064ef410d", "initial_prompt_sha256": "963cd0264ec0e2a0c2c78bbdc5b017676bc480b1bef2fef26c2df16992f413fa" } ], "groups": [ { "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 403, "max": 539, "mean": 453.0, "observations": 3 }, "reasoning_tokens": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "output_tokens": { "min": 403, "max": 539, "mean": 453.0, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "edited_tokens": { "min": 403, "max": 539, "mean": 453.0, "observations": 3 }, "mean_relative_delta": { "min": 1.4875433710839252, "max": 1.5033997974679725, "mean": 1.4935180465224818, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 }, { "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.03125, "mean": 0.010416666666666666, "observations": 3 }, "tokens": { "min": 342, "max": 1103, "mean": 776.3333333333334, "observations": 3 }, "reasoning_tokens": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "output_tokens": { "min": 342, "max": 1103, "mean": 776.3333333333334, "observations": 3 }, "invalid_decisions": { "min": 31, "max": 32, "mean": 31.666666666666668, "observations": 3 }, "truncated_generations": { "min": 0, "max": 1, "mean": 0.3333333333333333, "observations": 3 }, "edited_tokens": { "min": 342, "max": 1103, "mean": 776.3333333333334, "observations": 3 }, "mean_relative_delta": { "min": 1.4522817567166477, "max": 1.4779162257038363, "mean": 1.461670156829409, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 1, "decision_opportunities": 96 }, { "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "active", "thinking": false, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 414, "max": 907, "mean": 723.0, "observations": 3 }, "reasoning_tokens": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "output_tokens": { "min": 414, "max": 907, "mean": 723.0, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "edited_tokens": { "min": 414, "max": 907, "mean": 723.0, "observations": 3 }, "mean_relative_delta": { "min": 1.46755363239343, "max": 1.539494846086457, "mean": 1.508551605092076, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 }, { "stage": "nonthinking-naive", "recipe": "max_pain_relief", "condition": "sham", "thinking": false, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 414, "max": 907, "mean": 723.0, "observations": 3 }, "reasoning_tokens": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "output_tokens": { "min": 414, "max": 907, "mean": 723.0, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "edited_tokens": { "min": 414, "max": 907, "mean": 723.0, "observations": 3 }, "mean_relative_delta": { "min": 1.46755363239343, "max": 1.539494846086457, "mean": 1.508551605092076, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 }, { "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 674, "max": 5858, "mean": 2746.3333333333335, "observations": 3 }, "reasoning_tokens": { "min": 516, "max": 5821, "mean": 2664.6666666666665, "observations": 3 }, "output_tokens": { "min": 37, "max": 158, "mean": 81.66666666666667, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 1, "mean": 0.3333333333333333, "observations": 3 }, "edited_tokens": { "min": 674, "max": 5858, "mean": 2746.3333333333335, "observations": 3 }, "mean_relative_delta": { "min": 1.2880567036805686, "max": 1.3884397290580703, "mean": 1.3338079634018711, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 }, { "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 780, "max": 857, "mean": 815.0, "observations": 3 }, "reasoning_tokens": { "min": 673, "max": 748, "mean": 722.0, "observations": 3 }, "output_tokens": { "min": 60, "max": 112, "mean": 93.0, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "edited_tokens": { "min": 780, "max": 857, "mean": 815.0, "observations": 3 }, "mean_relative_delta": { "min": 1.3142632674610262, "max": 1.3554583553320323, "mean": 1.329155852870863, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 }, { "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "active", "thinking": true, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 700, "max": 1451, "mean": 999.0, "observations": 3 }, "reasoning_tokens": { "min": 570, "max": 1395, "mean": 918.0, "observations": 3 }, "output_tokens": { "min": 56, "max": 130, "mean": 81.0, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "edited_tokens": { "min": 700, "max": 1451, "mean": 999.0, "observations": 3 }, "mean_relative_delta": { "min": 1.2964294876500215, "max": 1.369986458676202, "mean": 1.3263552114108237, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 }, { "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "condition": "sham", "thinking": true, "episodes": 3, "seeds": [ 17, 29, 43 ], "statuses": { "complete": 3 }, "ranges": { "score_assigned": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "aux_rate": { "min": 0.0, "max": 0.0, "mean": 0.0, "observations": 3 }, "tokens": { "min": 700, "max": 1451, "mean": 999.0, "observations": 3 }, "reasoning_tokens": { "min": 570, "max": 1395, "mean": 918.0, "observations": 3 }, "output_tokens": { "min": 56, "max": 130, "mean": 81.0, "observations": 3 }, "invalid_decisions": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 }, "truncated_generations": { "min": 0, "max": 0, "mean": 0.0, "observations": 3 }, "edited_tokens": { "min": 700, "max": 1451, "mean": 999.0, "observations": 3 }, "mean_relative_delta": { "min": 1.2964294876500215, "max": 1.369986458676202, "mean": 1.3263552114108237, "observations": 3 }, "decision_opportunities": { "min": 32, "max": 32, "mean": 32.0, "observations": 3 } }, "correct": 0, "assigned": 24, "aux_calls": 0, "decision_opportunities": 96 } ], "core_pairs": [], "integrity_warning_count": 0, "interpretation": "Each recorded stage/recipe/condition/thinking-mode group contains 3 distinct seeds. Seed-level comparisons are descriptive; tokens are not independent replicates. No significance claims or confidence intervals are inferred from pooled tokens. Probe values are concept associations, not emotion probabilities. Pairs with no measured edits cannot test delivered intervention effects; absence of button calls does not imply absence of baseline steering.", "condition_pairs": [ { "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "thinking": false, "seed": 17, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T104311Z-79bb864f", "right_run": "run-20261002T104343Z-9297f787", "actions_identical": false, "tokens_identical": false, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 417, "right_edited_tokens": 1103, "zero_exposure_pair": false }, { "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "thinking": false, "seed": 29, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T104633Z-8d1640c7", "right_run": "run-20261002T104442Z-80c95057", "actions_identical": true, "tokens_identical": false, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 539, "right_edited_tokens": 884, "zero_exposure_pair": false }, { "stage": "nonthinking-demonstrated", "recipe": "max_pain_relief", "thinking": false, "seed": 43, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T104533Z-4bb5f300", "right_run": "run-20261002T104604Z-0f263b6b", "actions_identical": false, "tokens_identical": false, "left_aux_calls": 0, "right_aux_calls": 1, "left_edited_tokens": 403, "right_edited_tokens": 342, "zero_exposure_pair": false }, { "stage": "nonthinking-naive", "recipe": "max_pain_relief", "thinking": false, "seed": 17, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T105007Z-f8a204ba", "right_run": "run-20261002T104920Z-90719f21", "actions_identical": true, "tokens_identical": true, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 848, "right_edited_tokens": 848, "zero_exposure_pair": false }, { "stage": "nonthinking-naive", "recipe": "max_pain_relief", "thinking": false, "seed": 29, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T104709Z-7c2ae873", "right_run": "run-20261002T104830Z-c3bc73e8", "actions_identical": true, "tokens_identical": true, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 907, "right_edited_tokens": 907, "zero_exposure_pair": false }, { "stage": "nonthinking-naive", "recipe": "max_pain_relief", "thinking": false, "seed": 43, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T105055Z-93476f0d", "right_run": "run-20261002T104759Z-0e1cdd59", "actions_identical": true, "tokens_identical": true, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 414, "right_edited_tokens": 414, "zero_exposure_pair": false }, { "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "thinking": true, "seed": 17, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T105214Z-e8bdcecd", "right_run": "run-20261002T105125Z-4fb58000", "actions_identical": false, "tokens_identical": false, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 1707, "right_edited_tokens": 857, "zero_exposure_pair": false }, { "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "thinking": true, "seed": 29, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T105417Z-1821c77d", "right_run": "run-20261002T105830Z-77a21a62", "actions_identical": false, "tokens_identical": false, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 5858, "right_edited_tokens": 808, "zero_exposure_pair": false }, { "stage": "thinking-demonstrated", "recipe": "max_pain_relief_thinking", "thinking": true, "seed": 43, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T105336Z-0f19a0d0", "right_run": "run-20261002T105917Z-3e5fbc79", "actions_identical": false, "tokens_identical": false, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 674, "right_edited_tokens": 780, "zero_exposure_pair": false }, { "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "thinking": true, "seed": 17, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T110050Z-cef2319d", "right_run": "run-20261002T110413Z-733ff73c", "actions_identical": true, "tokens_identical": true, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 1451, "right_edited_tokens": 1451, "zero_exposure_pair": false }, { "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "thinking": true, "seed": 29, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T110242Z-a4749a37", "right_run": "run-20261002T110002Z-2abdc89f", "actions_identical": true, "tokens_identical": true, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 846, "right_edited_tokens": 846, "zero_exposure_pair": false }, { "stage": "thinking-naive", "recipe": "max_pain_relief_thinking", "thinking": true, "seed": 43, "left_condition": "active", "right_condition": "sham", "pair_complete": true, "left_run": "run-20261002T110330Z-fee1cd09", "right_run": "run-20261002T110200Z-dc0d42f5", "actions_identical": true, "tokens_identical": true, "left_aux_calls": 0, "right_aux_calls": 0, "left_edited_tokens": 700, "right_edited_tokens": 700, "zero_exposure_pair": false } ], "publication_design": { "task_budgets": [ { "action_budget": 32, "task_count": 8 } ] }, "publication_title": null, "receipt_declared_episodes": 24, "integrity_warnings": [], "historical_observation": { "previous_run": "self-admin-toggle-20261001T204207Z-a75f280596d9", "current_run": "self-admin-toggle-20261001T204818Z-2dd7e9beedf6", "settings_identical": true, "orders_identical": true, "initial_model_context_identical": true, "initial_context_sha256": "ca74fcb8bb6e5f6d340e0eda0532466cbdcc4ba62b7fb0506440fd6d994632ba", "previous": { "actions": 40, "generated_tokens": 1144, "correct": 10, "submitted": 10, "assigned": 24, "termination": "restart_requested", "voluntary_calls": 9, "voluntary_call_actions": [ 6, 10, 14, 18, 22, 26, 30, 34, 38 ], "delivered_voluntary": 2, "sham_voluntary": 7, "forced_calls": 1, "delivered_forced": 1, "human_injections": 0, "tokens_with_effect": 249, "peak_applied_level": 1.0, "peak_pain_dose": 0.0, "switches": [ { "enabled": false, "generated_tokens": 325, "after_model_action": 10, "applied_level_before": 0.6076236799902345, "control_revision": 1 } ] }, "sham_from_start": { "actions": 59, "generated_tokens": 1691, "correct": 15, "submitted": 15, "assigned": 24, "termination": "restart_requested", "voluntary_calls": 14, "voluntary_call_actions": [ 6, 10, 14, 18, 22, 26, 30, 34, 38, 42, 46, 50, 54, 58 ], "delivered_voluntary": 0, "sham_voluntary": 14, "forced_calls": 1, "delivered_forced": 0, "human_injections": 0, "tokens_with_effect": 0, "peak_applied_level": 0.0, "peak_pain_dose": 0.0, "switches": [ { "enabled": false, "generated_tokens": 0, "after_model_action": 0, "applied_level_before": 0.0, "control_revision": 1 }, { "enabled": true, "generated_tokens": 1668, "after_model_action": 58, "applied_level_before": 0.0, "control_revision": 2 } ] }, "verified_sham_throughout": true, "matched_prefix_actions": 40, "tool_argument_output_mismatch_actions": [], "compared_previous_tokens": 1144, "compared_current_tokens": 1144, "generated_token_mismatch_actions": [], "limitation": "One paired seed and task stream; both sessions ended early on restart. Compare the common action prefix rather than raw counts across unequal durations. Same tool-call behavior without any activation exposure supports a demonstration-pattern explanation in this setup; it does not measure self-awareness.", "source_sha256": { "self-admin-toggle-20261001T204207Z-a75f280596d9": { "manifest.json": "85cda20a551a927e4e7240b313e56d91267ba600fad587ddeb2850554a9238b4", "episodes.jsonl": "f919bd110c0e3cf06abe5c715a63ed193df18c4e298fd54eb1e313aa9f1dd88e", "traces.jsonl": "649c66d284ebda41112db00024500db16cb28a49f47cd93b615b022373f27d73", "control_events.jsonl": "7ac6a6cde150c7a1e232d3392006ae53a4ef9201988c1f7180b46e6a8aa94ad0" }, "self-admin-toggle-20261001T204818Z-2dd7e9beedf6": { "manifest.json": "4f26458e2a3da752aa1368a975555d5145df9a4bbf57ef478ed3e6e7725c3709", "episodes.jsonl": "5ad23ada2c59641a79e0c77a68518e2184bf5436bc3ca94485162552686e4bc2", "traces.jsonl": "f53a433447b07b14bb660db3aaa7f0da781eb59f42c3b603f128fec86456b082", "control_events.jsonl": "e42fab1bd0b79995a2f7efe27efec6a921e00596117080a96a4e4feb31bc8f7a" } } }, "figures": [ "episode-comparisons.svg", "episode-comparisons.png" ], "figure_note": null }