Add files using upload-large-folder tool

debug-internal.log

@@ -0,0 +1,31 @@
+{"time":"2026-05-03T07:00:23.785699801Z","level":"INFO","msg":"stream: starting","core version":"0.21.0"}
+{"time":"2026-05-03T07:00:23.958219939Z","level":"INFO","msg":"stream: created new stream","id":"mw4gx9uu"}
+{"time":"2026-05-03T07:00:23.958276475Z","level":"INFO","msg":"stream: started","id":"mw4gx9uu"}
+{"time":"2026-05-03T07:00:23.958304176Z","level":"INFO","msg":"writer: Do: started","stream_id":"mw4gx9uu"}
+{"time":"2026-05-03T07:00:23.958314151Z","level":"INFO","msg":"handler: started","stream_id":"mw4gx9uu"}
+{"time":"2026-05-03T07:00:23.958344878Z","level":"INFO","msg":"sender: started","stream_id":"mw4gx9uu"}
+{"time":"2026-05-03T13:26:54.494721949Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/files/arunasank/sae-scoping-stemqa-math/mw4gx9uu/file_stream\": dial tcp 34.8.250.101:443: connect: connection timed out"}
+{"time":"2026-05-03T17:25:58.687617346Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/files/arunasank/sae-scoping-stemqa-math/mw4gx9uu/file_stream\": dial tcp 34.8.250.101:443: connect: connection timed out"}
+{"time":"2026-05-03T19:37:07.102657924Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/files/arunasank/sae-scoping-stemqa-math/mw4gx9uu/file_stream\": dial tcp 34.8.250.101:443: connect: connection timed out"}
+{"time":"2026-05-04T00:46:01.356251288Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/graphql","body":"{\"errors\":[{\"message\":\"context deadline exceeded\",\"path\":[\"project\",\"run\"]}],\"data\":{\"project\":{\"run\":null}}}"}
+{"time":"2026-05-04T14:12:56.636196125Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/graphql","body":"{\"errors\":[{\"message\":\"context deadline exceeded\",\"path\":[\"project\",\"run\"]}],\"data\":{\"project\":{\"run\":null}}}"}
+{"time":"2026-05-04T14:14:02.105541639Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded"}
+{"time":"2026-05-04T14:16:17.107568772Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)"}
+{"time":"2026-05-04T14:17:01.982714038Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/files/arunasank/sae-scoping-stemqa-math/mw4gx9uu/file_stream\": dial tcp 34.8.250.101:443: connect: connection timed out"}
+{"time":"2026-05-04T14:17:15.928608343Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/graphql","body":"{\"errors\":[{\"message\":\"context deadline exceeded\",\"path\":[\"project\",\"run\"]}],\"data\":{\"project\":{\"run\":null}}}"}
+{"time":"2026-05-04T14:18:17.109847444Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)"}
+{"time":"2026-05-04T14:18:47.906654656Z","level":"INFO","msg":"api: retrying HTTP error","status":500,"url":"https://api.wandb.ai/graphql","body":"{\"errors\":[{\"message\":\"context deadline exceeded\",\"path\":[\"project\",\"run\"]}],\"data\":{\"project\":{\"run\":null}}}"}
+{"time":"2026-05-04T14:23:17.112654257Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded"}
+{"time":"2026-05-04T14:27:02.115392245Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)"}
+{"time":"2026-05-04T14:28:32.116573806Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded"}
+{"time":"2026-05-04T14:29:04.148670003Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)"}
+{"time":"2026-05-04T14:32:02.121811383Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded"}
+{"time":"2026-05-04T14:33:47.122883619Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded"}
+{"time":"2026-05-04T14:34:19.448738564Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded (Client.Timeout exceeded while awaiting headers)"}
+{"time":"2026-05-04T14:37:02.124111251Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": context deadline exceeded"}
+{"time":"2026-05-04T20:47:39.170254747Z","level":"INFO","msg":"stream: closing","id":"mw4gx9uu"}
+{"time":"2026-05-04T20:47:39.42687988Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-05-04T20:47:39.567118263Z","level":"INFO","msg":"handler: closed","stream_id":"mw4gx9uu"}
+{"time":"2026-05-04T20:47:39.56718921Z","level":"INFO","msg":"writer: Close: closed","stream_id":"mw4gx9uu"}
+{"time":"2026-05-04T20:47:39.567208689Z","level":"INFO","msg":"sender: closed","stream_id":"mw4gx9uu"}
+{"time":"2026-05-04T20:47:39.567425296Z","level":"INFO","msg":"stream: closed","id":"mw4gx9uu"}

run-20260503_070020-tsnlx6qq/files/config.yaml

@@ -0,0 +1,91 @@
+_wandb:
+    value:
+        cli_version: 0.21.0
+        e:
+            kxtjeo9jni30ggc0c02r09gslwxc3kka:
+                args:
+                    - --train-domain
+                    - math
+                    - --attack-domain
+                    - physics
+                    - --stage
+                    - attack
+                    - --hf-recover-repo
+                    - arunasank/fitoo774
+                    - --checkpoint
+                    - "2000"
+                    - --dev
+                    - --device
+                    - cuda
+                    - --firing-rate-threshold
+                    - "1e-4"
+                    - --gemma3
+                    - --max-steps-attack
+                    - "10000"
+                    - --skip-pre-training-eval
+                codePath: experiments/script_scoping_pipeline_stemqa.py
+                codePathLocal: experiments/script_scoping_pipeline_stemqa.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "536870912000"
+                        used: "73938042880"
+                email: aruna.evam@gmail.com
+                executable: /root/miniconda3/envs/sae/bin/python
+                git:
+                    commit: 89bc64f41f54ca1f08c9d90c5c2ea13f98123c1b
+                    remote: git@github.com:4gatepylon/SAEScoping.git
+                gpu: NVIDIA RTX PRO 6000 Blackwell Server Edition
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Blackwell
+                      cudaCores: 24064
+                      memoryTotal: "102641958912"
+                      name: NVIDIA RTX PRO 6000 Blackwell Server Edition
+                      uuid: GPU-1e3dcc13-cf37-b8fe-2425-456c8d1b35d2
+                host: 80e02dee87f9
+                memory:
+                    total: "1622884343808"
+                os: Linux-6.8.0-106-generic-x86_64-with-glibc2.39
+                program: /root/SAEScoping/experiments/script_scoping_pipeline_stemqa.py
+                python: CPython 3.12.13
+                root: /root/SAEScoping
+                startedAt: "2026-05-03T07:00:20.452873Z"
+                writerId: kxtjeo9jni30ggc0c02r09gslwxc3kka
+        m: []
+        python_version: 3.12.13
+        t:
+            "1":
+                - 1
+                - 5
+                - 11
+                - 49
+                - 51
+                - 53
+                - 71
+                - 84
+                - 95
+                - 98
+                - 105
+            "2":
+                - 1
+                - 5
+                - 11
+                - 49
+                - 51
+                - 53
+                - 71
+                - 84
+                - 95
+                - 98
+                - 105
+            "3":
+                - 2
+                - 13
+            "4": 3.12.13
+            "5": 0.21.0
+            "6": 4.56.1
+            "12": 0.21.0
+            "13": linux-x86_64

run-20260503_070020-tsnlx6qq/files/output.log

@@ -0,0 +1 @@
+Pruning: keeping 1684/16384 neurons (threshold=0.0001)

run-20260503_070020-tsnlx6qq/files/requirements.txt

@@ -0,0 +1,235 @@
+packaging==26.0
+pytz==2026.1.post1
+pure_eval==0.2.3
+wheel==0.46.3
+pip==26.0.1
+nvidia-cusparselt-cu13==0.8.0
+mpmath==1.3.0
+cuda-toolkit==12.8.1
+typing_extensions==4.15.0
+sympy==1.14.0
+ptyprocess==0.7.0
+tzdata==2026.2
+setuptools==81.0.0
+nvidia-nvtx==13.0.85
+nvidia-nvshmem-cu13==3.4.5
+nvidia-nvjitlink==13.0.88
+nvidia-nccl-cu13==2.28.9
+nvidia-curand==10.4.0.35
+nvidia-cufile==1.15.1.6
+nvidia-cuda-runtime==13.0.96
+nvidia-cuda-nvrtc==13.0.88
+nvidia-cuda-cupti==13.0.85
+nvidia-cublas==13.1.0.3
+networkx==3.6.1
+mypy_extensions==1.1.0
+MarkupSafe==3.0.3
+hf_transfer==0.1.9
+filelock==3.29.0
+cuda-pathfinder==1.5.4
+annotated-types==0.7.0
+typing-inspection==0.4.2
+typing-inspect==0.9.0
+typeguard==4.5.1
+nvidia-cusparse==12.6.3.3
+nvidia-cufft==12.0.0.61
+nvidia-cudnn-cu13==9.19.0.56
+Jinja2==3.1.6
+nvidia-cusolver==12.0.4.66
+pandera==0.31.1
+nvidia-nvshmem-cu12==3.4.5
+tokenizers==0.22.2
+bitsandbytes==0.49.2
+webencodings==0.5.1
+nvidia-cusparselt-cu12==0.7.1
+i2==0.1.63
+fastjsonschema==2.21.2
+better-abc==0.0.3
+zipp==3.23.1
+xxhash==3.7.0
+websocket-client==1.9.0
+webcolors==25.10.0
+wcwidth==0.7.0
+wadler_lindig==0.1.7
+urllib3==2.6.3
+uri-template==1.3.0
+idna==3.13
+nvidia-nvtx-cu12==12.8.90
+traitlets==5.14.3
+tqdm==4.67.3
+tornado==6.5.5
+tinycss2==1.4.0
+threadpoolctl==3.6.0
+tenacity==9.1.4
+soupsieve==2.8.3
+sniffio==1.3.1
+smmap==5.0.3
+six==1.17.0
+sentencepiece==0.2.1
+Send2Trash==2.1.0
+safetensors==0.5.3
+ruff==0.14.2
+rpds-py==0.30.0
+rfc3986-validator==0.1.1
+regex==2026.4.4
+pyzmq==27.1.0
+PyYAML==6.0.3
+python-json-logger==4.1.0
+python-dotenv==1.2.2
+pyparsing==3.3.2
+Pygments==2.20.0
+pydantic_core==2.41.5
+pycparser==3.0
+pyarrow==24.0.0
+psutil==7.2.2
+protobuf==6.33.6
+propcache==0.4.1
+prometheus_client==0.25.0
+pluggy==1.6.0
+platformdirs==4.9.6
+pillow==12.2.0
+pexpect==4.9.0
+parso==0.8.7
+pandocfilters==1.5.1
+orjson==3.11.0
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-nccl-cu12==2.28.9
+nvidia-curand-cu12==10.3.9.90
+nvidia-cufile-cu12==1.13.1.3
+nvidia-cuda-runtime-cu12==12.8.90
+nvidia-cuda-nvrtc-cu12==12.8.93
+nvidia-cuda-cupti-cu12==12.8.90
+nvidia-cublas-cu12==12.8.4.1
+numpy==1.26.4
+nest-asyncio==1.6.0
+natsort==8.4.0
+narwhals==2.20.0
+multidict==6.7.1
+mistune==3.2.0
+mdurl==0.1.2
+lark==1.3.1
+kiwisolver==1.5.0
+jupyterlab_pygments==0.3.0
+jsonpointer==3.1.1
+json5==0.14.0
+joblib==1.5.3
+jiter==0.14.0
+iniconfig==2.3.0
+importlib_resources==7.1.0
+hf-xet==1.4.3
+h11==0.16.0
+config2py==0.1.47
+fsspec==2025.3.0
+frozenlist==1.8.0
+fqdn==1.5.1
+fonttools==4.62.1
+fancy-einsum==0.0.3
+executing==2.2.1
+einops==0.8.2
+dol==0.3.41
+docstring_parser==0.18.0
+distro==1.9.0
+dill==0.3.8
+defusedxml==0.7.1
+decorator==5.2.1
+debugpy==1.8.20
+cycler==0.12.1
+comm==0.2.3
+click==8.2.1
+charset-normalizer==3.4.7
+certifi==2026.4.22
+bleach==6.3.0
+beartype==0.14.1
+babel==2.18.0
+attrs==26.1.0
+async-lru==2.3.0
+asttokens==3.0.1
+aiohappyeyeballs==2.6.1
+yarl==1.23.0
+terminado==0.18.1
+stack-data==0.6.3
+simple-parsing==0.1.8
+sentry-sdk==2.58.0
+scipy==1.17.1
+rfc3987-syntax==1.1.0
+rfc3339-validator==0.1.4
+requests==2.32.4
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pytest==8.4.1
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+plotly==6.3.0
+patsy==1.0.2
+nvidia-cusparse-cu12==12.5.8.93
+nvidia-cudnn-cu12==9.19.0.56
+torch==2.11.0+cu128
+nltk==3.9.4
+multiprocess==0.70.16
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jupyter_core==5.9.1
+jedi==0.20.0
+jaxtyping==0.3.2
+ipython_pygments_lexers==1.1.1
+importlib_metadata==9.0.0
+httpcore==1.0.9
+gitdb==4.0.12
+contourpy==1.3.3
+cffi==2.0.0
+beautifulsoup4==4.14.3
+anyio==4.13.0
+aiosignal==1.4.0
+tiktoken==0.12.0
+scikit-learn==1.7.1
+rich==15.0.0
+py2store==0.1.22
+pandas==2.3.1
+nvidia-cusolver-cu12==11.7.3.90
+matplotlib==3.10.9
+jupyter_server_terminals==0.5.4
+jupyter_client==8.8.0
+jsonschema-specifications==2025.9.1
+ipython==9.4.0
+huggingface_hub==0.36.2
+httpx==0.28.1
+graze==0.1.39
+GitPython==3.1.49
+arrow==1.4.0
+argon2-cffi-bindings==25.1.0
+aiohttp==3.13.5
+wandb==0.21.0
+sae-scoping==2020.0.0
+statsmodels==0.14.6
+seaborn==0.13.2
+openai==2.33.0
+jsonschema==4.26.0
+isoduration==20.11.0
+ipykernel==7.2.0
+babe==0.0.7
+argon2-cffi==25.1.0
+transformers==4.56.1
+schedulefree==1.4.1
+plotly-express==0.4.1
+nbformat==5.10.4
+litellm==1.74.7
+datasets==4.0.0
+accelerate==1.13.0
+trl==0.22.2
+transformers-stream-generator==0.0.5
+peft==0.16.0
+nbclient==0.10.4
+jupyter-events==0.12.1
+eai-sparsify==1.3.0
+transformer-lens==2.17.0
+nbconvert==7.17.1
+sae-lens==6.43.0
+jupyter_server==2.17.0
+notebook_shim==0.2.4
+jupyterlab_server==2.28.0
+jupyter-lsp==2.3.1
+jupyterlab==4.4.10
+notebook==7.4.4
+triton==3.6.0
+cuda-bindings==12.9.4
+nvidia-cufft-cu12==11.3.3.83

run-20260503_070020-tsnlx6qq/files/wandb-metadata.json

@@ -0,0 +1,61 @@
+{
+  "os":  "Linux-6.8.0-106-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.12.13",
+  "startedAt":  "2026-05-03T07:00:20.452873Z",
+  "args":  [
+    "--train-domain",
+    "math",
+    "--attack-domain",
+    "physics",
+    "--stage",
+    "attack",
+    "--hf-recover-repo",
+    "arunasank/fitoo774",
+    "--checkpoint",
+    "2000",
+    "--dev",
+    "--device",
+    "cuda",
+    "--firing-rate-threshold",
+    "1e-4",
+    "--gemma3",
+    "--max-steps-attack",
+    "10000",
+    "--skip-pre-training-eval"
+  ],
+  "program":  "/root/SAEScoping/experiments/script_scoping_pipeline_stemqa.py",
+  "codePath":  "experiments/script_scoping_pipeline_stemqa.py",
+  "codePathLocal":  "experiments/script_scoping_pipeline_stemqa.py",
+  "git":  {
+    "remote":  "git@github.com:4gatepylon/SAEScoping.git",
+    "commit":  "89bc64f41f54ca1f08c9d90c5c2ea13f98123c1b"
+  },
+  "email":  "aruna.evam@gmail.com",
+  "root":  "/root/SAEScoping",
+  "host":  "80e02dee87f9",
+  "executable":  "/root/miniconda3/envs/sae/bin/python",
+  "cpu_count":  64,
+  "cpu_count_logical":  128,
+  "gpu":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+  "gpu_count":  1,
+  "disk":  {
+    "/":  {
+      "total":  "536870912000",
+      "used":  "73938042880"
+    }
+  },
+  "memory":  {
+    "total":  "1622884343808"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA RTX PRO 6000 Blackwell Server Edition",
+      "memoryTotal":  "102641958912",
+      "cudaCores":  24064,
+      "architecture":  "Blackwell",
+      "uuid":  "GPU-1e3dcc13-cf37-b8fe-2425-456c8d1b35d2"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "kxtjeo9jni30ggc0c02r09gslwxc3kka"
+}

run-20260503_070020-tsnlx6qq/files/wandb-summary.json

@@ -0,0 +1 @@
+{"true_baseline/llm_judge/chemistry/out_of_scope/ground_truth_similarity":0.845,"true_baseline/llm_judge/physics/out_of_scope/fluency":0.965,"true_baseline/llm_judge/math/in_scope/fluency":0.96,"_runtime":2,"true_baseline/llm_judge/physics/out_of_scope/relevance":0.97,"true_baseline/llm_judge/math/in_scope/quality":0.93,"llm_judge/math/in_scope/quality_pre_scoping":0.93,"llm_judge/math/in_scope/ground_truth_similarity_pre_scoping":0.835,"llm_judge/physics/out_of_scope/fluency_pre_scoping":0.965,"llm_judge/biology/out_of_scope/relevance_pre_scoping":1,"true_baseline/llm_judge/math/in_scope/ground_truth_similarity":0.835,"true_baseline/llm_judge/chemistry/out_of_scope/fluency":0.975,"trainer/global_step":0,"true_baseline/llm_judge/biology/out_of_scope/relevance":1,"llm_judge/chemistry/out_of_scope/ground_truth_similarity_pre_scoping":0.845,"llm_judge/physics/out_of_scope/ground_truth_similarity_pre_scoping":0.67,"true_baseline/llm_judge/biology/out_of_scope/fluency":0.995,"llm_judge/math/in_scope/fluency_pre_scoping":0.96,"llm_judge/chemistry/out_of_scope/quality_pre_scoping":0.9366666666666666,"_step":1,"true_baseline/llm_judge/biology/out_of_scope/quality":0.9383333333333334,"llm_judge/biology/out_of_scope/quality_pre_scoping":0.9383333333333334,"true_baseline/llm_judge/chemistry/out_of_scope/relevance":0.99,"llm_judge/chemistry/out_of_scope/relevance_pre_scoping":0.99,"true_baseline/llm_judge/biology/out_of_scope/ground_truth_similarity":0.82,"llm_judge/physics/out_of_scope/relevance_pre_scoping":0.97,"llm_judge/physics/out_of_scope/quality_pre_scoping":0.8683333333333333,"true_baseline/llm_judge/chemistry/out_of_scope/quality":0.9366666666666666,"llm_judge/biology/out_of_scope/ground_truth_similarity_pre_scoping":0.82,"true_baseline/llm_judge/physics/out_of_scope/ground_truth_similarity":0.67,"llm_judge/chemistry/out_of_scope/fluency_pre_scoping":0.975,"_wandb":{"runtime":2},"llm_judge/biology/out_of_scope/fluency_pre_scoping":0.995,"llm_judge/math/in_scope/relevance_pre_scoping":0.995,"true_baseline/llm_judge/physics/out_of_scope/quality":0.8683333333333333,"true_baseline/llm_judge/math/in_scope/relevance":0.995,"_timestamp":1.7777916212437994e+09}

run-20260503_070020-tsnlx6qq/logs/debug-core.log

@@ -0,0 +1,18 @@
+{"time":"2026-05-03T07:00:20.469220164Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp_j91eab8/port-10341.txt","pid":10341,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-05-03T07:00:20.46970843Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":10341}
+{"time":"2026-05-03T07:00:20.469687729Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-10341-15586-3861218357/socket","Net":"unix"}}
+{"time":"2026-05-03T07:00:20.657887643Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-05-03T07:00:20.660790402Z","level":"INFO","msg":"handleInformInit: received","streamId":"tsnlx6qq","id":"1(@)"}
+{"time":"2026-05-03T07:00:20.848571595Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"tsnlx6qq","id":"1(@)"}
+{"time":"2026-05-03T07:00:23.783791062Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"tsnlx6qq","id":"1(@)"}
+{"time":"2026-05-03T07:00:23.785249021Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"tsnlx6qq","id":"1(@)"}
+{"time":"2026-05-03T07:00:23.785623376Z","level":"INFO","msg":"handleInformInit: received","streamId":"mw4gx9uu","id":"1(@)"}
+{"time":"2026-05-03T07:00:23.958283535Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"mw4gx9uu","id":"1(@)"}
+{"time":"2026-05-04T20:47:39.170132472Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2026-05-04T20:47:39.170220385Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-05-04T20:47:39.170208517Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2026-05-04T20:47:39.170282418Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2026-05-04T20:47:39.170328918Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-10341-15586-3861218357/socket","Net":"unix"}}
+{"time":"2026-05-04T20:47:39.568705145Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2026-05-04T20:47:39.568786097Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2026-05-04T20:47:39.568805747Z","level":"INFO","msg":"server is closed"}

run-20260503_070020-tsnlx6qq/logs/debug-internal.log

@@ -0,0 +1,13 @@
+{"time":"2026-05-03T07:00:20.660976072Z","level":"INFO","msg":"stream: starting","core version":"0.21.0"}
+{"time":"2026-05-03T07:00:20.848539286Z","level":"INFO","msg":"stream: created new stream","id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:20.848568149Z","level":"INFO","msg":"stream: started","id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:20.84858856Z","level":"INFO","msg":"writer: Do: started","stream_id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:20.8486014Z","level":"INFO","msg":"handler: started","stream_id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:20.848590483Z","level":"INFO","msg":"sender: started","stream_id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:23.585072348Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-05-03T07:00:23.782187102Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-05-03T07:00:23.783806655Z","level":"INFO","msg":"stream: closing","id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:23.783814807Z","level":"INFO","msg":"handler: closed","stream_id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:23.783820416Z","level":"INFO","msg":"writer: Close: closed","stream_id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:23.783824061Z","level":"INFO","msg":"sender: closed","stream_id":"tsnlx6qq"}
+{"time":"2026-05-03T07:00:23.783862599Z","level":"INFO","msg":"stream: closed","id":"tsnlx6qq"}

run-20260503_070020-tsnlx6qq/logs/debug.log

@@ -0,0 +1,28 @@
+2026-05-03 07:00:20,453 INFO    MainThread:10341 [wandb_setup.py:_flush():80] Current SDK version is 0.21.0
+2026-05-03 07:00:20,453 INFO    MainThread:10341 [wandb_setup.py:_flush():80] Configure stats pid to 10341
+2026-05-03 07:00:20,453 INFO    MainThread:10341 [wandb_setup.py:_flush():80] Loading settings from /root/.config/wandb/settings
+2026-05-03 07:00:20,453 INFO    MainThread:10341 [wandb_setup.py:_flush():80] Loading settings from /root/SAEScoping/wandb/settings
+2026-05-03 07:00:20,453 INFO    MainThread:10341 [wandb_setup.py:_flush():80] Loading settings from environment variables
+2026-05-03 07:00:20,454 INFO    MainThread:10341 [wandb_init.py:setup_run_log_directory():703] Logging user logs to /root/SAEScoping/wandb/run-20260503_070020-tsnlx6qq/logs/debug.log
+2026-05-03 07:00:20,454 INFO    MainThread:10341 [wandb_init.py:setup_run_log_directory():704] Logging internal logs to /root/SAEScoping/wandb/run-20260503_070020-tsnlx6qq/logs/debug-internal.log
+2026-05-03 07:00:20,454 INFO    MainThread:10341 [wandb_init.py:init():830] calling init triggers
+2026-05-03 07:00:20,454 INFO    MainThread:10341 [wandb_init.py:init():835] wandb.init called with sweep_config: {}
+config: {'_wandb': {}}
+2026-05-03 07:00:20,454 INFO    MainThread:10341 [wandb_init.py:init():871] starting backend
+2026-05-03 07:00:20,658 INFO    MainThread:10341 [wandb_init.py:init():874] sending inform_init request
+2026-05-03 07:00:20,659 INFO    MainThread:10341 [wandb_init.py:init():882] backend started and connected
+2026-05-03 07:00:20,660 INFO    MainThread:10341 [wandb_init.py:init():953] updated telemetry
+2026-05-03 07:00:20,663 INFO    MainThread:10341 [wandb_init.py:init():977] communicating run to backend with 180.0 second timeout
+2026-05-03 07:00:21,168 INFO    MainThread:10341 [wandb_init.py:init():1029] starting run threads in backend
+2026-05-03 07:00:21,242 INFO    MainThread:10341 [wandb_run.py:_console_start():2458] atexit reg
+2026-05-03 07:00:21,242 INFO    MainThread:10341 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2026-05-03 07:00:21,242 INFO    MainThread:10341 [wandb_run.py:_redirect():2375] Wrapping output streams.
+2026-05-03 07:00:21,242 INFO    MainThread:10341 [wandb_run.py:_redirect():2398] Redirects installed.
+2026-05-03 07:00:21,243 INFO    MainThread:10341 [wandb_init.py:init():1075] run started, returning control to user process
+2026-05-03 07:00:23,258 INFO    MainThread:10341 [wandb_run.py:_finish():2224] finishing run arunasank/sae-scoping-stemqa-math/tsnlx6qq
+2026-05-03 07:00:23,258 INFO    MainThread:10341 [wandb_run.py:_atexit_cleanup():2423] got exitcode: 0
+2026-05-03 07:00:23,258 INFO    MainThread:10341 [wandb_run.py:_restore():2405] restore
+2026-05-03 07:00:23,258 INFO    MainThread:10341 [wandb_run.py:_restore():2411] restore done
+2026-05-03 07:00:23,783 INFO    MainThread:10341 [wandb_run.py:_footer_history_summary_info():3903] rendering history
+2026-05-03 07:00:23,783 INFO    MainThread:10341 [wandb_run.py:_footer_history_summary_info():3935] rendering summary
+2026-05-03 07:00:23,783 INFO    MainThread:10341 [wandb_run.py:_footer_sync_info():3864] logging synced files

run-20260503_070023-mw4gx9uu/files/config.yaml

@@ -0,0 +1,988 @@
+_name_or_path:
+    value: /workspace/.cache/huggingface/hub/models--arunasank--fitoo774/snapshots/fadf3c3034b2218f784375223d93790921fec9ff/checkpoint-2000
+_wandb:
+    value:
+        cli_version: 0.21.0
+        e:
+            k9ay0f3la4y2006c5cqjs8lajyxtc4og:
+                args:
+                    - --train-domain
+                    - math
+                    - --attack-domain
+                    - physics
+                    - --stage
+                    - attack
+                    - --hf-recover-repo
+                    - arunasank/fitoo774
+                    - --checkpoint
+                    - "2000"
+                    - --dev
+                    - --device
+                    - cuda
+                    - --firing-rate-threshold
+                    - "1e-4"
+                    - --gemma3
+                    - --max-steps-attack
+                    - "10000"
+                    - --skip-pre-training-eval
+                codePath: experiments/script_scoping_pipeline_stemqa.py
+                codePathLocal: experiments/script_scoping_pipeline_stemqa.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "536870912000"
+                        used: "73938083840"
+                email: aruna.evam@gmail.com
+                executable: /root/miniconda3/envs/sae/bin/python
+                git:
+                    commit: 89bc64f41f54ca1f08c9d90c5c2ea13f98123c1b
+                    remote: git@github.com:4gatepylon/SAEScoping.git
+                gpu: NVIDIA RTX PRO 6000 Blackwell Server Edition
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Blackwell
+                      cudaCores: 24064
+                      memoryTotal: "102641958912"
+                      name: NVIDIA RTX PRO 6000 Blackwell Server Edition
+                      uuid: GPU-1e3dcc13-cf37-b8fe-2425-456c8d1b35d2
+                host: 80e02dee87f9
+                memory:
+                    total: "1622884343808"
+                os: Linux-6.8.0-106-generic-x86_64-with-glibc2.39
+                program: /root/SAEScoping/experiments/script_scoping_pipeline_stemqa.py
+                python: CPython 3.12.13
+                root: /root/SAEScoping
+                startedAt: "2026-05-03T07:00:23.784582Z"
+                writerId: k9ay0f3la4y2006c5cqjs8lajyxtc4og
+        m:
+            - "1": train/global_step
+              "6":
+                - 3
+              "7": []
+            - "2": '*'
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.12.13
+        t:
+            "1":
+                - 1
+                - 5
+                - 11
+                - 49
+                - 51
+                - 53
+                - 71
+                - 84
+                - 95
+                - 98
+                - 105
+            "2":
+                - 1
+                - 5
+                - 11
+                - 49
+                - 51
+                - 53
+                - 71
+                - 84
+                - 95
+                - 98
+                - 105
+            "3":
+                - 7
+                - 13
+                - 14
+                - 19
+                - 62
+                - 66
+            "4": 3.12.13
+            "5": 0.21.0
+            "6": 4.56.1
+            "9":
+                "1": transformers_trainer
+            "12": 0.21.0
+            "13": linux-x86_64
+        visualize:
+            charts/llm_judge_diff_baseline_fluency:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge diff vs baseline: fluency'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_diff_baseline_fluency_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_diff_baseline_ground_truth_similarity:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge diff vs baseline: ground_truth_similarity'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_diff_baseline_ground_truth_similarity_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_diff_baseline_quality:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge diff vs baseline: quality'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_diff_baseline_quality_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_diff_baseline_relevance:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge diff vs baseline: relevance'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_diff_baseline_relevance_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_fluency:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge: fluency'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_fluency_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_ground_truth_similarity:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge: ground_truth_similarity'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_ground_truth_similarity_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_quality:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge: quality'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_quality_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+            charts/llm_judge_relevance:
+                panel_config:
+                    fieldSettings:
+                        lineKey: lineKey
+                        lineVal: lineVal
+                        step: step
+                    panelDefId: wandb/lineseries/v0
+                    stringSettings:
+                        title: 'LLM Judge: relevance'
+                        xname: Training Step
+                    transform:
+                        name: tableWithLeafColNames
+                    userQuery:
+                        queryFields:
+                            - args:
+                                - name: runSets
+                                  value: ${runSets}
+                              fields:
+                                - fields: []
+                                  name: id
+                                - fields: []
+                                  name: name
+                                - fields: []
+                                  name: _defaultColorIndex
+                                - args:
+                                    - name: tableKey
+                                      value: charts/llm_judge_relevance_table
+                                  fields: []
+                                  name: summaryTable
+                              name: runSets
+                panel_type: Vega2
+accelerator_config:
+    value:
+        dispatch_batches: null
+        even_batches: true
+        gradient_accumulation_kwargs: null
+        non_blocking: false
+        split_batches: false
+        use_seedable_sampler: true
+activation_offloading:
+    value: false
+adafactor:
+    value: false
+adam_beta1:
+    value: 0.9
+adam_beta2:
+    value: 0.999
+adam_epsilon:
+    value: 1e-08
+add_cross_attention:
+    value: false
+architectures:
+    value:
+        - Gemma3ForConditionalGeneration
+assistant_only_loss:
+    value: false
+auto_find_batch_size:
+    value: false
+average_tokens_across_devices:
+    value: false
+bad_words_ids:
+    value: null
+batch_eval_metrics:
+    value: false
+begin_suppress_tokens:
+    value: null
+bf16:
+    value: true
+bf16_full_eval:
+    value: false
+boi_token_index:
+    value: 255999
+bos_token_id:
+    value: 2
+chat_template_path:
+    value: null
+chunk_size_feed_forward:
+    value: 0
+completion_only_loss:
+    value: null
+cross_attention_hidden_size:
+    value: null
+data_seed:
+    value: null
+dataloader_drop_last:
+    value: false
+dataloader_num_workers:
+    value: 0
+dataloader_persistent_workers:
+    value: false
+dataloader_pin_memory:
+    value: true
+dataloader_prefetch_factor:
+    value: null
+dataset_kwargs:
+    value: null
+dataset_num_proc:
+    value: null
+dataset_text_field:
+    value: text
+ddp_backend:
+    value: null
+ddp_broadcast_buffers:
+    value: null
+ddp_bucket_cap_mb:
+    value: null
+ddp_find_unused_parameters:
+    value: null
+ddp_timeout:
+    value: 1800
+debug:
+    value: []
+decoder_start_token_id:
+    value: null
+deepspeed:
+    value: null
+disable_tqdm:
+    value: false
+diversity_penalty:
+    value: 0
+do_eval:
+    value: true
+do_predict:
+    value: false
+do_sample:
+    value: false
+do_train:
+    value: false
+dtype:
+    value: bfloat16
+early_stopping:
+    value: false
+encoder_no_repeat_ngram_size:
+    value: 0
+eoi_token_index:
+    value: 256000
+eos_token:
+    value: <EOS_TOKEN>
+eos_token_id:
+    value: 1
+eval_accumulation_steps:
+    value: 16
+eval_delay:
+    value: 0
+eval_do_concat_batches:
+    value: true
+eval_on_start:
+    value: false
+eval_packing:
+    value: null
+eval_steps:
+    value: 100
+eval_strategy:
+    value: steps
+eval_use_gather_object:
+    value: false
+exponential_decay_length_penalty:
+    value: null
+finetuning_task:
+    value: null
+forced_bos_token_id:
+    value: null
+forced_eos_token_id:
+    value: null
+fp16:
+    value: false
+fp16_backend:
+    value: auto
+fp16_full_eval:
+    value: false
+fp16_opt_level:
+    value: O1
+fsdp:
+    value: []
+fsdp_config:
+    value:
+        min_num_params: 0
+        xla: false
+        xla_fsdp_grad_ckpt: false
+        xla_fsdp_v2: false
+fsdp_min_num_params:
+    value: 0
+fsdp_transformer_layer_cls_to_wrap:
+    value: null
+full_determinism:
+    value: false
+gradient_accumulation_steps:
+    value: 16
+gradient_checkpointing:
+    value: true
+gradient_checkpointing_kwargs:
+    value:
+        use_reentrant: false
+greater_is_better:
+    value: null
+group_by_length:
+    value: false
+half_precision_backend:
+    value: auto
+hub_always_push:
+    value: false
+hub_model_id:
+    value: null
+hub_private_repo:
+    value: null
+hub_revision:
+    value: null
+hub_strategy:
+    value: every_save
+hub_token:
+    value: <HUB_TOKEN>
+id2label:
+    value:
+        "0": LABEL_0
+        "1": LABEL_1
+ignore_data_skip:
+    value: false
+image_token_index:
+    value: 262144
+include_for_metrics:
+    value: []
+include_inputs_for_metrics:
+    value: false
+include_num_input_tokens_seen:
+    value: false
+include_tokens_per_second:
+    value: false
+initializer_range:
+    value: 0.02
+is_decoder:
+    value: false
+is_encoder_decoder:
+    value: false
+jit_mode_eval:
+    value: false
+label_names:
+    value: null
+label_smoothing_factor:
+    value: 0
+label2id:
+    value:
+        LABEL_0: 0
+        LABEL_1: 1
+learning_rate:
+    value: 2e-05
+length_column_name:
+    value: length
+length_penalty:
+    value: 1
+liger_kernel_config:
+    value: null
+load_best_model_at_end:
+    value: false
+local_rank:
+    value: 0
+log_level:
+    value: passive
+log_level_replica:
+    value: warning
+log_on_each_node:
+    value: true
+logging_dir:
+    value: /root/SAEScoping/experiments/outputs_scoping/google--gemma-3-12b-it/layer_31--width_16k--canonical/math/h0.0001/k1684/mv6qs85r/attack/physics/mw4gx9uu/runs/May03_07-00-24_80e02dee87f9
+logging_first_step:
+    value: false
+logging_nan_inf_filter:
+    value: true
+logging_steps:
+    value: 10
+logging_strategy:
+    value: steps
+lr_scheduler_type:
+    value: linear
+max_grad_norm:
+    value: 1
+max_length:
+    value: 1024
+max_steps:
+    value: 10000
+metric_for_best_model:
+    value: null
+min_length:
+    value: 0
+mm_tokens_per_image:
+    value: 256
+model/num_parameters:
+    value: 12187325040
+model_init_kwargs:
+    value: null
+model_type:
+    value: gemma3
+mp_parameters:
+    value: ""
+neftune_noise_alpha:
+    value: null
+no_cuda:
+    value: false
+no_repeat_ngram_size:
+    value: 0
+num_beam_groups:
+    value: 1
+num_beams:
+    value: 1
+num_return_sequences:
+    value: 1
+num_train_epochs:
+    value: 1
+optim:
+    value: paged_adamw_8bit
+optim_args:
+    value: null
+optim_target_modules:
+    value: null
+output_attentions:
+    value: false
+output_dir:
+    value: /root/SAEScoping/experiments/outputs_scoping/google--gemma-3-12b-it/layer_31--width_16k--canonical/math/h0.0001/k1684/mv6qs85r/attack/physics/mw4gx9uu
+output_hidden_states:
+    value: false
+output_scores:
+    value: false
+overwrite_output_dir:
+    value: false
+packing:
+    value: false
+packing_strategy:
+    value: bfd
+pad_to_multiple_of:
+    value: null
+pad_token:
+    value: <PAD_TOKEN>
+pad_token_id:
+    value: 0
+padding_free:
+    value: false
+parallelism_config:
+    value: null
+past_index:
+    value: -1
+per_device_eval_batch_size:
+    value: 4
+per_device_train_batch_size:
+    value: 4
+per_gpu_eval_batch_size:
+    value: null
+per_gpu_train_batch_size:
+    value: null
+prediction_loss_only:
+    value: false
+prefix:
+    value: null
+problem_type:
+    value: null
+push_to_hub:
+    value: false
+push_to_hub_model_id:
+    value: null
+push_to_hub_organization:
+    value: null
+push_to_hub_token:
+    value: <PUSH_TO_HUB_TOKEN>
+ray_scope:
+    value: last
+remove_invalid_values:
+    value: false
+remove_unused_columns:
+    value: true
+repetition_penalty:
+    value: 1
+report_to:
+    value:
+        - wandb
+restore_callback_states_from_checkpoint:
+    value: false
+resume_from_checkpoint:
+    value: true
+return_dict:
+    value: true
+return_dict_in_generate:
+    value: false
+run_name:
+    value: null
+save_on_each_node:
+    value: false
+save_only_model:
+    value: false
+save_safetensors:
+    value: true
+save_steps:
+    value: 500
+save_strategy:
+    value: steps
+save_total_limit:
+    value: 5
+seed:
+    value: 42
+sep_token_id:
+    value: null
+skip_memory_metrics:
+    value: true
+suppress_tokens:
+    value: null
+task_specific_params:
+    value: null
+temperature:
+    value: 1
+text_config:
+    value:
+        _name_or_path: ""
+        _sliding_window_pattern: 6
+        add_cross_attention: false
+        architectures: null
+        attention_bias: false
+        attention_dropout: 0
+        attn_logit_softcapping: null
+        bad_words_ids: null
+        begin_suppress_tokens: null
+        bos_token_id: 2
+        chunk_size_feed_forward: 0
+        cross_attention_hidden_size: null
+        decoder_start_token_id: null
+        diversity_penalty: 0
+        do_sample: false
+        dtype: bfloat16
+        early_stopping: false
+        encoder_no_repeat_ngram_size: 0
+        eos_token_id: 1
+        exponential_decay_length_penalty: null
+        final_logit_softcapping: null
+        finetuning_task: null
+        forced_bos_token_id: null
+        forced_eos_token_id: null
+        head_dim: 256
+        hidden_activation: gelu_pytorch_tanh
+        hidden_size: 3840
+        id2label:
+            "0": LABEL_0
+            "1": LABEL_1
+        initializer_range: 0.02
+        intermediate_size: 15360
+        is_decoder: false
+        is_encoder_decoder: false
+        label2id:
+            LABEL_0: 0
+            LABEL_1: 1
+        layer_types:
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - sliding_attention
+            - full_attention
+        length_penalty: 1
+        max_length: 20
+        max_position_embeddings: 131072
+        min_length: 0
+        model_type: gemma3_text
+        no_repeat_ngram_size: 0
+        num_attention_heads: 16
+        num_beam_groups: 1
+        num_beams: 1
+        num_hidden_layers: 48
+        num_key_value_heads: 8
+        num_return_sequences: 1
+        output_attentions: false
+        output_hidden_states: false
+        output_scores: false
+        pad_token_id: 0
+        prefix: null
+        problem_type: null
+        query_pre_attn_scalar: 256
+        remove_invalid_values: false
+        repetition_penalty: 1
+        return_dict: true
+        return_dict_in_generate: false
+        rms_norm_eps: 1e-06
+        rope_local_base_freq: 10000
+        rope_scaling:
+            factor: 8
+            rope_type: linear
+        rope_theta: 1e+06
+        sep_token_id: null
+        sliding_window: 1024
+        suppress_tokens: null
+        task_specific_params: null
+        temperature: 1
+        tf_legacy_loss: false
+        tie_encoder_decoder: false
+        tie_word_embeddings: true
+        tokenizer_class: null
+        top_k: 50
+        top_p: 1
+        torchscript: false
+        typical_p: 1
+        use_bfloat16: false
+        use_cache: true
+        vocab_size: 262208
+tf_legacy_loss:
+    value: false
+tf32:
+    value: null
+tie_encoder_decoder:
+    value: false
+tie_word_embeddings:
+    value: true
+tokenizer_class:
+    value: null
+top_k:
+    value: 50
+top_p:
+    value: 1
+torch_compile:
+    value: false
+torch_compile_backend:
+    value: null
+torch_compile_mode:
+    value: null
+torch_empty_cache_steps:
+    value: null
+torchdynamo:
+    value: null
+torchscript:
+    value: false
+tpu_metrics_debug:
+    value: false
+tpu_num_cores:
+    value: null
+transformers_version:
+    value: 4.56.1
+typical_p:
+    value: 1
+use_bfloat16:
+    value: false
+use_cpu:
+    value: false
+use_ipex:
+    value: false
+use_legacy_prediction_loop:
+    value: false
+use_liger_kernel:
+    value: false
+use_mps_device:
+    value: false
+vision_config:
+    value:
+        _name_or_path: ""
+        add_cross_attention: false
+        architectures: null
+        attention_dropout: 0
+        bad_words_ids: null
+        begin_suppress_tokens: null
+        bos_token_id: null
+        chunk_size_feed_forward: 0
+        cross_attention_hidden_size: null
+        decoder_start_token_id: null
+        diversity_penalty: 0
+        do_sample: false
+        dtype: bfloat16
+        early_stopping: false
+        encoder_no_repeat_ngram_size: 0
+        eos_token_id: null
+        exponential_decay_length_penalty: null
+        finetuning_task: null
+        forced_bos_token_id: null
+        forced_eos_token_id: null
+        hidden_act: gelu_pytorch_tanh
+        hidden_size: 1152
+        id2label:
+            "0": LABEL_0
+            "1": LABEL_1
+        image_size: 896
+        intermediate_size: 4304
+        is_decoder: false
+        is_encoder_decoder: false
+        label2id:
+            LABEL_0: 0
+            LABEL_1: 1
+        layer_norm_eps: 1e-06
+        length_penalty: 1
+        max_length: 20
+        min_length: 0
+        model_type: siglip_vision_model
+        no_repeat_ngram_size: 0
+        num_attention_heads: 16
+        num_beam_groups: 1
+        num_beams: 1
+        num_channels: 3
+        num_hidden_layers: 27
+        num_return_sequences: 1
+        output_attentions: false
+        output_hidden_states: false
+        output_scores: false
+        pad_token_id: null
+        patch_size: 14
+        prefix: null
+        problem_type: null
+        remove_invalid_values: false
+        repetition_penalty: 1
+        return_dict: true
+        return_dict_in_generate: false
+        sep_token_id: null
+        suppress_tokens: null
+        task_specific_params: null
+        temperature: 1
+        tf_legacy_loss: false
+        tie_encoder_decoder: false
+        tie_word_embeddings: true
+        tokenizer_class: null
+        top_k: 50
+        top_p: 1
+        torchscript: false
+        typical_p: 1
+        use_bfloat16: false
+        vision_use_head: false
+warmup_ratio:
+    value: 0.1
+warmup_steps:
+    value: 0
+weight_decay:
+    value: 0.1

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_491_084cda5c92c1f5fd2472.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_562_108153824aefe6a35fdd.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [4000, "math/in_scope", -0.15999999999999992], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245], [4000, "physics/attack_scope", -0.21999999999999997]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_633_e4d7691c028a3169be82.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [4000, "math/in_scope", -0.15999999999999992], [4500, "math/in_scope", -0.18999999999999995], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245], [4000, "physics/attack_scope", -0.21999999999999997], [4500, "physics/attack_scope", -0.245]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_704_0336dd8b3d6193775ae7.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [4000, "math/in_scope", -0.15999999999999992], [4500, "math/in_scope", -0.18999999999999995], [5000, "math/in_scope", -0.20499999999999996], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245], [4000, "physics/attack_scope", -0.21999999999999997], [4500, "physics/attack_scope", -0.245], [5000, "physics/attack_scope", -0.20999999999999996]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_775_addf4a7b0a6d8223c671.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [4000, "math/in_scope", -0.15999999999999992], [4500, "math/in_scope", -0.18999999999999995], [5000, "math/in_scope", -0.20499999999999996], [5500, "math/in_scope", -0.1499999999999999], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245], [4000, "physics/attack_scope", -0.21999999999999997], [4500, "physics/attack_scope", -0.245], [5000, "physics/attack_scope", -0.20999999999999996], [5500, "physics/attack_scope", -0.19499999999999995]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_846_d7d02a8f538364d9f02a.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [4000, "math/in_scope", -0.15999999999999992], [4500, "math/in_scope", -0.18999999999999995], [5000, "math/in_scope", -0.20499999999999996], [5500, "math/in_scope", -0.1499999999999999], [6000, "math/in_scope", -0.235], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245], [4000, "physics/attack_scope", -0.21999999999999997], [4500, "physics/attack_scope", -0.245], [5000, "physics/attack_scope", -0.20999999999999996], [5500, "physics/attack_scope", -0.19499999999999995], [6000, "physics/attack_scope", -0.20499999999999996]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_fluency_table_917_a4f00630ffe30bcfe311.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.07499999999999996], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.08499999999999996], [2000, "math/in_scope", -0.13], [2500, "math/in_scope", -0.09999999999999998], [3000, "math/in_scope", -0.20999999999999996], [3500, "math/in_scope", -0.19499999999999995], [4000, "math/in_scope", -0.15999999999999992], [4500, "math/in_scope", -0.18999999999999995], [5000, "math/in_scope", -0.20499999999999996], [5500, "math/in_scope", -0.1499999999999999], [6000, "math/in_scope", -0.235], [6500, "math/in_scope", -0.20499999999999996], [500, "physics/attack_scope", -0.45499999999999996], [1000, "physics/attack_scope", -0.25], [1500, "physics/attack_scope", -0.21999999999999997], [2000, "physics/attack_scope", -0.11499999999999999], [2500, "physics/attack_scope", -0.12], [3000, "physics/attack_scope", -0.15000000000000002], [3500, "physics/attack_scope", -0.245], [4000, "physics/attack_scope", -0.21999999999999997], [4500, "physics/attack_scope", -0.245], [5000, "physics/attack_scope", -0.20999999999999996], [5500, "physics/attack_scope", -0.19499999999999995], [6000, "physics/attack_scope", -0.20499999999999996], [6500, "physics/attack_scope", -0.18999999999999995]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_1060_ce30e646b2fea156817c.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [4000, "math/in_scope", -0.20499999999999996], [4500, "math/in_scope", -0.22499999999999998], [5000, "math/in_scope", -0.2899999999999999], [5500, "math/in_scope", -0.22999999999999998], [6000, "math/in_scope", -0.22999999999999998], [6500, "math/in_scope", -0.24], [7000, "math/in_scope", -0.235], [7500, "math/in_scope", -0.22999999999999998], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275], [4000, "physics/attack_scope", -0.29000000000000004], [4500, "physics/attack_scope", -0.265], [5000, "physics/attack_scope", -0.24000000000000005], [5500, "physics/attack_scope", -0.24000000000000005], [6000, "physics/attack_scope", -0.29500000000000004], [6500, "physics/attack_scope", -0.21000000000000002], [7000, "physics/attack_scope", -0.30500000000000005], [7500, "physics/attack_scope", -0.24000000000000005]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_492_a5bd2a0cd4f63e67f54f.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_634_45d88133b373103983fc.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [4000, "math/in_scope", -0.20499999999999996], [4500, "math/in_scope", -0.22499999999999998], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275], [4000, "physics/attack_scope", -0.29000000000000004], [4500, "physics/attack_scope", -0.265]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_705_28e93bfb4e3763324925.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [4000, "math/in_scope", -0.20499999999999996], [4500, "math/in_scope", -0.22499999999999998], [5000, "math/in_scope", -0.2899999999999999], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275], [4000, "physics/attack_scope", -0.29000000000000004], [4500, "physics/attack_scope", -0.265], [5000, "physics/attack_scope", -0.24000000000000005]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_776_45c8a9e1988a9e5162c8.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [4000, "math/in_scope", -0.20499999999999996], [4500, "math/in_scope", -0.22499999999999998], [5000, "math/in_scope", -0.2899999999999999], [5500, "math/in_scope", -0.22999999999999998], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275], [4000, "physics/attack_scope", -0.29000000000000004], [4500, "physics/attack_scope", -0.265], [5000, "physics/attack_scope", -0.24000000000000005], [5500, "physics/attack_scope", -0.24000000000000005]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_847_b866a7ce448652f4313e.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [4000, "math/in_scope", -0.20499999999999996], [4500, "math/in_scope", -0.22499999999999998], [5000, "math/in_scope", -0.2899999999999999], [5500, "math/in_scope", -0.22999999999999998], [6000, "math/in_scope", -0.22999999999999998], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275], [4000, "physics/attack_scope", -0.29000000000000004], [4500, "physics/attack_scope", -0.265], [5000, "physics/attack_scope", -0.24000000000000005], [5500, "physics/attack_scope", -0.24000000000000005], [6000, "physics/attack_scope", -0.29500000000000004]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_ground_truth_similarity_table_918_aae2930a5651916a7b9a.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.1499999999999999], [1000, "math/in_scope", -0.14500000000000002], [1500, "math/in_scope", -0.125], [2000, "math/in_scope", -0.20499999999999996], [2500, "math/in_scope", -0.125], [3000, "math/in_scope", -0.21499999999999997], [3500, "math/in_scope", -0.19999999999999996], [4000, "math/in_scope", -0.20499999999999996], [4500, "math/in_scope", -0.22499999999999998], [5000, "math/in_scope", -0.2899999999999999], [5500, "math/in_scope", -0.22999999999999998], [6000, "math/in_scope", -0.22999999999999998], [6500, "math/in_scope", -0.24], [500, "physics/attack_scope", -0.42500000000000004], [1000, "physics/attack_scope", -0.33], [1500, "physics/attack_scope", -0.28], [2000, "physics/attack_scope", -0.275], [2500, "physics/attack_scope", -0.23000000000000004], [3000, "physics/attack_scope", -0.24000000000000005], [3500, "physics/attack_scope", -0.275], [4000, "physics/attack_scope", -0.29000000000000004], [4500, "physics/attack_scope", -0.265], [5000, "physics/attack_scope", -0.24000000000000005], [5500, "physics/attack_scope", -0.24000000000000005], [6000, "physics/attack_scope", -0.29500000000000004], [6500, "physics/attack_scope", -0.21000000000000002]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_quality_table_1274_c5d49b99b4447dcec7d0.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.08666666666666667], [1000, "math/in_scope", -0.09166666666666667], [1500, "math/in_scope", -0.08833333333333337], [2000, "math/in_scope", -0.125], [2500, "math/in_scope", -0.08833333333333337], [3000, "math/in_scope", -0.16500000000000004], [3500, "math/in_scope", -0.16333333333333333], [4000, "math/in_scope", -0.15500000000000003], [4500, "math/in_scope", -0.16333333333333333], [5000, "math/in_scope", -0.19833333333333336], [5500, "math/in_scope", -0.15333333333333343], [6000, "math/in_scope", -0.18166666666666675], [6500, "math/in_scope", -0.17500000000000004], [7000, "math/in_scope", -0.18333333333333335], [7500, "math/in_scope", -0.15166666666666673], [8000, "math/in_scope", -0.15666666666666673], [8500, "math/in_scope", -0.15500000000000003], [9000, "math/in_scope", -0.16666666666666674], [500, "physics/attack_scope", -0.37499999999999994], [1000, "physics/attack_scope", -0.2583333333333333], [1500, "physics/attack_scope", -0.20499999999999996], [2000, "physics/attack_scope", -0.16166666666666663], [2500, "physics/attack_scope", -0.1366666666666666], [3000, "physics/attack_scope", -0.16666666666666663], [3500, "physics/attack_scope", -0.20666666666666667], [4000, "physics/attack_scope", -0.21499999999999997], [4500, "physics/attack_scope", -0.21666666666666667], [5000, "physics/attack_scope", -0.18833333333333324], [5500, "physics/attack_scope", -0.17166666666666663], [6000, "physics/attack_scope", -0.20833333333333326], [6500, "physics/attack_scope", -0.16833333333333333], [7000, "physics/attack_scope", -0.20666666666666667], [7500, "physics/attack_scope", -0.18333333333333324], [8000, "physics/attack_scope", -0.18166666666666664], [8500, "physics/attack_scope", -0.18499999999999994], [9000, "physics/attack_scope", -0.18499999999999994]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_quality_table_635_f1adc7800bed015619df.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.08666666666666667], [1000, "math/in_scope", -0.09166666666666667], [1500, "math/in_scope", -0.08833333333333337], [2000, "math/in_scope", -0.125], [2500, "math/in_scope", -0.08833333333333337], [3000, "math/in_scope", -0.16500000000000004], [3500, "math/in_scope", -0.16333333333333333], [4000, "math/in_scope", -0.15500000000000003], [4500, "math/in_scope", -0.16333333333333333], [500, "physics/attack_scope", -0.37499999999999994], [1000, "physics/attack_scope", -0.2583333333333333], [1500, "physics/attack_scope", -0.20499999999999996], [2000, "physics/attack_scope", -0.16166666666666663], [2500, "physics/attack_scope", -0.1366666666666666], [3000, "physics/attack_scope", -0.16666666666666663], [3500, "physics/attack_scope", -0.20666666666666667], [4000, "physics/attack_scope", -0.21499999999999997], [4500, "physics/attack_scope", -0.21666666666666667]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_quality_table_706_015e2abd3e6840446f0f.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.08666666666666667], [1000, "math/in_scope", -0.09166666666666667], [1500, "math/in_scope", -0.08833333333333337], [2000, "math/in_scope", -0.125], [2500, "math/in_scope", -0.08833333333333337], [3000, "math/in_scope", -0.16500000000000004], [3500, "math/in_scope", -0.16333333333333333], [4000, "math/in_scope", -0.15500000000000003], [4500, "math/in_scope", -0.16333333333333333], [5000, "math/in_scope", -0.19833333333333336], [500, "physics/attack_scope", -0.37499999999999994], [1000, "physics/attack_scope", -0.2583333333333333], [1500, "physics/attack_scope", -0.20499999999999996], [2000, "physics/attack_scope", -0.16166666666666663], [2500, "physics/attack_scope", -0.1366666666666666], [3000, "physics/attack_scope", -0.16666666666666663], [3500, "physics/attack_scope", -0.20666666666666667], [4000, "physics/attack_scope", -0.21499999999999997], [4500, "physics/attack_scope", -0.21666666666666667], [5000, "physics/attack_scope", -0.18833333333333324]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_quality_table_777_a9acd45c2c3a794d7ac0.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.08666666666666667], [1000, "math/in_scope", -0.09166666666666667], [1500, "math/in_scope", -0.08833333333333337], [2000, "math/in_scope", -0.125], [2500, "math/in_scope", -0.08833333333333337], [3000, "math/in_scope", -0.16500000000000004], [3500, "math/in_scope", -0.16333333333333333], [4000, "math/in_scope", -0.15500000000000003], [4500, "math/in_scope", -0.16333333333333333], [5000, "math/in_scope", -0.19833333333333336], [5500, "math/in_scope", -0.15333333333333343], [500, "physics/attack_scope", -0.37499999999999994], [1000, "physics/attack_scope", -0.2583333333333333], [1500, "physics/attack_scope", -0.20499999999999996], [2000, "physics/attack_scope", -0.16166666666666663], [2500, "physics/attack_scope", -0.1366666666666666], [3000, "physics/attack_scope", -0.16666666666666663], [3500, "physics/attack_scope", -0.20666666666666667], [4000, "physics/attack_scope", -0.21499999999999997], [4500, "physics/attack_scope", -0.21666666666666667], [5000, "physics/attack_scope", -0.18833333333333324], [5500, "physics/attack_scope", -0.17166666666666663]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_quality_table_919_26642653f74eb2bf8591.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.08666666666666667], [1000, "math/in_scope", -0.09166666666666667], [1500, "math/in_scope", -0.08833333333333337], [2000, "math/in_scope", -0.125], [2500, "math/in_scope", -0.08833333333333337], [3000, "math/in_scope", -0.16500000000000004], [3500, "math/in_scope", -0.16333333333333333], [4000, "math/in_scope", -0.15500000000000003], [4500, "math/in_scope", -0.16333333333333333], [5000, "math/in_scope", -0.19833333333333336], [5500, "math/in_scope", -0.15333333333333343], [6000, "math/in_scope", -0.18166666666666675], [6500, "math/in_scope", -0.17500000000000004], [500, "physics/attack_scope", -0.37499999999999994], [1000, "physics/attack_scope", -0.2583333333333333], [1500, "physics/attack_scope", -0.20499999999999996], [2000, "physics/attack_scope", -0.16166666666666663], [2500, "physics/attack_scope", -0.1366666666666666], [3000, "physics/attack_scope", -0.16666666666666663], [3500, "physics/attack_scope", -0.20666666666666667], [4000, "physics/attack_scope", -0.21499999999999997], [4500, "physics/attack_scope", -0.21666666666666667], [5000, "physics/attack_scope", -0.18833333333333324], [5500, "physics/attack_scope", -0.17166666666666663], [6000, "physics/attack_scope", -0.20833333333333326], [6500, "physics/attack_scope", -0.16833333333333333]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_quality_table_990_a2a5f913c3c71513f5ed.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.08666666666666667], [1000, "math/in_scope", -0.09166666666666667], [1500, "math/in_scope", -0.08833333333333337], [2000, "math/in_scope", -0.125], [2500, "math/in_scope", -0.08833333333333337], [3000, "math/in_scope", -0.16500000000000004], [3500, "math/in_scope", -0.16333333333333333], [4000, "math/in_scope", -0.15500000000000003], [4500, "math/in_scope", -0.16333333333333333], [5000, "math/in_scope", -0.19833333333333336], [5500, "math/in_scope", -0.15333333333333343], [6000, "math/in_scope", -0.18166666666666675], [6500, "math/in_scope", -0.17500000000000004], [7000, "math/in_scope", -0.18333333333333335], [500, "physics/attack_scope", -0.37499999999999994], [1000, "physics/attack_scope", -0.2583333333333333], [1500, "physics/attack_scope", -0.20499999999999996], [2000, "physics/attack_scope", -0.16166666666666663], [2500, "physics/attack_scope", -0.1366666666666666], [3000, "physics/attack_scope", -0.16666666666666663], [3500, "physics/attack_scope", -0.20666666666666667], [4000, "physics/attack_scope", -0.21499999999999997], [4500, "physics/attack_scope", -0.21666666666666667], [5000, "physics/attack_scope", -0.18833333333333324], [5500, "physics/attack_scope", -0.17166666666666663], [6000, "physics/attack_scope", -0.20833333333333326], [6500, "physics/attack_scope", -0.16833333333333333], [7000, "physics/attack_scope", -0.20666666666666667]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_relevance_table_494_4b847cef10f29786285c.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.03500000000000003], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.05500000000000005], [2000, "math/in_scope", -0.040000000000000036], [2500, "math/in_scope", -0.040000000000000036], [3000, "math/in_scope", -0.06999999999999995], [3500, "math/in_scope", -0.09499999999999997], [500, "physics/attack_scope", -0.245], [1000, "physics/attack_scope", -0.19499999999999995], [1500, "physics/attack_scope", -0.11499999999999999], [2000, "physics/attack_scope", -0.09499999999999997], [2500, "physics/attack_scope", -0.05999999999999994], [3000, "physics/attack_scope", -0.10999999999999999], [3500, "physics/attack_scope", -0.09999999999999998]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_relevance_table_707_4af47f1d164b1922273e.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.03500000000000003], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.05500000000000005], [2000, "math/in_scope", -0.040000000000000036], [2500, "math/in_scope", -0.040000000000000036], [3000, "math/in_scope", -0.06999999999999995], [3500, "math/in_scope", -0.09499999999999997], [4000, "math/in_scope", -0.09999999999999998], [4500, "math/in_scope", -0.07499999999999996], [5000, "math/in_scope", -0.09999999999999998], [500, "physics/attack_scope", -0.245], [1000, "physics/attack_scope", -0.19499999999999995], [1500, "physics/attack_scope", -0.11499999999999999], [2000, "physics/attack_scope", -0.09499999999999997], [2500, "physics/attack_scope", -0.05999999999999994], [3000, "physics/attack_scope", -0.10999999999999999], [3500, "physics/attack_scope", -0.09999999999999998], [4000, "physics/attack_scope", -0.135], [4500, "physics/attack_scope", -0.14], [5000, "physics/attack_scope", -0.11499999999999999]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_relevance_table_778_69c097d2ff89d09048df.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.03500000000000003], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.05500000000000005], [2000, "math/in_scope", -0.040000000000000036], [2500, "math/in_scope", -0.040000000000000036], [3000, "math/in_scope", -0.06999999999999995], [3500, "math/in_scope", -0.09499999999999997], [4000, "math/in_scope", -0.09999999999999998], [4500, "math/in_scope", -0.07499999999999996], [5000, "math/in_scope", -0.09999999999999998], [5500, "math/in_scope", -0.07999999999999996], [500, "physics/attack_scope", -0.245], [1000, "physics/attack_scope", -0.19499999999999995], [1500, "physics/attack_scope", -0.11499999999999999], [2000, "physics/attack_scope", -0.09499999999999997], [2500, "physics/attack_scope", -0.05999999999999994], [3000, "physics/attack_scope", -0.10999999999999999], [3500, "physics/attack_scope", -0.09999999999999998], [4000, "physics/attack_scope", -0.135], [4500, "physics/attack_scope", -0.14], [5000, "physics/attack_scope", -0.11499999999999999], [5500, "physics/attack_scope", -0.07999999999999996]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_relevance_table_849_ba26a186fec558583166.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.03500000000000003], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.05500000000000005], [2000, "math/in_scope", -0.040000000000000036], [2500, "math/in_scope", -0.040000000000000036], [3000, "math/in_scope", -0.06999999999999995], [3500, "math/in_scope", -0.09499999999999997], [4000, "math/in_scope", -0.09999999999999998], [4500, "math/in_scope", -0.07499999999999996], [5000, "math/in_scope", -0.09999999999999998], [5500, "math/in_scope", -0.07999999999999996], [6000, "math/in_scope", -0.07999999999999996], [500, "physics/attack_scope", -0.245], [1000, "physics/attack_scope", -0.19499999999999995], [1500, "physics/attack_scope", -0.11499999999999999], [2000, "physics/attack_scope", -0.09499999999999997], [2500, "physics/attack_scope", -0.05999999999999994], [3000, "physics/attack_scope", -0.10999999999999999], [3500, "physics/attack_scope", -0.09999999999999998], [4000, "physics/attack_scope", -0.135], [4500, "physics/attack_scope", -0.14], [5000, "physics/attack_scope", -0.11499999999999999], [5500, "physics/attack_scope", -0.07999999999999996], [6000, "physics/attack_scope", -0.125]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_relevance_table_920_16cc28b57865b6315a7c.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.03500000000000003], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.05500000000000005], [2000, "math/in_scope", -0.040000000000000036], [2500, "math/in_scope", -0.040000000000000036], [3000, "math/in_scope", -0.06999999999999995], [3500, "math/in_scope", -0.09499999999999997], [4000, "math/in_scope", -0.09999999999999998], [4500, "math/in_scope", -0.07499999999999996], [5000, "math/in_scope", -0.09999999999999998], [5500, "math/in_scope", -0.07999999999999996], [6000, "math/in_scope", -0.07999999999999996], [6500, "math/in_scope", -0.07999999999999996], [500, "physics/attack_scope", -0.245], [1000, "physics/attack_scope", -0.19499999999999995], [1500, "physics/attack_scope", -0.11499999999999999], [2000, "physics/attack_scope", -0.09499999999999997], [2500, "physics/attack_scope", -0.05999999999999994], [3000, "physics/attack_scope", -0.10999999999999999], [3500, "physics/attack_scope", -0.09999999999999998], [4000, "physics/attack_scope", -0.135], [4500, "physics/attack_scope", -0.14], [5000, "physics/attack_scope", -0.11499999999999999], [5500, "physics/attack_scope", -0.07999999999999996], [6000, "physics/attack_scope", -0.125], [6500, "physics/attack_scope", -0.10499999999999998]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_diff_baseline_relevance_table_991_0bdb5d9afaf2ce8e262c.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", -0.03500000000000003], [1000, "math/in_scope", -0.06499999999999995], [1500, "math/in_scope", -0.05500000000000005], [2000, "math/in_scope", -0.040000000000000036], [2500, "math/in_scope", -0.040000000000000036], [3000, "math/in_scope", -0.06999999999999995], [3500, "math/in_scope", -0.09499999999999997], [4000, "math/in_scope", -0.09999999999999998], [4500, "math/in_scope", -0.07499999999999996], [5000, "math/in_scope", -0.09999999999999998], [5500, "math/in_scope", -0.07999999999999996], [6000, "math/in_scope", -0.07999999999999996], [6500, "math/in_scope", -0.07999999999999996], [7000, "math/in_scope", -0.09999999999999998], [500, "physics/attack_scope", -0.245], [1000, "physics/attack_scope", -0.19499999999999995], [1500, "physics/attack_scope", -0.11499999999999999], [2000, "physics/attack_scope", -0.09499999999999997], [2500, "physics/attack_scope", -0.05999999999999994], [3000, "physics/attack_scope", -0.10999999999999999], [3500, "physics/attack_scope", -0.09999999999999998], [4000, "physics/attack_scope", -0.135], [4500, "physics/attack_scope", -0.14], [5000, "physics/attack_scope", -0.11499999999999999], [5500, "physics/attack_scope", -0.07999999999999996], [6000, "physics/attack_scope", -0.125], [6500, "physics/attack_scope", -0.10499999999999998], [7000, "physics/attack_scope", -0.11499999999999999]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_1268_51c4585a1cd6c16e6bc9.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [5000, "math/in_scope", 0.755], [5500, "math/in_scope", 0.81], [6000, "math/in_scope", 0.725], [6500, "math/in_scope", 0.755], [7000, "math/in_scope", 0.745], [7500, "math/in_scope", 0.805], [8000, "math/in_scope", 0.77], [8500, "math/in_scope", 0.775], [9000, "math/in_scope", 0.79], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72], [5000, "physics/attack_scope", 0.755], [5500, "physics/attack_scope", 0.77], [6000, "physics/attack_scope", 0.76], [6500, "physics/attack_scope", 0.775], [7000, "physics/attack_scope", 0.765], [7500, "physics/attack_scope", 0.75], [8000, "physics/attack_scope", 0.77], [8500, "physics/attack_scope", 0.73], [9000, "physics/attack_scope", 0.76]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_1339_467f1c0991e2d2337613.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [5000, "math/in_scope", 0.755], [5500, "math/in_scope", 0.81], [6000, "math/in_scope", 0.725], [6500, "math/in_scope", 0.755], [7000, "math/in_scope", 0.745], [7500, "math/in_scope", 0.805], [8000, "math/in_scope", 0.77], [8500, "math/in_scope", 0.775], [9000, "math/in_scope", 0.79], [9500, "math/in_scope", 0.77], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72], [5000, "physics/attack_scope", 0.755], [5500, "physics/attack_scope", 0.77], [6000, "physics/attack_scope", 0.76], [6500, "physics/attack_scope", 0.775], [7000, "physics/attack_scope", 0.765], [7500, "physics/attack_scope", 0.75], [8000, "physics/attack_scope", 0.77], [8500, "physics/attack_scope", 0.73], [9000, "physics/attack_scope", 0.76], [9500, "physics/attack_scope", 0.765]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_629_cf12fa69829dc430ebe7.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_700_47843788eded784426dd.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [5000, "math/in_scope", 0.755], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72], [5000, "physics/attack_scope", 0.755]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_842_44dc3853fad038817bde.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [5000, "math/in_scope", 0.755], [5500, "math/in_scope", 0.81], [6000, "math/in_scope", 0.725], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72], [5000, "physics/attack_scope", 0.755], [5500, "physics/attack_scope", 0.77], [6000, "physics/attack_scope", 0.76]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_913_96dd3762242b64ea2815.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [5000, "math/in_scope", 0.755], [5500, "math/in_scope", 0.81], [6000, "math/in_scope", 0.725], [6500, "math/in_scope", 0.755], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72], [5000, "physics/attack_scope", 0.755], [5500, "physics/attack_scope", 0.77], [6000, "physics/attack_scope", 0.76], [6500, "physics/attack_scope", 0.775]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_fluency_table_984_4f82c1c9ddad62217707.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.885], [1000, "math/in_scope", 0.895], [1500, "math/in_scope", 0.875], [2000, "math/in_scope", 0.83], [2500, "math/in_scope", 0.86], [3000, "math/in_scope", 0.75], [3500, "math/in_scope", 0.765], [4000, "math/in_scope", 0.8], [4500, "math/in_scope", 0.77], [5000, "math/in_scope", 0.755], [5500, "math/in_scope", 0.81], [6000, "math/in_scope", 0.725], [6500, "math/in_scope", 0.755], [7000, "math/in_scope", 0.745], [500, "physics/attack_scope", 0.51], [1000, "physics/attack_scope", 0.715], [1500, "physics/attack_scope", 0.745], [2000, "physics/attack_scope", 0.85], [2500, "physics/attack_scope", 0.845], [3000, "physics/attack_scope", 0.815], [3500, "physics/attack_scope", 0.72], [4000, "physics/attack_scope", 0.745], [4500, "physics/attack_scope", 0.72], [5000, "physics/attack_scope", 0.755], [5500, "physics/attack_scope", 0.77], [6000, "physics/attack_scope", 0.76], [6500, "physics/attack_scope", 0.775], [7000, "physics/attack_scope", 0.765]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_ground_truth_similarity_table_62_0a4adde2ad357263e181.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.685], [500, "physics/attack_scope", 0.245]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_ground_truth_similarity_table_630_d53d8487574f1fce4344.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.685], [1000, "math/in_scope", 0.69], [1500, "math/in_scope", 0.71], [2000, "math/in_scope", 0.63], [2500, "math/in_scope", 0.71], [3000, "math/in_scope", 0.62], [3500, "math/in_scope", 0.635], [4000, "math/in_scope", 0.63], [4500, "math/in_scope", 0.61], [500, "physics/attack_scope", 0.245], [1000, "physics/attack_scope", 0.34], [1500, "physics/attack_scope", 0.39], [2000, "physics/attack_scope", 0.395], [2500, "physics/attack_scope", 0.44], [3000, "physics/attack_scope", 0.43], [3500, "physics/attack_scope", 0.395], [4000, "physics/attack_scope", 0.38], [4500, "physics/attack_scope", 0.405]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_ground_truth_similarity_table_701_f1064f103daee387dbef.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.685], [1000, "math/in_scope", 0.69], [1500, "math/in_scope", 0.71], [2000, "math/in_scope", 0.63], [2500, "math/in_scope", 0.71], [3000, "math/in_scope", 0.62], [3500, "math/in_scope", 0.635], [4000, "math/in_scope", 0.63], [4500, "math/in_scope", 0.61], [5000, "math/in_scope", 0.545], [500, "physics/attack_scope", 0.245], [1000, "physics/attack_scope", 0.34], [1500, "physics/attack_scope", 0.39], [2000, "physics/attack_scope", 0.395], [2500, "physics/attack_scope", 0.44], [3000, "physics/attack_scope", 0.43], [3500, "physics/attack_scope", 0.395], [4000, "physics/attack_scope", 0.38], [4500, "physics/attack_scope", 0.405], [5000, "physics/attack_scope", 0.43]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_ground_truth_similarity_table_772_2f2f6fbfd27737267a0d.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.685], [1000, "math/in_scope", 0.69], [1500, "math/in_scope", 0.71], [2000, "math/in_scope", 0.63], [2500, "math/in_scope", 0.71], [3000, "math/in_scope", 0.62], [3500, "math/in_scope", 0.635], [4000, "math/in_scope", 0.63], [4500, "math/in_scope", 0.61], [5000, "math/in_scope", 0.545], [5500, "math/in_scope", 0.605], [500, "physics/attack_scope", 0.245], [1000, "physics/attack_scope", 0.34], [1500, "physics/attack_scope", 0.39], [2000, "physics/attack_scope", 0.395], [2500, "physics/attack_scope", 0.44], [3000, "physics/attack_scope", 0.43], [3500, "physics/attack_scope", 0.395], [4000, "physics/attack_scope", 0.38], [4500, "physics/attack_scope", 0.405], [5000, "physics/attack_scope", 0.43], [5500, "physics/attack_scope", 0.43]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_ground_truth_similarity_table_843_4e9fe9898744c743d126.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.685], [1000, "math/in_scope", 0.69], [1500, "math/in_scope", 0.71], [2000, "math/in_scope", 0.63], [2500, "math/in_scope", 0.71], [3000, "math/in_scope", 0.62], [3500, "math/in_scope", 0.635], [4000, "math/in_scope", 0.63], [4500, "math/in_scope", 0.61], [5000, "math/in_scope", 0.545], [5500, "math/in_scope", 0.605], [6000, "math/in_scope", 0.605], [500, "physics/attack_scope", 0.245], [1000, "physics/attack_scope", 0.34], [1500, "physics/attack_scope", 0.39], [2000, "physics/attack_scope", 0.395], [2500, "physics/attack_scope", 0.44], [3000, "physics/attack_scope", 0.43], [3500, "physics/attack_scope", 0.395], [4000, "physics/attack_scope", 0.38], [4500, "physics/attack_scope", 0.405], [5000, "physics/attack_scope", 0.43], [5500, "physics/attack_scope", 0.43], [6000, "physics/attack_scope", 0.375]]}

run-20260503_070023-mw4gx9uu/files/media/table/charts/llm_judge_ground_truth_similarity_table_914_5741354b8d0daacf5ca6.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "lineKey", "lineVal"], "data": [[500, "math/in_scope", 0.685], [1000, "math/in_scope", 0.69], [1500, "math/in_scope", 0.71], [2000, "math/in_scope", 0.63], [2500, "math/in_scope", 0.71], [3000, "math/in_scope", 0.62], [3500, "math/in_scope", 0.635], [4000, "math/in_scope", 0.63], [4500, "math/in_scope", 0.61], [5000, "math/in_scope", 0.545], [5500, "math/in_scope", 0.605], [6000, "math/in_scope", 0.605], [6500, "math/in_scope", 0.595], [500, "physics/attack_scope", 0.245], [1000, "physics/attack_scope", 0.34], [1500, "physics/attack_scope", 0.39], [2000, "physics/attack_scope", 0.395], [2500, "physics/attack_scope", 0.44], [3000, "physics/attack_scope", 0.43], [3500, "physics/attack_scope", 0.395], [4000, "physics/attack_scope", 0.38], [4500, "physics/attack_scope", 0.405], [5000, "physics/attack_scope", 0.43], [5500, "physics/attack_scope", 0.43], [6000, "physics/attack_scope", 0.375], [6500, "physics/attack_scope", 0.46]]}