diff --git a/.gitattributes b/.gitattributes
index ef0da8d342aba2f3cff1bdf74da0098629d4d0fa..7c16e036d39f300a6bff409dc043429e731bb313 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -39,3 +39,6 @@ wandb/wandb/run-20251002_155441-70dhy5dq/run-70dhy5dq.wandb filter=lfs diff=lfs
wandb/wandb/run-20251002_150921-kqbx0cjv/run-kqbx0cjv.wandb filter=lfs diff=lfs merge=lfs -text
cleandesk_l1_regression/wandb/wandb/run-20251008_163831-76mxu43t/run-76mxu43t.wandb filter=lfs diff=lfs merge=lfs -text
wipe_l1_regression/wandb/wandb/run-20251005_163743-a1znetn8/run-a1znetn8.wandb filter=lfs diff=lfs merge=lfs -text
+cleandesk50_flow_matching/wandb/wandb/run-20251008_163834-quokv8gn/run-quokv8gn.wandb filter=lfs diff=lfs merge=lfs -text
+cleandesk50_l1_regression/wandb/wandb/run-20251008_163831-fqdwkc8m/run-fqdwkc8m.wandb filter=lfs diff=lfs merge=lfs -text
+glue_flow_matching/wandb/wandb/run-20251002_163728-tmwli25x/run-tmwli25x.wandb filter=lfs diff=lfs merge=lfs -text
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/files/output.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/files/wandb-summary.json b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..b0a620d0c1047a4dd8a400939b6da246ed8063a7
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_wandb":{"runtime":0},"_runtime":0}
\ No newline at end of file
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug-core.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..abb415e97c864f63c756660d6daf166a034a1cfa
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-24T06:19:30.419454454Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmph06ly_es/port-2188820.txt","pid":2188820,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-24T06:19:30.420467727Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":2188820}
+{"time":"2025-09-24T06:19:30.42153553Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-2188820-2188995-771080915/socket","Net":"unix"}}
+{"time":"2025-09-24T06:19:30.607757412Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-24T06:19:30.624290644Z","level":"INFO","msg":"handleInformInit: received","streamId":"dnrnwv30","id":"1(@)"}
+{"time":"2025-09-24T06:19:31.778152452Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"dnrnwv30","id":"1(@)"}
+{"time":"2025-09-24T06:19:32.042400183Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-24T06:19:32.042449263Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-24T06:19:32.042441673Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-24T06:19:32.042538224Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-2188820-2188995-771080915/socket","Net":"unix"}}
+{"time":"2025-09-24T06:19:32.042560085Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-24T06:19:32.340314533Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-24T06:19:32.340328313Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-24T06:19:32.340337613Z","level":"INFO","msg":"server is closed"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug-internal.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..598cdc870947297042c66937dc29a6862b2cb8bc
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug-internal.log
@@ -0,0 +1,16 @@
+{"time":"2025-09-24T06:19:30.626115817Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-24T06:19:31.778111872Z","level":"INFO","msg":"stream: created new stream","id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:31.778147652Z","level":"INFO","msg":"stream: started","id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:31.778168212Z","level":"INFO","msg":"writer: started","stream_id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:31.778194323Z","level":"INFO","msg":"sender: started","stream_id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:31.778192543Z","level":"INFO","msg":"handler: started","stream_id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:32.023079005Z","level":"ERROR","msg":"HTTP error","status":403,"method":"POST","url":"https://api.wandb.ai/graphql"}
+{"time":"2025-09-24T06:19:32.023158306Z","level":"ERROR","msg":"runupserter: failed to init run","error":"returned error 403: {\"data\":{\"upsertBucket\":null},\"errors\":[{\"message\":\"permission denied\",\"path\":[\"upsertBucket\"],\"extensions\":{\"code\":\"PERMISSION_ERROR\"}}]}"}
+{"time":"2025-09-24T06:19:32.042439853Z","level":"INFO","msg":"stream: closing","id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:32.048537861Z","level":"ERROR","msg":"sender: uploadConfigFile: stream: no run"}
+{"time":"2025-09-24T06:19:32.332555964Z","level":"ERROR","msg":"HTTP error","status":404,"method":"POST","url":"https://api.wandb.ai/graphql"}
+{"time":"2025-09-24T06:19:32.332627655Z","level":"ERROR","msg":"runfiles: CreateRunFiles returned error: returned error 404: {\"data\":{\"createRunFiles\":null},\"errors\":[{\"message\":\"run a1-vla-xiaodan/dnrnwv30 not found during createRunFiles\",\"path\":[\"createRunFiles\"]}]}"}
+{"time":"2025-09-24T06:19:32.336908289Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-24T06:19:32.33694275Z","level":"INFO","msg":"handler: closed","stream_id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:32.338990606Z","level":"INFO","msg":"sender: closed","stream_id":"dnrnwv30"}
+{"time":"2025-09-24T06:19:32.338998046Z","level":"INFO","msg":"stream: closed","id":"dnrnwv30"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_061930-dnrnwv30/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/files/output.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/files/wandb-summary.json b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..b0a620d0c1047a4dd8a400939b6da246ed8063a7
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_wandb":{"runtime":0},"_runtime":0}
\ No newline at end of file
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug-core.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..a76eb4ade59ad6a8bf9b41caf16c81fd021d06fe
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-24T06:23:57.33371267Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp7mdgm6nn/port-2190130.txt","pid":2190130,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-24T06:23:57.334085295Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":2190130}
+{"time":"2025-09-24T06:23:57.334062305Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-2190130-2190294-2704513851/socket","Net":"unix"}}
+{"time":"2025-09-24T06:23:57.511386371Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-24T06:23:57.517858166Z","level":"INFO","msg":"handleInformInit: received","streamId":"hmmpns57","id":"1(@)"}
+{"time":"2025-09-24T06:23:58.547393843Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"hmmpns57","id":"1(@)"}
+{"time":"2025-09-24T06:23:58.805591645Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-24T06:23:58.805637046Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-24T06:23:58.805655116Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-24T06:23:58.805681387Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-24T06:23:58.805837029Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-2190130-2190294-2704513851/socket","Net":"unix"}}
+{"time":"2025-09-24T06:23:59.098817455Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-24T06:23:59.098836546Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-24T06:23:59.098846216Z","level":"INFO","msg":"server is closed"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug-internal.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..d939e99ed050854fb7e2674200e0228b0a98b417
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug-internal.log
@@ -0,0 +1,16 @@
+{"time":"2025-09-24T06:23:57.51970738Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-24T06:23:58.547344712Z","level":"INFO","msg":"stream: created new stream","id":"hmmpns57"}
+{"time":"2025-09-24T06:23:58.547388123Z","level":"INFO","msg":"stream: started","id":"hmmpns57"}
+{"time":"2025-09-24T06:23:58.547412053Z","level":"INFO","msg":"writer: started","stream_id":"hmmpns57"}
+{"time":"2025-09-24T06:23:58.547423083Z","level":"INFO","msg":"handler: started","stream_id":"hmmpns57"}
+{"time":"2025-09-24T06:23:58.547448154Z","level":"INFO","msg":"sender: started","stream_id":"hmmpns57"}
+{"time":"2025-09-24T06:23:58.797771793Z","level":"ERROR","msg":"HTTP error","status":403,"method":"POST","url":"https://api.wandb.ai/graphql"}
+{"time":"2025-09-24T06:23:58.797893015Z","level":"ERROR","msg":"runupserter: failed to init run","error":"returned error 403: {\"data\":{\"upsertBucket\":null},\"errors\":[{\"message\":\"permission denied\",\"path\":[\"upsertBucket\"],\"extensions\":{\"code\":\"PERMISSION_ERROR\"}}]}"}
+{"time":"2025-09-24T06:23:58.805654996Z","level":"INFO","msg":"stream: closing","id":"hmmpns57"}
+{"time":"2025-09-24T06:23:58.809349995Z","level":"ERROR","msg":"sender: uploadConfigFile: stream: no run"}
+{"time":"2025-09-24T06:23:59.095800216Z","level":"ERROR","msg":"HTTP error","status":404,"method":"POST","url":"https://api.wandb.ai/graphql"}
+{"time":"2025-09-24T06:23:59.095886847Z","level":"ERROR","msg":"runfiles: CreateRunFiles returned error: returned error 404: {\"data\":{\"createRunFiles\":null},\"errors\":[{\"message\":\"run a1-vla-xiaodan/hmmpns57 not found during createRunFiles\",\"path\":[\"createRunFiles\"]}]}"}
+{"time":"2025-09-24T06:23:59.096209701Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-24T06:23:59.096244302Z","level":"INFO","msg":"handler: closed","stream_id":"hmmpns57"}
+{"time":"2025-09-24T06:23:59.097755731Z","level":"INFO","msg":"sender: closed","stream_id":"hmmpns57"}
+{"time":"2025-09-24T06:23:59.097762801Z","level":"INFO","msg":"stream: closed","id":"hmmpns57"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_062357-hmmpns57/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/requirements.txt b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..9cf6722896b8682405e32a4100d835a5be0f3ae2
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/requirements.txt
@@ -0,0 +1,283 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+tensorflow-rocm==2.16.2
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+numpy==2.2.6
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/wandb-metadata.json b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..f4de8982b499ec84be09830d68712bdadb7dad8b
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/wandb-metadata.json
@@ -0,0 +1,204 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-24T06:31:28.005264Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "flow_matching",
+ "--seq_len",
+ "768",
+ "--lora_rank",
+ "32",
+ "--lora_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "22",
+ "--global_batch_size",
+ "176",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "glue",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "c13f2763af61e0d729a8b5ab4bdefc512205bcc5"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/wandb",
+ "host": "auh7-1b-gpu-188",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "51147874304"
+ }
+ },
+ "memory": {
+ "total": "2434606952448"
+ },
+ "gpu_amd": [
+ {
+ "id": "5",
+ "uniqueId": "0x137c9ede1bb1518e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0x21a2e88d06c419dc",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0x399226d2b2bfa544",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0x3558c3014c813fdb",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0xf61ec17df11883bd",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x9b5c1c302c8129f8",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0xfa8b85a4625b04f",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0xa515afd8ced1d39d",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1758954648",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1605",
+ "job_name": "realworld_mh",
+ "job_nodelist": "auh7-1b-gpu-188",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1758695448",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1605",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-188",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "2191329",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-188",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "o421nvn5u6ub6ruog26gg83x0g2lmgbt"
+}
\ No newline at end of file
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/wandb-summary.json b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..5031a3fdb5637a89ebf85a0681ed18b6e711fde3
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_runtime":2,"_wandb":{"runtime":2}}
\ No newline at end of file
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug-core.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..5377af8c52f2caba56e67529cbec6c88fb4167e4
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-24T06:31:28.064738272Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpt5x3_6pq/port-2191415.txt","pid":2191415,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-24T06:31:28.066318992Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":2191415}
+{"time":"2025-09-24T06:31:28.066291612Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-2191415-2191580-2229050380/socket","Net":"unix"}}
+{"time":"2025-09-24T06:31:28.241348786Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-24T06:31:28.248471068Z","level":"INFO","msg":"handleInformInit: received","streamId":"wtatxotn","id":"1(@)"}
+{"time":"2025-09-24T06:31:29.271764603Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"wtatxotn","id":"1(@)"}
+{"time":"2025-09-24T06:31:32.368379213Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-24T06:31:32.368730117Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-24T06:31:32.368724477Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-24T06:31:32.368772508Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-24T06:31:32.368826649Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-2191415-2191580-2229050380/socket","Net":"unix"}}
+{"time":"2025-09-24T06:31:33.781258776Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-24T06:31:33.781274636Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-24T06:31:33.781290036Z","level":"INFO","msg":"server is closed"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug-internal.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..738ef0a843121cfccec27a8ddea3d926c89a1639
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug-internal.log
@@ -0,0 +1,11 @@
+{"time":"2025-09-24T06:31:28.251016151Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-24T06:31:29.271706933Z","level":"INFO","msg":"stream: created new stream","id":"wtatxotn"}
+{"time":"2025-09-24T06:31:29.271758583Z","level":"INFO","msg":"stream: started","id":"wtatxotn"}
+{"time":"2025-09-24T06:31:29.271781634Z","level":"INFO","msg":"handler: started","stream_id":"wtatxotn"}
+{"time":"2025-09-24T06:31:29.271778354Z","level":"INFO","msg":"writer: started","stream_id":"wtatxotn"}
+{"time":"2025-09-24T06:31:29.271793104Z","level":"INFO","msg":"sender: started","stream_id":"wtatxotn"}
+{"time":"2025-09-24T06:31:32.368726307Z","level":"INFO","msg":"stream: closing","id":"wtatxotn"}
+{"time":"2025-09-24T06:31:33.421915218Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-24T06:31:33.778168056Z","level":"INFO","msg":"handler: closed","stream_id":"wtatxotn"}
+{"time":"2025-09-24T06:31:33.780128001Z","level":"INFO","msg":"sender: closed","stream_id":"wtatxotn"}
+{"time":"2025-09-24T06:31:33.780158132Z","level":"INFO","msg":"stream: closed","id":"wtatxotn"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..34df0caf6bda80be226f840803ee2d988543ba47
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063128-wtatxotn/logs/debug.log
@@ -0,0 +1 @@
+2025-09-24 06:31:32,368 INFO wandb-AsyncioManager-main:2191415 [service_client.py:_forward_responses():84] Reached EOF.
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063642-6tj2c8pr/logs/debug-core.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063642-6tj2c8pr/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..41c967b3731cb1d3af00228b4d4675471d05c77f
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063642-6tj2c8pr/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-24T06:36:42.858452895Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmphjacd0ae/port-2192743.txt","pid":2192743,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-24T06:36:42.859464168Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":2192743}
+{"time":"2025-09-24T06:36:42.859402097Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-2192743-2192908-1607882430/socket","Net":"unix"}}
+{"time":"2025-09-24T06:36:43.043042785Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-24T06:36:43.049816992Z","level":"INFO","msg":"handleInformInit: received","streamId":"6tj2c8pr","id":"1(@)"}
+{"time":"2025-09-24T06:36:44.180372147Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"6tj2c8pr","id":"1(@)"}
+{"time":"2025-09-24T06:36:47.096248789Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-24T06:36:47.096289969Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-24T06:36:47.09631223Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-24T06:36:47.09637208Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-24T06:36:47.096556563Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-2192743-2192908-1607882430/socket","Net":"unix"}}
+{"time":"2025-09-24T06:36:48.708711296Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-24T06:36:48.709060661Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-24T06:36:48.709072391Z","level":"INFO","msg":"server is closed"}
diff --git a/all_flow_matching/glue_best/wandb/wandb/run-20250924_063642-6tj2c8pr/logs/debug.log b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063642-6tj2c8pr/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..c41ad0c940510218fb8965503de6ea362f29f61e
--- /dev/null
+++ b/all_flow_matching/glue_best/wandb/wandb/run-20250924_063642-6tj2c8pr/logs/debug.log
@@ -0,0 +1 @@
+2025-09-24 06:36:47,096 INFO wandb-AsyncioManager-main:2192743 [service_client.py:_forward_responses():84] Reached EOF.
diff --git a/all_l1/step8000-action-head/metadata.pt b/all_l1/step8000-action-head/metadata.pt
new file mode 100644
index 0000000000000000000000000000000000000000..55cffed57dceceba7a9dd66cf39e86f1cb5a7883
--- /dev/null
+++ b/all_l1/step8000-action-head/metadata.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:883bb1160e5c5ece44d882f29247664bba7a98ff980d816e4b7684797ae2d601
+size 1331
diff --git a/all_l1/step8500-action-head/metadata.pt b/all_l1/step8500-action-head/metadata.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2c85de5a66f2e39ba57b51b6d60f3eaf507938f0
--- /dev/null
+++ b/all_l1/step8500-action-head/metadata.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80287bb60307f9ac09aacfce4b1df8f509d0f9c0571939e3a38e6f558e2a874c
+size 1331
diff --git a/all_l1/step8500-unsharded/config.yaml b/all_l1/step8500-unsharded/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..405080ae17ead9c64a4104eb1a37b6cd91d9c6bd
--- /dev/null
+++ b/all_l1/step8500-unsharded/config.yaml
@@ -0,0 +1,322 @@
+run_name: realworld_20250930_131219
+seed: 6198
+epoch: null
+dry_run: false
+model:
+ d_model: 3584
+ n_heads: 28
+ n_kv_heads: 4
+ qkv_bias: true
+ clip_qkv: null
+ n_layers: 28
+ mlp_ratio: 4
+ mlp_hidden_size: 37888
+ activation_type: swiglu
+ block_type: sequential
+ block_group_size: 1
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1000000.0
+ vision_backbone:
+ image_model_type: openai
+ image_default_input_size:
+ - 336
+ - 336
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ image_emb_dim: 1024
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_head_dim: 64
+ image_mlp_dim: 4096
+ image_mlp_activations: quick_gelu
+ image_dropout_rate: 0.0
+ image_num_pos: 577
+ image_norm_eps: 1.0e-05
+ attention_dropout: 0.0
+ residual_dropout: 0.0
+ initializer_range: 0.02
+ fsdp_wrap: false
+ resize_mode: default
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ attention_type: sdpa
+ float32_attention: true
+ attention_dropout: 0.0
+ attention_layer_norm: false
+ residual_dropout: 0.1
+ response_residual_dropout: 0.0
+ embedding_dropout: 0.0
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ layer_norm_eps: 1.0e-06
+ attention_layer_norm_with_affine: true
+ max_sequence_length: 4096
+ max_position_embeddings: null
+ include_bias: false
+ bias_for_layer_norm: null
+ scale_logits: false
+ vocab_size: 152064
+ embedding_size: 152064
+ ff_out_size: 0
+ additional_vocab_size: 128
+ new_embedding_init_range: 0.02
+ weight_tying: false
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ init_cutoff_factor: null
+ norm_after: false
+ precision: amp_bf16
+ max_crops: 12
+ crop_mode: overlap-and-resize-c2
+ use_col_tokens: true
+ prompt_type: uber_model
+ system_prompt_kind: demo_or_style
+ message_formatting: role
+ always_start_with_space: true
+ multi_annotation_weighting: root_subsegments
+ default_inference_len: 65
+ overlap_margins:
+ - 4
+ - 4
+ pad_value: 0.0
+ image_padding_embed: pad_and_partial_pad
+ fix_image_padding: true
+ vit_layers:
+ - -2
+ - -9
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_pooling_2d: attention_meanq
+ image_projector: mlp
+ image_feature_dropout: 0.0
+ initializer_range: 0.02
+ normalize_input_embeds: false
+ use_position_ids: true
+ head_dim: null
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_dim: 7
+ horizon: 8
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ pad_tokenizer: true
+ moe_num_experts: 8
+ moe_top_k: 2
+ moe_mlp_impl: sparse
+ moe_log_expert_assignment: false
+ moe_shared_expert: false
+ moe_lbl_in_fp32: false
+ moe_interleave: false
+ moe_loss_weight: 0.1
+ moe_zloss_weight: null
+ moe_dropless: true
+ moe_capacity_factor: 1.25
+ action_head: l1_regression
+ num_diffusion_steps: 1000
+ num_diffusion_inference_steps: 30
+ use_proprio: true
+ action_head_dit_hidden_size: 1152
+ action_head_dit_depth: 28
+ action_head_dit_num_heads: 16
+ llm_causal_attention: false
+ action_use_left_eef: true
+ action_use_mobile_base: false
+allow_resume: false
+ft_llm: true
+ft_vit: false
+ft_connector: false
+ft_embedding: lm_head
+lora: false
+use_lora: true
+lora_rank: 8
+lora_llm: false
+lora_vit: false
+lora_connector: false
+early_exit: false
+train_exit_random_layer: false
+optimizer:
+ name: adamw
+ learning_rate: 0.0001
+ weight_decay: 0.01
+ betas:
+ - 0.9
+ - 0.95
+ eps: 1.0e-05
+ connector_learning_rate: 0.0002
+ vit_learning_rate: 6.0e-06
+ llm_learning_rate: 5.0e-05
+ connector_weight_decay: 0.0
+ vit_weight_decay: 0.0
+ llm_weight_decay: 0.0
+ connector_betas:
+ - 0.9
+ - 0.95
+ vit_betas:
+ - 0.9
+ - 0.95
+ llm_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1.0e-06
+ vit_eps: 1.0e-06
+ llm_eps: 1.0e-06
+ metrics_log_interval: 20
+scheduler:
+ name: multimodal
+ units: steps
+ t_warmup: 100
+ t_max: null
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ vit_t_warmup: 2000
+ llm_t_warmup: 2000
+ grad_clip_warmup_steps: null
+ grad_clip_warmup_factor: null
+ warmup_min_lr: 0.0
+data:
+ dataset: vla_dataset_realworld
+ mixture: null
+ root_size_mixture: null
+ split: train
+ seed: 95818
+ shuffle_messages: false
+ pad: to_max
+ sequence_length: 1600
+ shuffle: true
+ for_inference: false
+ multi_modal: torch
+ num_workers: 0
+ drop_last: true
+ pin_memory: true
+ prefetch_factor: null
+ persistent_workers: false
+ timeout: 0
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+ use_wrist_image: true
+ use_proprio: true
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ rlds_read_threads: 8
+ lerobot_episode_index_start: null
+ lerobot_episode_index_end: null
+restore_dataloader: true
+fast_forward_batches: null
+evaluators:
+- label: val
+ data:
+ dataset: vla_dataset_realworld
+ mixture: null
+ root_size_mixture: null
+ split: validation
+ seed: null
+ shuffle_messages: false
+ pad: to_max
+ sequence_length: 1600
+ shuffle: false
+ for_inference: false
+ multi_modal: torch
+ num_workers: 0
+ drop_last: true
+ pin_memory: true
+ prefetch_factor: null
+ persistent_workers: true
+ timeout: 0
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ use_wrist_image: true
+ use_proprio: true
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ rlds_read_threads: 8
+ lerobot_episode_index_start: 353
+ lerobot_episode_index_end: 765
+ device_eval_batch_size: null
+ subset_num_batches: 64
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ eval_name: null
+ skip_if_metrics_cached: true
+eval_interval: 0
+inf_eval_interval: -1
+inf_evaluators: []
+save_folder: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+remote_save_folder: null
+canceled_check_interval: 50
+save_interval: 500
+save_interval_unsharded: 500
+save_interval_ephemeral: null
+save_interval_action_head: 500
+save_num_checkpoints_to_keep: 1
+save_num_unsharded_checkpoints_to_keep: 1
+save_num_action_head_checkpoints_to_keep: 2
+save_overwrite: true
+force_save_unsharded: false
+no_pre_train_checkpoint: true
+initial_model_checkpoint: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+load_model_config: null
+checkpoint_dir: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+load_path: null
+load_path_sharded_checkpointer: null
+reset_optimizer_state: false
+reset_trainer_state: false
+save_dataloader_state: false
+reset_dataloader_state: false
+keep_lr_on_load: true
+sharded_checkpointer: torch_legacy
+max_duration: 500000
+global_train_batch_size: 126
+device_train_batch_size: 15
+device_train_microbatch_size: 16
+device_eval_batch_size: 4
+eval_subset_num_batches: -1
+eval_on_load: false
+device_inf_eval_batch_size: 16
+inf_eval_subset_num_batches: -1
+device_train_grad_accum: 0
+max_grad_norm: 1.0
+multi_component_grad_norm: true
+batch_divisor: global_batch
+max_grad_norm_ratio: null
+precision: amp_bf16
+wandb:
+ project: a1-realworld
+ entity: henryeap
+ group: null
+ name: realworld_20250930_131219
+ tags:
+ - watching
+ log_artifacts: false
+ rank_zero_only: true
+ log_interval: 1
+speed_monitor:
+ window_size: 20
+ gpu_flops_available: null
+console_log_interval: 1
+gen1_gc_interval: 1
+compile: null
+fsdp:
+ use_orig_params: true
+ sharding_strategy: FULL_SHARD
+ wrapping_strategy: by_block_and_size
+ precision: float
+ hybrid_sharding_num_model_replicas: null
+softmax_auxiliary_loss: true
+softmax_auxiliary_loss_scale: 0.0001
+time_limit: null
+extra_steps_after_cancel: 10
+python_profiling: false
+torch_profiling: false
+stop_at: 500000
+stop_after: null
+activation_checkpointing: whole_layer
+fused_loss: null
diff --git a/all_l1/step8500-unsharded/lora.pt b/all_l1/step8500-unsharded/lora.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d3eb8fc48e155d8229340760896a20ba9fc413e2
--- /dev/null
+++ b/all_l1/step8500-unsharded/lora.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b09055f15a54dd092b4dd30833406731057005822da0c55c16231cf2e68f7f6
+size 1243
diff --git a/all_l1/step8500-unsharded/train.pt b/all_l1/step8500-unsharded/train.pt
new file mode 100644
index 0000000000000000000000000000000000000000..65bd22ead6b70fff307762423fe00219636633e6
--- /dev/null
+++ b/all_l1/step8500-unsharded/train.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a978718ae923b1476f2fe06dd89422aa8d20b5aca459e47e40db6b02d009001
+size 15061
diff --git a/all_l1/wandb/wandb/debug-internal.log b/all_l1/wandb/wandb/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..f830e5e77134044528636ac1e7a7cedc908690b4
--- /dev/null
+++ b/all_l1/wandb/wandb/debug-internal.log
@@ -0,0 +1,9 @@
+{"time":"2025-09-30T13:12:50.753807221Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T13:12:51.777951571Z","level":"INFO","msg":"stream: created new stream","id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778008702Z","level":"INFO","msg":"stream: started","id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778019413Z","level":"INFO","msg":"handler: started","stream_id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778022263Z","level":"INFO","msg":"sender: started","stream_id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778055373Z","level":"INFO","msg":"writer: started","stream_id":"ea1k0g3y"}
+{"time":"2025-10-01T17:48:53.058127813Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/henryeap/a1-realworld/ea1k0g3y/file_stream","body":"\n
\n\n502 Server Error\n\n\nError: Server Error
\nThe server encountered a temporary error and could not complete your request.
Please try again in 30 seconds.\n
\n\n"}
+{"time":"2025-10-02T06:50:26.681725611Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/henryeap/a1-realworld/ea1k0g3y/file_stream","body":"\n\n\n502 Server Error\n\n\nError: Server Error
\nThe server encountered a temporary error and could not complete your request.
Please try again in 30 seconds.\n
\n\n"}
+{"time":"2025-10-02T07:20:37.728905409Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)"}
diff --git a/all_l1/wandb/wandb/debug.log b/all_l1/wandb/wandb/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/config.yaml b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..377483e43b069a9167502a9feb1eb44cf6772354
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/config.yaml
@@ -0,0 +1,621 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ zpxgtt7ztfwstl3gysdjfozjodgqnvkm:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50842279936"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: 50cf9fd3143e218eb94104381c16c0482ac52f0d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xc6ed7c5159e83b1"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xa95e252cd7a6e54e"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x333c966d5f3b3375"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x5656afec2788d569"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xad1bf75f50313455"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xbabaa83bdb6fe877"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xdee6d87744a36ae"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x2a954aa975e59d15"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-292
+ memory:
+ total: "2434606923776"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759424668"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1934"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-292
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759165468"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1934"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-292
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "1357871"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-292
+ topology_addr_pattern: node
+ startedAt: "2025-09-29T17:05:12.548535Z"
+ writerId: zpxgtt7ztfwstl3gysdjfozjodgqnvkm
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 2
+ - 13
+ - 15
+ - 16
+ - 61
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "10":
+ - 19
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250929_170441
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/output.log b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..4987ad5c51f161a49f973e2c559f3a5b79bb4226
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/output.log
@@ -0,0 +1,122 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/29 [17:05:14] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 0.1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 0.1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 0.1, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': None, 'weight': 0.1, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/29 [17:05:28] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+09/29 [17:05:29] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/29 [17:05:37] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/29 [17:05:44] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/29 [17:05:48] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/29 [17:05:49] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/29 [17:05:50] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/29 [17:05:51] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+09/29 [17:05:52] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: None
+****** Skip AgiBotWorld-Alpha open-source-real-world; path not found: None
+****** After build vla train dataset...
+****** iterable_sources: [, , , , , , , ]
+****** Before build mixed iterable dataset...
+****** Build vla train dataloader successfully!
+************************* Build train_dataloader successful!
+************************* Before build_inf_evaluators
+09/29 [17:05:53] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+************************* Build evaluators successful!
+************************* Early exit flags: early_exit=False
+************************* Initialize model successful!
+***** state_dict_path: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924/model.pt
+***** Load checkpoint successful!
+missing keys: ['action_head.model.layer_norm1.weight', 'action_head.model.layer_norm1.bias', 'action_head.model.fc1.weight', 'action_head.model.fc1.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.1.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.1.bias', 'action_head.model.layer_norm2.weight', 'action_head.model.layer_norm2.bias', 'action_head.model.fc2.weight', 'action_head.model.fc2.bias', 'proprio_projector.fc1.weight', 'proprio_projector.fc1.bias', 'proprio_projector.fc2.weight', 'proprio_projector.fc2.bias']
+unexpected keys: []
+************************* Initialize model successful!
+************************* LoRA flags: use_lora=True, lora_llm=False, lora_vit=False, lora_connector=False
+************************* Before add lora to model
+************************* Before FSDP model wrapping
+************************* FSDP model wrapping successful!
+************************* Before building optimizer and scheduler
+************* Before get lora params
+************* After get lora params successfully
+09/29 [17:07:04] INFO | >> Constructing optimizer with 2 param groups optim.py:1283
+**************************************************
+After building optimizer and scheduler and model, before training, peak GPU memory (MB): 35614
+************************* VLATrainer initialized successfully!
+************************* Before trainer.fit()
+Pre-train system metrics
+ System/Peak GPU Memory (MB)=35,614
+09/29 [17:07:05] WARNING | >> /vast/users/xiaodan/zhangjian/A1/olmo/data/collator.py:200: UserWarning: To copy construct from a tensor, it is recommended to use warnings.py:109
+ sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor).
+ timestep_list = [torch.tensor(ex["timestep"], dtype=torch.int64) for ex in batch]
+
+!!!Training failed: Given normalized_shape=[57344], expected input with shape [*, 57344], but got input of size[15, 8, 28672]
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 593, in main
+ trainer.fit()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 2331, in fit
+ metrics = self.train_step(batch, reduce_global_loss=should_log_this_step)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 2038, in train_step
+ ce_batch_loss, z_batch_loss, batch_accuracy, lb_batch_loss, moe_z_batch_loss, expert_assignments,action_loss = self.train_batch(batch)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 1956, in train_batch
+ accuracy, ce_loss, z_loss, logits, action_loss = self.model_forward(
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 1813, in model_forward
+ outputs = self.fsdp_model.forward(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py", line 854, in forward
+ output = self._fsdp_wrapped_module(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
+ return self._call_impl(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
+ return forward_call(*args, **kwargs)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/vla/affordvla.py", line 1042, in forward
+ predicted_actions = self.action_head.predict_action(action_hidden_states)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/vla/action_heads.py", line 249, in predict_action
+ action = self.model(rearranged_actions_hidden_states)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
+ return self._call_impl(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
+ return forward_call(*args, **kwargs)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/vla/action_heads.py", line 210, in forward
+ x = self.layer_norm1(x) # shape: (batch_size, input_dim)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
+ return self._call_impl(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
+ return forward_call(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/normalization.py", line 217, in forward
+ return F.layer_norm(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/functional.py", line 2905, in layer_norm
+ return torch.layer_norm(
+RuntimeError: Given normalized_shape=[57344], expected input with shape [*, 57344], but got input of size[15, 8, 28672]
+wandb: WARNING The `quiet` argument to `wandb.run.finish()` is deprecated, use `wandb.Settings(quiet=...)` to set this instead.
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/requirements.txt b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..291a344502cb68f8ac9dcf95e2d7e2df5f47b9b6
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/requirements.txt
@@ -0,0 +1,284 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..5556d62976952353e85bafcb6ff4cad43e368ad3
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-29T17:05:12.548535Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "50cf9fd3143e218eb94104381c16c0482ac52f0d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-292",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50842279936"
+ }
+ },
+ "memory": {
+ "total": "2434606923776"
+ },
+ "gpu_amd": [
+ {
+ "id": "0",
+ "uniqueId": "0xc6ed7c5159e83b1",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "5",
+ "uniqueId": "0xa95e252cd7a6e54e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x333c966d5f3b3375",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0x5656afec2788d569",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0xad1bf75f50313455",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0xbabaa83bdb6fe877",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0xdee6d87744a36ae",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0x2a954aa975e59d15",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759424668",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1934",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-292",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759165468",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1934",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-292",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "1357871",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-292",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "zpxgtt7ztfwstl3gysdjfozjodgqnvkm"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..a79bb9cbafb2005ea305802a28cb776082c6cc20
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_step":0,"System/Peak GPU Memory (MB)":35614.78125,"_timestamp":1.759165624842901e+09,"_wandb":{"runtime":118},"_runtime":118.9149238}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug-core.log b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..df1a03c8f4db65d039dc4b3c65f77f97fd647b1e
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug-core.log
@@ -0,0 +1,16 @@
+{"time":"2025-09-29T17:05:12.599767574Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpblcfcrd2/port-1358048.txt","pid":1358048,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-29T17:05:12.601129497Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":1358048}
+{"time":"2025-09-29T17:05:12.601134567Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-1358048-1358213-1657513860/socket","Net":"unix"}}
+{"time":"2025-09-29T17:05:12.784034321Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-29T17:05:12.788802342Z","level":"INFO","msg":"handleInformInit: received","streamId":"2lq20p1f","id":"1(@)"}
+{"time":"2025-09-29T17:05:13.92427811Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"2lq20p1f","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.469972392Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"2lq20p1f","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473846438Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"2lq20p1f","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473852038Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473858278Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473865358Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-29T17:07:15.473865688Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473901259Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473904939Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-29T17:07:15.473942719Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-1358048-1358213-1657513860/socket","Net":"unix"}}
+{"time":"2025-09-29T17:07:15.47396587Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..11d44621118f40b354d734fd97a3801bb26d17fb
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2025-09-29T17:05:12.790661794Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-29T17:05:13.924227629Z","level":"INFO","msg":"stream: created new stream","id":"2lq20p1f"}
+{"time":"2025-09-29T17:05:13.92427238Z","level":"INFO","msg":"stream: started","id":"2lq20p1f"}
+{"time":"2025-09-29T17:05:13.92428108Z","level":"INFO","msg":"writer: started","stream_id":"2lq20p1f"}
+{"time":"2025-09-29T17:05:13.924301101Z","level":"INFO","msg":"handler: started","stream_id":"2lq20p1f"}
+{"time":"2025-09-29T17:05:13.924330541Z","level":"INFO","msg":"sender: started","stream_id":"2lq20p1f"}
+{"time":"2025-09-29T17:07:13.196500373Z","level":"INFO","msg":"handler: operation stats","stats":{"operations":[{"desc":"updating run metadata","runtime_seconds":0.001296762}],"total_operations":1}}
+{"time":"2025-09-29T17:07:15.018224116Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-29T17:07:15.470258007Z","level":"INFO","msg":"stream: closing","id":"2lq20p1f"}
+{"time":"2025-09-29T17:07:15.470272437Z","level":"INFO","msg":"handler: closed","stream_id":"2lq20p1f"}
+{"time":"2025-09-29T17:07:15.472123178Z","level":"INFO","msg":"sender: closed","stream_id":"2lq20p1f"}
+{"time":"2025-09-29T17:07:15.472129189Z","level":"INFO","msg":"stream: closed","id":"2lq20p1f"}
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug.log b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/run-2lq20p1f.wandb b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/run-2lq20p1f.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..c35abd43d66a97ec65d3daea60d9fa5e2f9bbc48
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250929_170512-2lq20p1f/run-2lq20p1f.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/config.yaml b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..d2fd9fdcb448ffcf4bb3c69f5f62d8eda4e524e9
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/config.yaml
@@ -0,0 +1,617 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ xw28krbghcx1p9g77lw10b5hvgwgqmco:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50128396288"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: f2afcc15e05f491a8e50add64395fc1db0a1188d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x413935505e32b8da"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x95be8fdc770fcfd7"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x27087f06439a527d"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x82728d7f9bd937e4"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xa0442ab3bdd405c1"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xaabcddaa244a3d6e"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x12140cd9e24f12e9"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x24ee801b7c402006"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-306
+ memory:
+ total: "2434611519488"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759481466"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1970"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-306
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759222266"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1970"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-306
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "589646"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-306
+ topology_addr_pattern: node
+ startedAt: "2025-09-30T08:52:06.337927Z"
+ writerId: xw28krbghcx1p9g77lw10b5hvgwgqmco
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 13
+ - 15
+ - 16
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250930_085126
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/output.log b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..9c4ae3ffc3eb99091b436cc7dbdbe8bc0402d935
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/output.log
@@ -0,0 +1,86 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [08:52:08] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': '/vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha', 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [08:52:19] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+09/30 [08:52:20] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [08:52:26] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [08:52:27] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [08:52:28] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [08:52:29] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [08:52:31] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/30 [08:52:32] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: /vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha
+****** before AgiBotWorldAlpha dataset...
+09/30 [08:52:33] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:485
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/compat/_optional.py", line 135, in import_optional_dependency
+ module = importlib.import_module(name)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/importlib/__init__.py", line 126, in import_module
+ return _bootstrap._gcd_import(name[level:], package, level)
+ File "", line 1050, in _gcd_import
+ File "", line 1027, in _find_and_load
+ File "", line 1004, in _find_and_load_unlocked
+ModuleNotFoundError: No module named 'openpyxl'
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py", line 397, in
+ train(cfg)
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 160, in main
+ train_loader = build_train_dataloader(cfg, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 197, in build_train_dataloader
+ return build_vla_train_dataloader(train_config, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 384, in build_vla_train_dataloader
+ ds = build_agibot_train_dataset(train_config, normalization_type, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 487, in build_agibot_train_dataset
+ dataset = AgiBotWorldAlphaDataset(
+ File "", line 13, in __init__
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 145, in __post_init__
+ self._frame_ranges_map: Optional[Dict[Tuple[str, int], Tuple[int, int]]] = self._load_frame_ranges_excel()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 411, in _load_frame_ranges_excel
+ df = pd.read_excel(excel_path)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_base.py", line 495, in read_excel
+ io = ExcelFile(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_base.py", line 1567, in __init__
+ self._reader = self._engines[engine](
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_openpyxl.py", line 552, in __init__
+ import_optional_dependency("openpyxl")
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/compat/_optional.py", line 138, in import_optional_dependency
+ raise ImportError(msg)
+ImportError: Missing optional dependency 'openpyxl'. Use pip or conda to install openpyxl.
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..291a344502cb68f8ac9dcf95e2d7e2df5f47b9b6
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/requirements.txt
@@ -0,0 +1,284 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..92aad0de7fec10c8e0e8ec90e6a661f7bd5f712b
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T08:52:06.337927Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-306",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50128396288"
+ }
+ },
+ "memory": {
+ "total": "2434611519488"
+ },
+ "gpu_amd": [
+ {
+ "id": "5",
+ "uniqueId": "0x413935505e32b8da",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0x95be8fdc770fcfd7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x27087f06439a527d",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0x82728d7f9bd937e4",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xa0442ab3bdd405c1",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0xaabcddaa244a3d6e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0x12140cd9e24f12e9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0x24ee801b7c402006",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759481466",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1970",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-306",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759222266",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1970",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-306",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "589646",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-306",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "xw28krbghcx1p9g77lw10b5hvgwgqmco"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..2d7b7348866aa5fab906f41ff70c8e018458d95f
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_wandb":{"runtime":25},"_runtime":25}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..faaa15e20c7c1d1dfdc8501804845084e1bcdfc0
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-30T08:52:06.638339714Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpyjj93vky/port-589823.txt","pid":589823,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T08:52:06.641262155Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":589823}
+{"time":"2025-09-30T08:52:06.642250983Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-589823-590216-2576664124/socket","Net":"unix"}}
+{"time":"2025-09-30T08:52:06.792135331Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T08:52:06.805602998Z","level":"INFO","msg":"handleInformInit: received","streamId":"50kj35c0","id":"1(@)"}
+{"time":"2025-09-30T08:52:08.062408989Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"50kj35c0","id":"1(@)"}
+{"time":"2025-09-30T08:52:34.430005899Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-30T08:52:34.430229963Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-30T08:52:34.430227373Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-30T08:52:34.430267094Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-30T08:52:34.430322415Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-589823-590216-2576664124/socket","Net":"unix"}}
+{"time":"2025-09-30T08:52:36.423142352Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-30T08:52:36.423156882Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-30T08:52:36.423167352Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..1f5398aaef6dbbef94aa18f18575fdc1c4aad1f2
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug-internal.log
@@ -0,0 +1,11 @@
+{"time":"2025-09-30T08:52:06.807490411Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T08:52:08.062344498Z","level":"INFO","msg":"stream: created new stream","id":"50kj35c0"}
+{"time":"2025-09-30T08:52:08.062403319Z","level":"INFO","msg":"stream: started","id":"50kj35c0"}
+{"time":"2025-09-30T08:52:08.062425769Z","level":"INFO","msg":"handler: started","stream_id":"50kj35c0"}
+{"time":"2025-09-30T08:52:08.06244382Z","level":"INFO","msg":"sender: started","stream_id":"50kj35c0"}
+{"time":"2025-09-30T08:52:08.06243294Z","level":"INFO","msg":"writer: started","stream_id":"50kj35c0"}
+{"time":"2025-09-30T08:52:34.430242003Z","level":"INFO","msg":"stream: closing","id":"50kj35c0"}
+{"time":"2025-09-30T08:52:36.099290171Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-30T08:52:36.421230078Z","level":"INFO","msg":"handler: closed","stream_id":"50kj35c0"}
+{"time":"2025-09-30T08:52:36.422272446Z","level":"INFO","msg":"sender: closed","stream_id":"50kj35c0"}
+{"time":"2025-09-30T08:52:36.422279596Z","level":"INFO","msg":"stream: closed","id":"50kj35c0"}
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug.log b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..a572bc39e168280fb6c83f9d1eeb52bec35c3516
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/logs/debug.log
@@ -0,0 +1 @@
+2025-09-30 08:52:34,430 INFO wandb-AsyncioManager-main:589823 [service_client.py:_forward_responses():84] Reached EOF.
diff --git a/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/run-50kj35c0.wandb b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/run-50kj35c0.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..3d24940d7b458ba0975dbfc5eca3c9ce112b6b72
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250930_085206-50kj35c0/run-50kj35c0.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/config.yaml b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..ecb4c5eaa0fa5353e84b8b6c3714ddad5dfdd4f4
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/config.yaml
@@ -0,0 +1,617 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ e39r496xjes4qj7ky2l3e9tlyr84a0v3:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50128465920"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: f2afcc15e05f491a8e50add64395fc1db0a1188d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x12140cd9e24f12e9"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x95be8fdc770fcfd7"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xa0442ab3bdd405c1"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x24ee801b7c402006"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xaabcddaa244a3d6e"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x82728d7f9bd937e4"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x413935505e32b8da"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x27087f06439a527d"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-306
+ memory:
+ total: "2434611519488"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759482071"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1973"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-306
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759222871"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1973"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-306
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "594412"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-306
+ topology_addr_pattern: node
+ startedAt: "2025-09-30T09:01:55.101472Z"
+ writerId: e39r496xjes4qj7ky2l3e9tlyr84a0v3
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 13
+ - 15
+ - 16
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250930_090124
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/output.log b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..db1e3d2bf6ab35e4dc4f3055481e93e63d6b2d67
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/output.log
@@ -0,0 +1,85 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [09:01:56] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': None, 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': '/vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha', 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [09:01:58] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [09:02:07] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [09:02:09] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [09:02:10] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [09:02:11] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [09:02:12] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/30 [09:02:13] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+09/30 [09:02:14] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; path not found: None
+****** path: /vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha
+****** before AgiBotWorldAlpha dataset...
+09/30 [09:02:15] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:485
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/compat/_optional.py", line 135, in import_optional_dependency
+ module = importlib.import_module(name)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/importlib/__init__.py", line 126, in import_module
+ return _bootstrap._gcd_import(name[level:], package, level)
+ File "", line 1050, in _gcd_import
+ File "", line 1027, in _find_and_load
+ File "", line 1004, in _find_and_load_unlocked
+ModuleNotFoundError: No module named 'openpyxl'
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py", line 397, in
+ train(cfg)
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 160, in main
+ train_loader = build_train_dataloader(cfg, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 197, in build_train_dataloader
+ return build_vla_train_dataloader(train_config, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 384, in build_vla_train_dataloader
+ ds = build_agibot_train_dataset(train_config, normalization_type, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 487, in build_agibot_train_dataset
+ dataset = AgiBotWorldAlphaDataset(
+ File "", line 13, in __init__
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 145, in __post_init__
+ self._frame_ranges_map: Optional[Dict[Tuple[str, int], Tuple[int, int]]] = self._load_frame_ranges_excel()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 411, in _load_frame_ranges_excel
+ df = pd.read_excel(excel_path)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_base.py", line 495, in read_excel
+ io = ExcelFile(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_base.py", line 1567, in __init__
+ self._reader = self._engines[engine](
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_openpyxl.py", line 552, in __init__
+ import_optional_dependency("openpyxl")
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/compat/_optional.py", line 138, in import_optional_dependency
+ raise ImportError(msg)
+ImportError: Missing optional dependency 'openpyxl'. Use pip or conda to install openpyxl.
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..291a344502cb68f8ac9dcf95e2d7e2df5f47b9b6
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/requirements.txt
@@ -0,0 +1,284 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef09c5afb76a3b20ff17fa87d97ebe18d8dec765
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T09:01:55.101472Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-306",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50128465920"
+ }
+ },
+ "memory": {
+ "total": "2434611519488"
+ },
+ "gpu_amd": [
+ {
+ "id": "6",
+ "uniqueId": "0x12140cd9e24f12e9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0x95be8fdc770fcfd7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xa0442ab3bdd405c1",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0x24ee801b7c402006",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0xaabcddaa244a3d6e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0x82728d7f9bd937e4",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "5",
+ "uniqueId": "0x413935505e32b8da",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x27087f06439a527d",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759482071",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1973",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-306",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759222871",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1973",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-306",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "594412",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-306",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "e39r496xjes4qj7ky2l3e9tlyr84a0v3"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..ba532d630a2f6e9086432ef98a4e1304e8ad3f55
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_runtime":19,"_wandb":{"runtime":19}}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..eb4ed164e5ec7ad85a29c2d981a6e70aab817ef6
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-30T09:01:55.162904331Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmprwjho3ja/port-594589.txt","pid":594589,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T09:01:55.164089452Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":594589}
+{"time":"2025-09-30T09:01:55.164075022Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-594589-594975-1877083663/socket","Net":"unix"}}
+{"time":"2025-09-30T09:01:55.335963017Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T09:01:55.342714526Z","level":"INFO","msg":"handleInformInit: received","streamId":"w2yi62pb","id":"1(@)"}
+{"time":"2025-09-30T09:01:56.270439585Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"w2yi62pb","id":"1(@)"}
+{"time":"2025-09-30T09:02:15.777414547Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-30T09:02:15.777645561Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-30T09:02:15.777654891Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-30T09:02:15.777680572Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-30T09:02:15.778959994Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-594589-594975-1877083663/socket","Net":"unix"}}
+{"time":"2025-09-30T09:02:17.890949248Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-30T09:02:17.890967089Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-30T09:02:17.890978859Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..9a15c9d9796c266949732881b3c2ca4c286b8f5c
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug-internal.log
@@ -0,0 +1,11 @@
+{"time":"2025-09-30T09:01:55.344751712Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T09:01:56.270392104Z","level":"INFO","msg":"stream: created new stream","id":"w2yi62pb"}
+{"time":"2025-09-30T09:01:56.270434675Z","level":"INFO","msg":"stream: started","id":"w2yi62pb"}
+{"time":"2025-09-30T09:01:56.270440165Z","level":"INFO","msg":"writer: started","stream_id":"w2yi62pb"}
+{"time":"2025-09-30T09:01:56.270445835Z","level":"INFO","msg":"handler: started","stream_id":"w2yi62pb"}
+{"time":"2025-09-30T09:01:56.270482246Z","level":"INFO","msg":"sender: started","stream_id":"w2yi62pb"}
+{"time":"2025-09-30T09:02:15.777655041Z","level":"INFO","msg":"stream: closing","id":"w2yi62pb"}
+{"time":"2025-09-30T09:02:17.562495857Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-30T09:02:17.888649918Z","level":"INFO","msg":"handler: closed","stream_id":"w2yi62pb"}
+{"time":"2025-09-30T09:02:17.889765777Z","level":"INFO","msg":"sender: closed","stream_id":"w2yi62pb"}
+{"time":"2025-09-30T09:02:17.889789108Z","level":"INFO","msg":"stream: closed","id":"w2yi62pb"}
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug.log b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..f3f5468fff145d700a46bd183b6314a3c5c2acbd
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/logs/debug.log
@@ -0,0 +1 @@
+2025-09-30 09:02:15,777 INFO wandb-AsyncioManager-main:594589 [service_client.py:_forward_responses():84] Reached EOF.
diff --git a/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/run-w2yi62pb.wandb b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/run-w2yi62pb.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..aee1b3147d9acd103d95275cd9b078df05a4710e
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250930_090155-w2yi62pb/run-w2yi62pb.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/config.yaml b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..b71b52fd7e20a10e4f53101b5646e30ec536c373
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/config.yaml
@@ -0,0 +1,617 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ esn3ubylhv9g1ubwtuf03ejgtuan58ph:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50128633856"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: f2afcc15e05f491a8e50add64395fc1db0a1188d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x27087f06439a527d"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xa0442ab3bdd405c1"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x95be8fdc770fcfd7"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x24ee801b7c402006"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x82728d7f9bd937e4"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x413935505e32b8da"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xaabcddaa244a3d6e"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x12140cd9e24f12e9"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-306
+ memory:
+ total: "2434611519488"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759489779"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1988"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-306
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759230579"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1988"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-306
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "604425"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-306
+ topology_addr_pattern: node
+ startedAt: "2025-09-30T11:10:22.173666Z"
+ writerId: esn3ubylhv9g1ubwtuf03ejgtuan58ph
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 13
+ - 15
+ - 16
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250930_110952
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/output.log b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..4cffad9d38480bae6f92c39855a7d292d0ab7a21
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/output.log
@@ -0,0 +1,86 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [11:10:24] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': '/vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha', 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [11:10:30] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+09/30 [11:10:31] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [11:10:39] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [11:10:45] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [11:10:47] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [11:10:48] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [11:10:50] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+09/30 [11:10:51] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: /vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha
+****** before AgiBotWorldAlpha dataset...
+09/30 [11:10:52] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:485
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/compat/_optional.py", line 135, in import_optional_dependency
+ module = importlib.import_module(name)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/importlib/__init__.py", line 126, in import_module
+ return _bootstrap._gcd_import(name[level:], package, level)
+ File "", line 1050, in _gcd_import
+ File "", line 1027, in _find_and_load
+ File "", line 1004, in _find_and_load_unlocked
+ModuleNotFoundError: No module named 'openpyxl'
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py", line 398, in
+ train(cfg)
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 160, in main
+ train_loader = build_train_dataloader(cfg, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 197, in build_train_dataloader
+ return build_vla_train_dataloader(train_config, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 384, in build_vla_train_dataloader
+ ds = build_agibot_train_dataset(train_config, normalization_type, device)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/__init__.py", line 487, in build_agibot_train_dataset
+ dataset = AgiBotWorldAlphaDataset(
+ File "", line 13, in __init__
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 145, in __post_init__
+ self._frame_ranges_map: Optional[Dict[Tuple[str, int], Tuple[int, int]]] = self._load_frame_ranges_excel()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 411, in _load_frame_ranges_excel
+ df = pd.read_excel(excel_path)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_base.py", line 495, in read_excel
+ io = ExcelFile(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_base.py", line 1567, in __init__
+ self._reader = self._engines[engine](
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/io/excel/_openpyxl.py", line 552, in __init__
+ import_optional_dependency("openpyxl")
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/pandas/compat/_optional.py", line 138, in import_optional_dependency
+ raise ImportError(msg)
+ImportError: Missing optional dependency 'openpyxl'. Use pip or conda to install openpyxl.
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..291a344502cb68f8ac9dcf95e2d7e2df5f47b9b6
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/requirements.txt
@@ -0,0 +1,284 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..06b689a7e69499e3a72f95f8a37bb0d06f408597
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T11:10:22.173666Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-306",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50128633856"
+ }
+ },
+ "memory": {
+ "total": "2434611519488"
+ },
+ "gpu_amd": [
+ {
+ "id": "1",
+ "uniqueId": "0x27087f06439a527d",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xa0442ab3bdd405c1",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0x95be8fdc770fcfd7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0x24ee801b7c402006",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0x82728d7f9bd937e4",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "5",
+ "uniqueId": "0x413935505e32b8da",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0xaabcddaa244a3d6e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0x12140cd9e24f12e9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759489779",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1988",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-306",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759230579",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1988",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-306",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "604425",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-306",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "esn3ubylhv9g1ubwtuf03ejgtuan58ph"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..2809f7f6be2cd810753ae1188aee6dbc55b345f9
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_runtime":29,"_wandb":{"runtime":29}}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..2f7de8e11512cadf2d69c36f1cfa18bc3cb53f9e
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug-core.log
@@ -0,0 +1,14 @@
+{"time":"2025-09-30T11:10:22.224998099Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpodvish0l/port-604602.txt","pid":604602,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T11:10:22.22560875Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":604602}
+{"time":"2025-09-30T11:10:22.22558572Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-604602-604772-538545887/socket","Net":"unix"}}
+{"time":"2025-09-30T11:10:22.408964708Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T11:10:22.415973911Z","level":"INFO","msg":"handleInformInit: received","streamId":"tyi87rpc","id":"1(@)"}
+{"time":"2025-09-30T11:10:23.336017725Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"tyi87rpc","id":"1(@)"}
+{"time":"2025-09-30T11:10:53.09639777Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-30T11:10:53.097801345Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-30T11:10:53.096660235Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-30T11:10:53.097854056Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-30T11:10:53.098180352Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-604602-604772-538545887/socket","Net":"unix"}}
+{"time":"2025-09-30T11:10:55.079824912Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-30T11:10:55.079840642Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-30T11:10:55.079850132Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..dcaff13accce8758b19c1abb75f378c13336c34c
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug-internal.log
@@ -0,0 +1,11 @@
+{"time":"2025-09-30T11:10:22.417648611Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T11:10:23.335963304Z","level":"INFO","msg":"stream: created new stream","id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:23.336011325Z","level":"INFO","msg":"stream: started","id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:23.336024875Z","level":"INFO","msg":"writer: started","stream_id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:23.336032926Z","level":"INFO","msg":"sender: started","stream_id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:23.336095647Z","level":"INFO","msg":"handler: started","stream_id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:53.096617934Z","level":"INFO","msg":"stream: closing","id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:54.757731193Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-30T11:10:55.076398492Z","level":"INFO","msg":"handler: closed","stream_id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:55.078483378Z","level":"INFO","msg":"sender: closed","stream_id":"tyi87rpc"}
+{"time":"2025-09-30T11:10:55.078509419Z","level":"INFO","msg":"stream: closed","id":"tyi87rpc"}
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug.log b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..16ba2eb808b0855afd61d4277ebbfcb90aeb9cf9
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/logs/debug.log
@@ -0,0 +1 @@
+2025-09-30 11:10:53,096 INFO wandb-AsyncioManager-main:604602 [service_client.py:_forward_responses():84] Reached EOF.
diff --git a/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/run-tyi87rpc.wandb b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/run-tyi87rpc.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..fcec9484fd66434121d446b02b769f62db983c47
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250930_111022-tyi87rpc/run-tyi87rpc.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/config.yaml b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..4c8d562f37e05268c721ceadaf57c54657a09e54
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/config.yaml
@@ -0,0 +1,621 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ lwz9u7135qzjvugxcx5lynr52j566znm:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50128723968"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: f2afcc15e05f491a8e50add64395fc1db0a1188d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xaabcddaa244a3d6e"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x82728d7f9bd937e4"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x27087f06439a527d"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x24ee801b7c402006"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x413935505e32b8da"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x95be8fdc770fcfd7"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x12140cd9e24f12e9"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xa0442ab3bdd405c1"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-306
+ memory:
+ total: "2434611519488"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759492424"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1989"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-306
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759233224"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1989"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-306
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "610629"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-306
+ topology_addr_pattern: node
+ startedAt: "2025-09-30T11:54:26.454498Z"
+ writerId: lwz9u7135qzjvugxcx5lynr52j566znm
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 2
+ - 13
+ - 15
+ - 16
+ - 61
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "10":
+ - 19
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250930_115356
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/output.log b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..4c6498c5464eaa9218f7990906e0a0ab83f7ee2c
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/output.log
@@ -0,0 +1,103 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [11:54:28] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': '/vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha', 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [11:54:34] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [11:54:37] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [11:54:38] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [11:54:39] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [11:54:40] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [11:54:41] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/30 [11:54:42] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+09/30 [11:54:43] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: /vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha
+****** before AgiBotWorldAlpha dataset...
+09/30 [11:54:44] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:485
+****** length of the AgiBotWorldAlphaDataset: 1395683
+****** length of the AgiBotWorldAlphaDataset: 1395683
+****** After build vla train dataset...
+****** iterable_sources: [, , , , , , , , ]
+****** Before build mixed iterable dataset...
+****** Build vla train dataloader successfully!
+************************* Build train_dataloader successful!
+************************* Before build_inf_evaluators
+09/30 [11:54:45] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+************************* Build evaluators successful!
+************************* Early exit flags: early_exit=False
+************************* Initialize model successful!
+***** state_dict_path: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924/model.pt
+***** Load checkpoint successful!
+missing keys: ['action_head.model.layer_norm1.weight', 'action_head.model.layer_norm1.bias', 'action_head.model.fc1.weight', 'action_head.model.fc1.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.1.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.1.bias', 'action_head.model.layer_norm2.weight', 'action_head.model.layer_norm2.bias', 'action_head.model.fc2.weight', 'action_head.model.fc2.bias', 'proprio_projector.fc1.weight', 'proprio_projector.fc1.bias', 'proprio_projector.fc2.weight', 'proprio_projector.fc2.bias']
+unexpected keys: []
+************************* Initialize model successful!
+************************* LoRA flags: use_lora=True, lora_llm=False, lora_vit=False, lora_connector=False
+************************* Before add lora to model
+************************* Before FSDP model wrapping
+************************* FSDP model wrapping successful!
+************************* Before building optimizer and scheduler
+************* Before get lora params
+************* After get lora params successfully
+09/30 [11:56:10] INFO | >> Constructing optimizer with 2 param groups optim.py:1283
+**************************************************
+After building optimizer and scheduler and model, before training, peak GPU memory (MB): 35614
+************************* VLATrainer initialized successfully!
+************************* Before trainer.fit()
+Pre-train system metrics
+ System/Peak GPU Memory (MB)=35,614
+!!!Training failed: __enter__
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 593, in main
+ trainer.fit()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 2284, in fit
+ for batch in self.train_loader:
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 734, in __next__
+ data = self._next_data()
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 790, in _next_data
+ data = self._dataset_fetcher.fetch(index) # may raise StopIteration
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/utils/data/_utils/fetch.py", line 33, in fetch
+ data.append(next(self.dataset_iter))
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/iterable_dataset_mixture.py", line 156, in __iter__
+ yield next(source_iters[src_idx])
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/dataset.py", line 111, in __iter__
+ item = self.dataset.get(int(idx), rng)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 350, in get
+ with self._open_h5(set_id, episode_id) as h5:
+AttributeError: __enter__
+wandb: WARNING The `quiet` argument to `wandb.run.finish()` is deprecated, use `wandb.Settings(quiet=...)` to set this instead.
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5a5e971a5d9f3b6f6a75248e203682c5fbcd69e8
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/requirements.txt
@@ -0,0 +1,286 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+et_xmlfile==2.0.0
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+openpyxl==3.1.5
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..b3f30af8129228073f6da90798d7edb75ad90c66
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T11:54:26.454498Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-306",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50128723968"
+ }
+ },
+ "memory": {
+ "total": "2434611519488"
+ },
+ "gpu_amd": [
+ {
+ "id": "2",
+ "uniqueId": "0xaabcddaa244a3d6e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0x82728d7f9bd937e4",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x27087f06439a527d",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0x24ee801b7c402006",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "5",
+ "uniqueId": "0x413935505e32b8da",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0x95be8fdc770fcfd7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0x12140cd9e24f12e9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xa0442ab3bdd405c1",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759492424",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1989",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-306",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759233224",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1989",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-306",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "610629",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-306",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "lwz9u7135qzjvugxcx5lynr52j566znm"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..7080d373195ac5cbd74b89b0b4a01a70bcc29b8d
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_wandb":{"runtime":104},"_runtime":104.60638138,"_step":0,"System/Peak GPU Memory (MB)":35614.78125,"_timestamp":1.7592333707721887e+09}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..c55fdf4dcebf7604e1d3780f20dc28f2baca6c30
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug-core.log
@@ -0,0 +1,16 @@
+{"time":"2025-09-30T11:54:26.521817705Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmprzszz_9w/port-610806.txt","pid":610806,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T11:54:26.522336125Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":610806}
+{"time":"2025-09-30T11:54:26.522321354Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-610806-610974-3557492925/socket","Net":"unix"}}
+{"time":"2025-09-30T11:54:26.705925546Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T11:54:26.712556793Z","level":"INFO","msg":"handleInformInit: received","streamId":"4y0fn6ka","id":"1(@)"}
+{"time":"2025-09-30T11:54:27.766430612Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"4y0fn6ka","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.452974036Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"4y0fn6ka","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.456105611Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"4y0fn6ka","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.483739548Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.483758558Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.483763869Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-30T11:56:14.483789359Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.48383279Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-610806-610974-3557492925/socket","Net":"unix"}}
+{"time":"2025-09-30T11:56:14.483901891Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.483907841Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-30T11:56:14.483922101Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..89122e33061292032942ab7695935c6170d04920
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2025-09-30T11:54:26.714354905Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T11:54:27.766379841Z","level":"INFO","msg":"stream: created new stream","id":"4y0fn6ka"}
+{"time":"2025-09-30T11:54:27.766424922Z","level":"INFO","msg":"stream: started","id":"4y0fn6ka"}
+{"time":"2025-09-30T11:54:27.766455023Z","level":"INFO","msg":"sender: started","stream_id":"4y0fn6ka"}
+{"time":"2025-09-30T11:54:27.766455013Z","level":"INFO","msg":"handler: started","stream_id":"4y0fn6ka"}
+{"time":"2025-09-30T11:54:27.766467603Z","level":"INFO","msg":"writer: started","stream_id":"4y0fn6ka"}
+{"time":"2025-09-30T11:56:12.736999283Z","level":"INFO","msg":"handler: operation stats","stats":{"operations":[{"desc":"updating run metadata","runtime_seconds":1.9632518270000001}],"total_operations":1}}
+{"time":"2025-09-30T11:56:13.795981592Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-30T11:56:14.453298712Z","level":"INFO","msg":"stream: closing","id":"4y0fn6ka"}
+{"time":"2025-09-30T11:56:14.453319212Z","level":"INFO","msg":"handler: closed","stream_id":"4y0fn6ka"}
+{"time":"2025-09-30T11:56:14.454730867Z","level":"INFO","msg":"sender: closed","stream_id":"4y0fn6ka"}
+{"time":"2025-09-30T11:56:14.454741967Z","level":"INFO","msg":"stream: closed","id":"4y0fn6ka"}
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug.log b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/run-4y0fn6ka.wandb b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/run-4y0fn6ka.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..6a97d89b06d298e670806f72900c4be302ebf45e
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250930_115426-4y0fn6ka/run-4y0fn6ka.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/config.yaml b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..5900e54610df9e31d8e1a4cd62cef55a64fc7a50
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/config.yaml
@@ -0,0 +1,621 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ np3futv3ypcdn5zstcz9ip385cmxztpr:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50128797696"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: f2afcc15e05f491a8e50add64395fc1db0a1188d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x27087f06439a527d"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x24ee801b7c402006"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xaabcddaa244a3d6e"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x82728d7f9bd937e4"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x95be8fdc770fcfd7"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xa0442ab3bdd405c1"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x12140cd9e24f12e9"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x413935505e32b8da"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-306
+ memory:
+ total: "2434611519488"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759493653"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1990"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-306
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759234453"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1990"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-306
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "627093"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-306
+ topology_addr_pattern: node
+ startedAt: "2025-09-30T12:14:56.412445Z"
+ writerId: np3futv3ypcdn5zstcz9ip385cmxztpr
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 2
+ - 13
+ - 15
+ - 16
+ - 61
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "10":
+ - 19
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250930_121426
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/output.log b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..506b7ae0552729d61b64efd7e0c0ce8cd8548c70
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/output.log
@@ -0,0 +1,103 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [12:14:58] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': '/vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha', 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [12:15:04] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+09/30 [12:15:05] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [12:15:13] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [12:15:14] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [12:15:15] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [12:15:16] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [12:15:17] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/30 [12:15:18] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+09/30 [12:15:19] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: /vast/users/xiaodan/zhangjian/datasets/AgiBotWorld-Alpha
+****** before AgiBotWorldAlpha dataset...
+09/30 [12:15:20] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:485
+****** length of the AgiBotWorldAlphaDataset: 1395683
+****** length of the AgiBotWorldAlphaDataset: 1395683
+****** After build vla train dataset...
+****** iterable_sources: [, , , , , , , , ]
+****** Before build mixed iterable dataset...
+****** Build vla train dataloader successfully!
+************************* Build train_dataloader successful!
+************************* Before build_inf_evaluators
+ WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+************************* Build evaluators successful!
+************************* Early exit flags: early_exit=False
+************************* Initialize model successful!
+***** state_dict_path: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924/model.pt
+***** Load checkpoint successful!
+missing keys: ['action_head.model.layer_norm1.weight', 'action_head.model.layer_norm1.bias', 'action_head.model.fc1.weight', 'action_head.model.fc1.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.1.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.1.bias', 'action_head.model.layer_norm2.weight', 'action_head.model.layer_norm2.bias', 'action_head.model.fc2.weight', 'action_head.model.fc2.bias', 'proprio_projector.fc1.weight', 'proprio_projector.fc1.bias', 'proprio_projector.fc2.weight', 'proprio_projector.fc2.bias']
+unexpected keys: []
+************************* Initialize model successful!
+************************* LoRA flags: use_lora=True, lora_llm=False, lora_vit=False, lora_connector=False
+************************* Before add lora to model
+************************* Before FSDP model wrapping
+************************* FSDP model wrapping successful!
+************************* Before building optimizer and scheduler
+************* Before get lora params
+************* After get lora params successfully
+09/30 [12:16:34] INFO | >> Constructing optimizer with 2 param groups optim.py:1283
+**************************************************
+After building optimizer and scheduler and model, before training, peak GPU memory (MB): 35614
+************************* VLATrainer initialized successfully!
+************************* Before trainer.fit()
+Pre-train system metrics
+ System/Peak GPU Memory (MB)=35,614
+!!!Training failed: __enter__
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 593, in main
+ trainer.fit()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 2284, in fit
+ for batch in self.train_loader:
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 734, in __next__
+ data = self._next_data()
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 790, in _next_data
+ data = self._dataset_fetcher.fetch(index) # may raise StopIteration
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/utils/data/_utils/fetch.py", line 33, in fetch
+ data.append(next(self.dataset_iter))
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/iterable_dataset_mixture.py", line 156, in __iter__
+ yield next(source_iters[src_idx])
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/dataset.py", line 111, in __iter__
+ item = self.dataset.get(int(idx), rng)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/data/vla/agibot_dataset.py", line 350, in get
+ with self._open_h5(set_id, episode_id) as h5:
+AttributeError: __enter__
+wandb: WARNING The `quiet` argument to `wandb.run.finish()` is deprecated, use `wandb.Settings(quiet=...)` to set this instead.
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5a5e971a5d9f3b6f6a75248e203682c5fbcd69e8
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/requirements.txt
@@ -0,0 +1,286 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+et_xmlfile==2.0.0
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+openpyxl==3.1.5
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..b1cf7c08c3085cc327b55a5a47ae80e71639e12f
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T12:14:56.412445Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-306",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50128797696"
+ }
+ },
+ "memory": {
+ "total": "2434611519488"
+ },
+ "gpu_amd": [
+ {
+ "id": "1",
+ "uniqueId": "0x27087f06439a527d",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0x24ee801b7c402006",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0xaabcddaa244a3d6e",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0x82728d7f9bd937e4",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0x95be8fdc770fcfd7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xa0442ab3bdd405c1",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0x12140cd9e24f12e9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "5",
+ "uniqueId": "0x413935505e32b8da",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759493653",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1990",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-306",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759234453",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1990",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-306",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "627093",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-306",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "np3futv3ypcdn5zstcz9ip385cmxztpr"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..0b21b78871c4f850d046142ad98d6a73df4542da
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/files/wandb-summary.json
@@ -0,0 +1 @@
+{"_wandb":{"runtime":98},"_runtime":98.460616905,"_timestamp":1.7592345947656755e+09,"_step":0,"System/Peak GPU Memory (MB)":35614.78125}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..29809d1714c841fc7a1e1aef9cc5ba3bc5e0c5dd
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug-core.log
@@ -0,0 +1,16 @@
+{"time":"2025-09-30T12:14:56.463776788Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpak81ny0c/port-627270.txt","pid":627270,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T12:14:56.464292447Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":627270}
+{"time":"2025-09-30T12:14:56.464288217Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-627270-627439-1346290193/socket","Net":"unix"}}
+{"time":"2025-09-30T12:14:56.647255937Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T12:14:56.65366928Z","level":"INFO","msg":"handleInformInit: received","streamId":"q6s75me3","id":"1(@)"}
+{"time":"2025-09-30T12:14:57.578834615Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"q6s75me3","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.901143997Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"q6s75me3","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.904637269Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"q6s75me3","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.904643079Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.904651619Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.904661199Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-30T12:16:37.904660139Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.90470073Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.90470506Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-30T12:16:37.904758371Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-627270-627439-1346290193/socket","Net":"unix"}}
+{"time":"2025-09-30T12:16:37.904779601Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..d86c456286adc102e15c8133d669b7ccfe327259
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2025-09-30T12:14:56.655582724Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T12:14:57.578785874Z","level":"INFO","msg":"stream: created new stream","id":"q6s75me3"}
+{"time":"2025-09-30T12:14:57.578828744Z","level":"INFO","msg":"stream: started","id":"q6s75me3"}
+{"time":"2025-09-30T12:14:57.578853335Z","level":"INFO","msg":"writer: started","stream_id":"q6s75me3"}
+{"time":"2025-09-30T12:14:57.578853335Z","level":"INFO","msg":"handler: started","stream_id":"q6s75me3"}
+{"time":"2025-09-30T12:14:57.578883645Z","level":"INFO","msg":"sender: started","stream_id":"q6s75me3"}
+{"time":"2025-09-30T12:16:36.433285201Z","level":"INFO","msg":"handler: operation stats","stats":{"operations":[{"desc":"updating run metadata","runtime_seconds":1.666228789}],"total_operations":1}}
+{"time":"2025-09-30T12:16:37.447855539Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-30T12:16:37.901604365Z","level":"INFO","msg":"stream: closing","id":"q6s75me3"}
+{"time":"2025-09-30T12:16:37.901626916Z","level":"INFO","msg":"handler: closed","stream_id":"q6s75me3"}
+{"time":"2025-09-30T12:16:37.903184523Z","level":"INFO","msg":"sender: closed","stream_id":"q6s75me3"}
+{"time":"2025-09-30T12:16:37.903192913Z","level":"INFO","msg":"stream: closed","id":"q6s75me3"}
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug.log b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/run-q6s75me3.wandb b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/run-q6s75me3.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..b3edfb0a38469ef79c48ea54a2f7fb48f39262ac
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250930_121456-q6s75me3/run-q6s75me3.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/config.yaml b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..567aa0b7dc6c0873ec5703d649c56dac7ebc5b37
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/config.yaml
@@ -0,0 +1,621 @@
+_wandb:
+ value:
+ cli_version: 0.21.4
+ e:
+ ddpjl58x45oxalea5he84xtx2t0hmwvi:
+ args:
+ - qwen2_7b
+ - save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+ - --vision_backbone
+ - openai
+ - --action_head
+ - l1_regression
+ - --seq_len
+ - "1600"
+ - --ft_llm
+ - --checkpoint
+ - /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+ - --device_train_microbatch_size
+ - "16"
+ - --global_batch_size
+ - "126"
+ - --dataset
+ - vla_dataset_realworld
+ - --llm_learning_rate
+ - "5e-5"
+ - --wandb_entity
+ - henryeap
+ - --wandb_project
+ - a1-realworld
+ - --wandb_run_name
+ - realworld
+ - --save_overwrite
+ codePath: launch_scripts/train_vla.py
+ codePathLocal: launch_scripts/train_vla.py
+ cpu_count: 64
+ cpu_count_logical: 128
+ disk:
+ /:
+ total: "470343073792"
+ used: "50365325312"
+ email: ihenrykwok@outlook.com
+ executable: /vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10
+ git:
+ commit: f2afcc15e05f491a8e50add64395fc1db0a1188d
+ remote: https://github.com/Spatialtemporal-AI/A1.git
+ gpu: Instinct MI210
+ gpu_amd:
+ - id: "5"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x55366988951d07ad"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "2"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x433fd00772eb3fb7"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "7"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xec87d8b9ba4d4aa9"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "3"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xdaf8107e479aecf0"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "4"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xe7ce40a1ea1cbac2"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "1"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x1633d5497b099b7f"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "0"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0xef820a5c24b0c740"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ - id: "6"
+ maxPower: "300.0"
+ mclkRange: 400Mhz - 1600Mhz
+ model: "0x740f"
+ performanceLevel: auto
+ sclkRange: 500Mhz - 1700Mhz
+ series: Instinct MI210
+ sku: D67301V
+ uniqueId: "0x38b61e086b79539"
+ vbiosVersion: 113-D67301V-073
+ vendor: Advanced Micro Devices, Inc. [AMD/ATI]
+ gpu_count: 8
+ host: auh7-1b-gpu-266
+ memory:
+ total: "2434606915584"
+ os: Linux-5.15.0-140-generic-x86_64-with-glibc2.35
+ program: /vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py
+ python: CPython 3.10.18
+ root: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb
+ slurm:
+ cluster_name: ai-04r
+ conf: /etc/slurm/slurm.conf
+ cpus_on_node: "128"
+ gpus_on_node: "8"
+ gtids: "0"
+ job_account: faculty-acc
+ job_cpus_per_node: "128"
+ job_end_time: "1759495098"
+ job_gid: "2000"
+ job_gpus: 0,1,2,3,4,5,6,7
+ job_id: "1992"
+ job_name: mh_realworld
+ job_nodelist: auh7-1b-gpu-266
+ job_num_nodes: "1"
+ job_partition: faculty
+ job_qos: xdqos
+ job_start_time: "1759235898"
+ job_uid: "2013"
+ job_user: xiaodan
+ jobid: "1992"
+ localid: "0"
+ nnodes: "1"
+ nodeid: "0"
+ nodelist: auh7-1b-gpu-266
+ nprocs: "1"
+ ntasks: "1"
+ ntasks_per_node: "1"
+ oom_kill_step: "0"
+ prio_process: "0"
+ procid: "0"
+ submit_dir: /vast/users/xiaodan/zhangjian/A1/launch_scripts
+ submit_host: auh-1b-cpu-login-001
+ task_pid: "3348477"
+ tasks_per_node: "1"
+ topology_addr: auh7-1b-gpu-266
+ topology_addr_pattern: node
+ startedAt: "2025-09-30T12:39:35.212827Z"
+ writerId: ddpjl58x45oxalea5he84xtx2t0hmwvi
+ m: []
+ python_version: 3.10.18
+ t:
+ "1":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "2":
+ - 1
+ - 2
+ - 3
+ - 5
+ - 11
+ - 41
+ - 49
+ - 51
+ - 53
+ - 63
+ - 71
+ - 83
+ - 95
+ - 105
+ "3":
+ - 2
+ - 13
+ - 15
+ - 16
+ - 61
+ "4": 3.10.18
+ "5": 0.21.4
+ "6": 4.56.1
+ "10":
+ - 19
+ "12": 0.21.4
+ "13": linux-x86_64
+activation_checkpointing:
+ value: whole_layer
+allow_resume:
+ value: false
+batch_divisor:
+ value: global_batch
+canceled_check_interval:
+ value: 50
+checkpoint_dir:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+compile:
+ value: null
+console_log_interval:
+ value: 1
+data:
+ value:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: null
+ lerobot_episode_index_start: null
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: false
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 100000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: 95818
+ sequence_length: 1600
+ shuffle: true
+ shuffle_messages: false
+ split: train
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+device_eval_batch_size:
+ value: 4
+device_inf_eval_batch_size:
+ value: 16
+device_train_batch_size:
+ value: 15
+device_train_grad_accum:
+ value: 0
+device_train_microbatch_size:
+ value: 16
+dry_run:
+ value: false
+early_exit:
+ value: false
+epoch:
+ value: null
+eval_interval:
+ value: 0
+eval_on_load:
+ value: false
+eval_subset_num_batches:
+ value: -1
+evaluators:
+ value:
+ - data:
+ dataset: vla_dataset_realworld
+ drop_last: true
+ for_inference: false
+ lerobot_episode_index_end: 765
+ lerobot_episode_index_start: 353
+ mixture: null
+ multi_modal: torch
+ num_workers: 0
+ pad: to_max
+ persistent_workers: true
+ pin_memory: true
+ prefetch_factor: null
+ rlds_data_root_dir: /mnt/data/zhangjian/dataset/Simulation/datasets--openvla--modified_libero_rlds
+ rlds_dataset_name: libero_4_task_suites_no_noops
+ rlds_read_threads: 8
+ rlds_shuffle_buffer_size: 256000
+ rlds_traj_threads: 8
+ root_size_mixture: null
+ seed: null
+ sequence_length: 1600
+ shuffle: false
+ shuffle_messages: false
+ split: validation
+ timeout: 0
+ use_proprio: true
+ use_wrist_image: true
+ device_eval_batch_size: null
+ eval_name: null
+ label: val
+ max_examples: null
+ max_new_tokens: 448
+ mm_evaluator: null
+ save_dir: null
+ save_to_checkpoint_dir: false
+ skip_if_metrics_cached: true
+ subset_num_batches: 64
+extra_steps_after_cancel:
+ value: 10
+fast_forward_batches:
+ value: null
+force_save_unsharded:
+ value: false
+fsdp:
+ value:
+ hybrid_sharding_num_model_replicas: null
+ precision: float
+ sharding_strategy: FULL_SHARD
+ use_orig_params: true
+ wrapping_strategy: by_block_and_size
+ft_connector:
+ value: false
+ft_embedding:
+ value: lm_head
+ft_llm:
+ value: true
+ft_vit:
+ value: false
+fused_loss:
+ value: null
+gen1_gc_interval:
+ value: 1
+global_train_batch_size:
+ value: 126
+inf_eval_interval:
+ value: -1
+inf_eval_subset_num_batches:
+ value: -1
+inf_evaluators:
+ value: []
+initial_model_checkpoint:
+ value: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924
+keep_lr_on_load:
+ value: true
+load_model_config:
+ value: null
+load_path:
+ value: null
+load_path_sharded_checkpointer:
+ value: null
+lora:
+ value: false
+lora_connector:
+ value: false
+lora_llm:
+ value: false
+lora_rank:
+ value: 8
+lora_vit:
+ value: false
+max_duration:
+ value: 500000
+max_grad_norm:
+ value: 1
+max_grad_norm_ratio:
+ value: null
+model:
+ value:
+ action_dim: 7
+ action_head: l1_regression
+ action_head_dit_depth: 28
+ action_head_dit_hidden_size: 1152
+ action_head_dit_num_heads: 16
+ action_tokenizer:
+ identifier: physical-intelligence/fast
+ tokenizer_dir: null
+ action_use_left_eef: false
+ action_use_mobile_base: false
+ activation_type: swiglu
+ additional_vocab_size: 128
+ always_start_with_space: true
+ attention_dropout: 0
+ attention_layer_norm: false
+ attention_layer_norm_with_affine: true
+ attention_type: sdpa
+ bias_for_layer_norm: null
+ block_group_size: 1
+ block_type: sequential
+ clip_qkv: null
+ crop_mode: overlap-and-resize-c2
+ d_model: 3584
+ default_inference_len: 65
+ embedding_dropout: 0
+ embedding_size: 152064
+ ff_out_size: 0
+ fix_image_padding: true
+ float32_attention: true
+ head_dim: null
+ horizon: 8
+ image_feature_dropout: 0
+ image_padding_embed: pad_and_partial_pad
+ image_pooling_2d: attention_meanq
+ image_pooling_h: 2
+ image_pooling_w: 2
+ image_projector: mlp
+ include_bias: false
+ init_cutoff_factor: null
+ init_device: null
+ init_fn: normal
+ init_std: 0.02
+ initializer_range: 0.02
+ layer_norm_eps: 1e-06
+ layer_norm_type: rms
+ layer_norm_with_affine: true
+ llm_causal_attention: false
+ llm_load_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen2-7b.pt
+ low_cpu_fsdp: true
+ max_crops: 12
+ max_position_embeddings: null
+ max_sequence_length: 4096
+ message_formatting: role
+ mlp_hidden_size: 37888
+ mlp_ratio: 4
+ moe_capacity_factor: 1.25
+ moe_dropless: true
+ moe_interleave: false
+ moe_lbl_in_fp32: false
+ moe_log_expert_assignment: false
+ moe_loss_weight: 0.1
+ moe_mlp_impl: sparse
+ moe_num_experts: 8
+ moe_shared_expert: false
+ moe_top_k: 2
+ moe_zloss_weight: null
+ multi_annotation_weighting: root_subsegments
+ n_heads: 28
+ n_kv_heads: 4
+ n_layers: 28
+ new_embedding_init_range: 0.02
+ norm_after: false
+ normalize_input_embeds: false
+ num_diffusion_inference_steps: 30
+ num_diffusion_steps: 1000
+ overlap_margins:
+ - 4
+ - 4
+ pad_tokenizer: true
+ pad_value: 0
+ precision: amp_bf16
+ prompt_type: uber_model
+ qkv_bias: true
+ residual_dropout: 0.1
+ response_residual_dropout: 0
+ rope: true
+ rope_full_precision: true
+ rope_theta: 1e+06
+ scale_logits: false
+ system_prompt_kind: demo_or_style
+ tokenizer:
+ identifier: Qwen/Qwen2-7B
+ tokenizer_dir: null
+ use_col_tokens: true
+ use_position_ids: true
+ use_proprio: true
+ vision_backbone:
+ attention_dropout: 0
+ fsdp_wrap: false
+ image_default_input_size:
+ - 336
+ - 336
+ image_dropout_rate: 0
+ image_emb_dim: 1024
+ image_head_dim: 64
+ image_mlp_activations: quick_gelu
+ image_mlp_dim: 4096
+ image_model_type: openai
+ image_norm_eps: 1e-05
+ image_num_heads: 16
+ image_num_key_value_heads: 16
+ image_num_layers: 23
+ image_num_pos: 577
+ image_patch_size: 14
+ image_pos_patch_size: 14
+ initializer_range: 0.02
+ residual_dropout: 0
+ resize_mode: default
+ vit_layers:
+ - -2
+ - -9
+ vit_load_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/vit-l-14-336.pt
+ vocab_size: 152064
+ weight_tying: false
+multi_component_grad_norm:
+ value: true
+no_pre_train_checkpoint:
+ value: true
+optimizer:
+ value:
+ betas:
+ - 0.9
+ - 0.95
+ connector_betas:
+ - 0.9
+ - 0.95
+ connector_eps: 1e-06
+ connector_learning_rate: 0.0002
+ connector_weight_decay: 0
+ eps: 1e-05
+ learning_rate: 0.0001
+ llm_betas:
+ - 0.9
+ - 0.95
+ llm_eps: 1e-06
+ llm_learning_rate: 5e-05
+ llm_weight_decay: 0
+ metrics_log_interval: 20
+ name: adamw
+ vit_betas:
+ - 0.9
+ - 0.95
+ vit_eps: 1e-06
+ vit_learning_rate: 6e-06
+ vit_weight_decay: 0
+ weight_decay: 0.01
+precision:
+ value: amp_bf16
+python_profiling:
+ value: false
+remote_save_folder:
+ value: null
+reset_dataloader_state:
+ value: false
+reset_optimizer_state:
+ value: false
+reset_trainer_state:
+ value: false
+restore_dataloader:
+ value: true
+run_name:
+ value: realworld_20250930_123854
+save_dataloader_state:
+ value: false
+save_folder:
+ value: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1
+save_interval:
+ value: 500
+save_interval_action_head:
+ value: 500
+save_interval_ephemeral:
+ value: null
+save_interval_unsharded:
+ value: 500
+save_num_action_head_checkpoints_to_keep:
+ value: 2
+save_num_checkpoints_to_keep:
+ value: 1
+save_num_unsharded_checkpoints_to_keep:
+ value: 1
+save_overwrite:
+ value: true
+scheduler:
+ value:
+ alpha_f: 0.1
+ connector_t_warmup: 200
+ grad_clip_warmup_factor: null
+ grad_clip_warmup_steps: null
+ llm_t_warmup: 2000
+ name: multimodal
+ t_max: null
+ t_warmup: 100
+ units: steps
+ vit_t_warmup: 2000
+ warmup_min_lr: 0
+seed:
+ value: 6198
+sharded_checkpointer:
+ value: torch_legacy
+softmax_auxiliary_loss:
+ value: true
+softmax_auxiliary_loss_scale:
+ value: 0.0001
+speed_monitor:
+ value:
+ gpu_flops_available: null
+ window_size: 20
+stop_after:
+ value: null
+stop_at:
+ value: 500000
+time_limit:
+ value: null
+torch_profiling:
+ value: false
+train_exit_random_layer:
+ value: false
+use_lora:
+ value: true
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/output.log b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..585b382786b5a8abbabdb5e699595af70046f329
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/output.log
@@ -0,0 +1,122 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [12:39:37] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': None, 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [12:39:43] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [12:39:52] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [12:39:53] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [12:39:54] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [12:39:55] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [12:39:57] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/30 [12:39:58] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: None
+****** Skip AgiBotWorld-Alpha open-source-real-world; path not found: None
+****** After build vla train dataset...
+****** iterable_sources: [, , , , , , , ]
+****** Before build mixed iterable dataset...
+****** Build vla train dataloader successfully!
+************************* Build train_dataloader successful!
+************************* Before build_inf_evaluators
+09/30 [12:39:59] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+************************* Build evaluators successful!
+************************* Early exit flags: early_exit=False
+************************* Initialize model successful!
+***** state_dict_path: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924/model.pt
+***** Load checkpoint successful!
+missing keys: ['action_head.model.layer_norm1.weight', 'action_head.model.layer_norm1.bias', 'action_head.model.fc1.weight', 'action_head.model.fc1.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.1.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.1.bias', 'action_head.model.layer_norm2.weight', 'action_head.model.layer_norm2.bias', 'action_head.model.fc2.weight', 'action_head.model.fc2.bias', 'proprio_projector.fc1.weight', 'proprio_projector.fc1.bias', 'proprio_projector.fc2.weight', 'proprio_projector.fc2.bias']
+unexpected keys: []
+************************* Initialize model successful!
+************************* LoRA flags: use_lora=True, lora_llm=False, lora_vit=False, lora_connector=False
+************************* Before add lora to model
+************************* Before FSDP model wrapping
+************************* FSDP model wrapping successful!
+************************* Before building optimizer and scheduler
+************* Before get lora params
+************* After get lora params successfully
+09/30 [12:41:26] INFO | >> Constructing optimizer with 2 param groups optim.py:1283
+**************************************************
+After building optimizer and scheduler and model, before training, peak GPU memory (MB): 35614
+************************* VLATrainer initialized successfully!
+************************* Before trainer.fit()
+Pre-train system metrics
+ System/Peak GPU Memory (MB)=35,614
+ WARNING | >> /vast/users/xiaodan/zhangjian/A1/olmo/data/collator.py:200: UserWarning: To copy construct from a tensor, it is recommended to use warnings.py:109
+ sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor).
+ timestep_list = [torch.tensor(ex["timestep"], dtype=torch.int64) for ex in batch]
+
+!!!Training failed: Given normalized_shape=[57344], expected input with shape [*, 57344], but got input of size[15, 8, 28672]
+Traceback (most recent call last):
+ File "/vast/users/xiaodan/zhangjian/A1/scripts/train_for_action.py", line 593, in main
+ trainer.fit()
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 2331, in fit
+ metrics = self.train_step(batch, reduce_global_loss=should_log_this_step)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 2038, in train_step
+ ce_batch_loss, z_batch_loss, batch_accuracy, lb_batch_loss, moe_z_batch_loss, expert_assignments,action_loss = self.train_batch(batch)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 1956, in train_batch
+ accuracy, ce_loss, z_loss, logits, action_loss = self.model_forward(
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/train.py", line 1813, in model_forward
+ outputs = self.fsdp_model.forward(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py", line 854, in forward
+ output = self._fsdp_wrapped_module(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
+ return self._call_impl(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
+ return forward_call(*args, **kwargs)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/vla/affordvla.py", line 1042, in forward
+ predicted_actions = self.action_head.predict_action(action_hidden_states)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/vla/action_heads.py", line 249, in predict_action
+ action = self.model(rearranged_actions_hidden_states)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
+ return self._call_impl(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
+ return forward_call(*args, **kwargs)
+ File "/vast/users/xiaodan/zhangjian/A1/olmo/vla/action_heads.py", line 210, in forward
+ x = self.layer_norm1(x) # shape: (batch_size, input_dim)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1773, in _wrapped_call_impl
+ return self._call_impl(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1784, in _call_impl
+ return forward_call(*args, **kwargs)
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/modules/normalization.py", line 217, in forward
+ return F.layer_norm(
+ File "/vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/nn/functional.py", line 2905, in layer_norm
+ return torch.layer_norm(
+RuntimeError: Given normalized_shape=[57344], expected input with shape [*, 57344], but got input of size[15, 8, 28672]
+wandb: WARNING The `quiet` argument to `wandb.run.finish()` is deprecated, use `wandb.Settings(quiet=...)` to set this instead.
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5a5e971a5d9f3b6f6a75248e203682c5fbcd69e8
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/requirements.txt
@@ -0,0 +1,286 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+et_xmlfile==2.0.0
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+openpyxl==3.1.5
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..5762a7bb91d928d9a80f47410b88ca55fb42a8a6
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T12:39:35.212827Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-266",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50365325312"
+ }
+ },
+ "memory": {
+ "total": "2434606915584"
+ },
+ "gpu_amd": [
+ {
+ "id": "5",
+ "uniqueId": "0x55366988951d07ad",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0x433fd00772eb3fb7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xec87d8b9ba4d4aa9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0xdaf8107e479aecf0",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0xe7ce40a1ea1cbac2",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x1633d5497b099b7f",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0xef820a5c24b0c740",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0x38b61e086b79539",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759495098",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "1992",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-266",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759235898",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "1992",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-266",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "3348477",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-266",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "ddpjl58x45oxalea5he84xtx2t0hmwvi"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/wandb-summary.json b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/wandb-summary.json
new file mode 100644
index 0000000000000000000000000000000000000000..2845286b5b86f811ffeb1607ee5b370273cdb1d3
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/files/wandb-summary.json
@@ -0,0 +1 @@
+{"System/Peak GPU Memory (MB)":35614.78125,"_wandb":{"runtime":117},"_runtime":117.495642133,"_timestamp":1.7592360860861056e+09,"_step":0}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..7da1929de6f736f50f9ef711615d13821c5a9f9e
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug-core.log
@@ -0,0 +1,16 @@
+{"time":"2025-09-30T12:39:35.531288321Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpfxbw7uln/port-3348659.txt","pid":3348659,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T12:39:35.533267395Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3348659}
+{"time":"2025-09-30T12:39:35.534759111Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3348659-3348834-3851720927/socket","Net":"unix"}}
+{"time":"2025-09-30T12:39:35.667946335Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T12:39:35.680795333Z","level":"INFO","msg":"handleInformInit: received","streamId":"on919ofh","id":"1(@)"}
+{"time":"2025-09-30T12:39:37.028776178Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"on919ofh","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.891520431Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"on919ofh","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894745416Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"on919ofh","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894752626Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894759416Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894768886Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-30T12:41:36.894767326Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894807567Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894811637Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2025-09-30T12:41:36.894869608Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-3348659-3348834-3851720927/socket","Net":"unix"}}
+{"time":"2025-09-30T12:41:36.894890678Z","level":"INFO","msg":"server is closed"}
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..0361c0ad20a583e125b493480b5bf29a6b61525a
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug-internal.log
@@ -0,0 +1,12 @@
+{"time":"2025-09-30T12:39:35.682847778Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T12:39:37.028720157Z","level":"INFO","msg":"stream: created new stream","id":"on919ofh"}
+{"time":"2025-09-30T12:39:37.028770978Z","level":"INFO","msg":"stream: started","id":"on919ofh"}
+{"time":"2025-09-30T12:39:37.028806479Z","level":"INFO","msg":"writer: started","stream_id":"on919ofh"}
+{"time":"2025-09-30T12:39:37.028813859Z","level":"INFO","msg":"handler: started","stream_id":"on919ofh"}
+{"time":"2025-09-30T12:39:37.028829659Z","level":"INFO","msg":"sender: started","stream_id":"on919ofh"}
+{"time":"2025-09-30T12:41:34.878202555Z","level":"INFO","msg":"handler: operation stats","stats":{"operations":[{"desc":"updating run metadata","runtime_seconds":0.001991154}],"total_operations":1}}
+{"time":"2025-09-30T12:41:36.521701684Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-30T12:41:36.891885208Z","level":"INFO","msg":"stream: closing","id":"on919ofh"}
+{"time":"2025-09-30T12:41:36.891903148Z","level":"INFO","msg":"handler: closed","stream_id":"on919ofh"}
+{"time":"2025-09-30T12:41:36.893444534Z","level":"INFO","msg":"sender: closed","stream_id":"on919ofh"}
+{"time":"2025-09-30T12:41:36.893451044Z","level":"INFO","msg":"stream: closed","id":"on919ofh"}
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug.log b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/all_l1/wandb/wandb/run-20250930_123935-on919ofh/run-on919ofh.wandb b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/run-on919ofh.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..e87d6aa220ca0a894a8c1eac0b724132e5fa6348
Binary files /dev/null and b/all_l1/wandb/wandb/run-20250930_123935-on919ofh/run-on919ofh.wandb differ
diff --git a/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/output.log b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/output.log
new file mode 100644
index 0000000000000000000000000000000000000000..1bb88f27b27896eb4b02da358511d15067d0b892
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/output.log
@@ -0,0 +1,45224 @@
+wandb: Detected [openai] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+09/30 [13:12:52] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+****** vla_cfg: {'datasets': {'rlds': {'name': None, 'path': None, 'weight': 1.0, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue', 1, 'bounds'], ['/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe', 1, 'bounds']], 'open-source-real-world': {'rlds': {'name': 'a1_real_world', 'path': '/vast/users/xiaodan/zhangjian/datasets/OXE', 'weight': 8, 'action_proprio_normalization_type': 'bounds_q99', 'image_augmentation': False}, 'lerobot': [], 'agibot': {'path': None, 'weight': 8, 'action_proprio_normalization_type': None}}}, 'model': {'action_head': {'action_dim': 16, 'proprio_dim': 16, 'num_actions_chunk': 8, 'action_tokens_mapping': {'left_end_effector': 8, 'right_end_effector': 8}, 'use_left_eef': True, 'use_mobile_base': False}}}
+****** Skip RLDS main; path not found: None
+****** start build LeRobot main...
+build_tokenizer, cache_dir None tokenizer_dir None
+09/30 [13:12:53] INFO | >> Padding tokenizer with 418 tokens tokenizer.py:130
+09/30 [13:12:54] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk
+****** length of the dataset: 72641
+09/30 [13:13:02] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_CleanDesk50
+****** length of the dataset: 27906
+09/30 [13:13:09] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Eraser
+****** length of the dataset: 13441
+09/30 [13:13:10] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Glue
+****** length of the dataset: 10316
+09/30 [13:13:11] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Pen
+****** length of the dataset: 17131
+09/30 [13:13:12] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_StickyNote
+****** length of the dataset: 15765
+09/30 [13:13:13] INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Test_Glue
+****** length of the dataset: 90
+ INFO | >> Loading train dataset: vla_dataset_realworld/train __init__.py:435
+****** before LeRobot dataset...
+****** data_config.rlds_data_root_dir: /vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/dataset/realworld/Lerobot_Wipe
+****** length of the dataset: 18397
+****** Skip RLDS open-source-real-world; mixture 'a1_real_world' not found under: /vast/users/xiaodan/zhangjian/datasets/OXE
+****** Expect one of: []
+****** path: None
+****** Skip AgiBotWorld-Alpha open-source-real-world; path not found: None
+****** After build vla train dataset...
+****** iterable_sources: [, , , , , , , ]
+****** Before build mixed iterable dataset...
+****** Build vla train dataloader successfully!
+************************* Build train_dataloader successful!
+************************* Before build_inf_evaluators
+09/30 [13:13:14] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:4807: UserWarning: No warnings.py:109
+ device id is provided via `init_process_group` or `barrier `. Using the current device set by the user.
+ warnings.warn( # warn only once
+
+************************* Build evaluators successful!
+************************* Early exit flags: early_exit=False
+************************* Initialize model successful!
+***** state_dict_path: /vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924/model.pt
+***** Load checkpoint successful!
+missing keys: ['action_head.model.layer_norm1.weight', 'action_head.model.layer_norm1.bias', 'action_head.model.fc1.weight', 'action_head.model.fc1.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.0.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.0.ffn.1.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.0.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.0.bias', 'action_head.model.mlp_resnet_blocks.1.ffn.1.weight', 'action_head.model.mlp_resnet_blocks.1.ffn.1.bias', 'action_head.model.layer_norm2.weight', 'action_head.model.layer_norm2.bias', 'action_head.model.fc2.weight', 'action_head.model.fc2.bias', 'proprio_projector.fc1.weight', 'proprio_projector.fc1.bias', 'proprio_projector.fc2.weight', 'proprio_projector.fc2.bias']
+unexpected keys: []
+************************* Initialize model successful!
+************************* LoRA flags: use_lora=True, lora_llm=False, lora_vit=False, lora_connector=False
+************************* Before add lora to model
+************************* Before FSDP model wrapping
+************************* FSDP model wrapping successful!
+************************* Before building optimizer and scheduler
+************* Before get lora params
+************* After get lora params successfully
+09/30 [13:14:27] INFO | >> Constructing optimizer with 2 param groups optim.py:1283
+**************************************************
+After building optimizer and scheduler and model, before training, peak GPU memory (MB): 35614
+************************* VLATrainer initialized successfully!
+************************* Before trainer.fit()
+Pre-train system metrics
+ System/Peak GPU Memory (MB)=35,614
+ WARNING | >> /vast/users/xiaodan/zhangjian/A1/olmo/data/collator.py:200: UserWarning: To copy construct from a tensor, it is recommended to use warnings.py:109
+ sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor).
+ timestep_list = [torch.tensor(ex["timestep"], dtype=torch.int64) for ex in batch]
+
+[step=1/500000]
+ train/ActionL1Loss=0.6793
+ throughput/total_tokens=192,000
+ System/Peak GPU Memory (MB)=40,145
+[step=2/500000]
+ train/ActionL1Loss=0.5930
+ throughput/total_tokens=384,000
+ throughput/device/tokens_per_second=1,200
+ throughput/device/batches_per_second=0.0500
+ System/Peak GPU Memory (MB)=46,917
+[step=3/500000]
+ train/ActionL1Loss=0.6559
+ throughput/total_tokens=576,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=4/500000]
+ train/ActionL1Loss=0.6283
+ throughput/total_tokens=768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5/500000]
+ train/ActionL1Loss=0.5817
+ throughput/total_tokens=960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=6/500000]
+ train/ActionL1Loss=0.5941
+ throughput/total_tokens=1,152,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=7/500000]
+ train/ActionL1Loss=0.5921
+ throughput/total_tokens=1,344,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=8/500000]
+ train/ActionL1Loss=0.5627
+ throughput/total_tokens=1,536,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=9/500000]
+ train/ActionL1Loss=0.5788
+ throughput/total_tokens=1,728,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=10/500000]
+ train/ActionL1Loss=0.5913
+ throughput/total_tokens=1,920,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=11/500000]
+ train/ActionL1Loss=0.5938
+ throughput/total_tokens=2,112,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=12/500000]
+ train/ActionL1Loss=0.5775
+ throughput/total_tokens=2,304,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=13/500000]
+ train/ActionL1Loss=0.5672
+ throughput/total_tokens=2,496,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=14/500000]
+ train/ActionL1Loss=0.5261
+ throughput/total_tokens=2,688,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=15/500000]
+ train/ActionL1Loss=0.5075
+ throughput/total_tokens=2,880,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=16/500000]
+ train/ActionL1Loss=0.5681
+ throughput/total_tokens=3,072,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=17/500000]
+ train/ActionL1Loss=0.5502
+ throughput/total_tokens=3,264,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=18/500000]
+ train/ActionL1Loss=0.5460
+ throughput/total_tokens=3,456,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=19/500000]
+ train/ActionL1Loss=0.5226
+ throughput/total_tokens=3,648,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=20/500000]
+ optim/total_grad_norm=19.49
+ train/ActionL1Loss=0.5248
+ throughput/total_tokens=3,840,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=21/500000]
+ train/ActionL1Loss=0.6231
+ throughput/total_tokens=4,032,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=22/500000]
+ train/ActionL1Loss=0.5242
+ throughput/total_tokens=4,224,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=23/500000]
+ train/ActionL1Loss=0.5158
+ throughput/total_tokens=4,416,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=24/500000]
+ train/ActionL1Loss=0.5881
+ throughput/total_tokens=4,608,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=25/500000]
+ train/ActionL1Loss=0.4934
+ throughput/total_tokens=4,800,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=26/500000]
+ train/ActionL1Loss=0.5082
+ throughput/total_tokens=4,992,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=27/500000]
+ train/ActionL1Loss=0.5737
+ throughput/total_tokens=5,184,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=28/500000]
+ train/ActionL1Loss=0.5663
+ throughput/total_tokens=5,376,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=29/500000]
+ train/ActionL1Loss=0.5887
+ throughput/total_tokens=5,568,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=30/500000]
+ train/ActionL1Loss=0.5154
+ throughput/total_tokens=5,760,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=31/500000]
+ train/ActionL1Loss=0.5592
+ throughput/total_tokens=5,952,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=32/500000]
+ train/ActionL1Loss=0.5018
+ throughput/total_tokens=6,144,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=33/500000]
+ train/ActionL1Loss=0.5105
+ throughput/total_tokens=6,336,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=34/500000]
+ train/ActionL1Loss=0.6220
+ throughput/total_tokens=6,528,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=35/500000]
+ train/ActionL1Loss=0.5397
+ throughput/total_tokens=6,720,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=36/500000]
+ train/ActionL1Loss=0.5327
+ throughput/total_tokens=6,912,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=37/500000]
+ train/ActionL1Loss=0.5355
+ throughput/total_tokens=7,104,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=38/500000]
+ train/ActionL1Loss=0.4923
+ throughput/total_tokens=7,296,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=39/500000]
+ train/ActionL1Loss=0.5807
+ throughput/total_tokens=7,488,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=40/500000]
+ optim/total_grad_norm=15.46
+ train/ActionL1Loss=0.4993
+ throughput/total_tokens=7,680,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=41/500000]
+ train/ActionL1Loss=0.5361
+ throughput/total_tokens=7,872,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=42/500000]
+ train/ActionL1Loss=0.5536
+ throughput/total_tokens=8,064,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=43/500000]
+ train/ActionL1Loss=0.5528
+ throughput/total_tokens=8,256,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=44/500000]
+ train/ActionL1Loss=0.6008
+ throughput/total_tokens=8,448,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=45/500000]
+ train/ActionL1Loss=0.5148
+ throughput/total_tokens=8,640,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=46/500000]
+ train/ActionL1Loss=0.5229
+ throughput/total_tokens=8,832,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=47/500000]
+ train/ActionL1Loss=0.5188
+ throughput/total_tokens=9,024,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=48/500000]
+ train/ActionL1Loss=0.5267
+ throughput/total_tokens=9,216,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=49/500000]
+ train/ActionL1Loss=0.5068
+ throughput/total_tokens=9,408,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=50/500000]
+ train/ActionL1Loss=0.5800
+ throughput/total_tokens=9,600,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=51/500000]
+ train/ActionL1Loss=0.5701
+ throughput/total_tokens=9,792,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=52/500000]
+ train/ActionL1Loss=0.5594
+ throughput/total_tokens=9,984,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=53/500000]
+ train/ActionL1Loss=0.5317
+ throughput/total_tokens=10,176,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=54/500000]
+ train/ActionL1Loss=0.5531
+ throughput/total_tokens=10,368,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=55/500000]
+ train/ActionL1Loss=0.4996
+ throughput/total_tokens=10,560,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=56/500000]
+ train/ActionL1Loss=0.5236
+ throughput/total_tokens=10,752,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=57/500000]
+ train/ActionL1Loss=0.5663
+ throughput/total_tokens=10,944,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=58/500000]
+ train/ActionL1Loss=0.5745
+ throughput/total_tokens=11,136,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=59/500000]
+ train/ActionL1Loss=0.5446
+ throughput/total_tokens=11,328,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=60/500000]
+ optim/total_grad_norm=23.71
+ train/ActionL1Loss=0.5159
+ throughput/total_tokens=11,520,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+ System/Peak GPU Memory (MB)=46,917
+[step=61/500000]
+ train/ActionL1Loss=0.5215
+ throughput/total_tokens=11,712,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=62/500000]
+ train/ActionL1Loss=0.5102
+ throughput/total_tokens=11,904,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=63/500000]
+ train/ActionL1Loss=0.4673
+ throughput/total_tokens=12,096,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=64/500000]
+ train/ActionL1Loss=0.5513
+ throughput/total_tokens=12,288,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=65/500000]
+ train/ActionL1Loss=0.5553
+ throughput/total_tokens=12,480,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=66/500000]
+ train/ActionL1Loss=0.5202
+ throughput/total_tokens=12,672,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=67/500000]
+ train/ActionL1Loss=0.5049
+ throughput/total_tokens=12,864,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=68/500000]
+ train/ActionL1Loss=0.5740
+ throughput/total_tokens=13,056,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=69/500000]
+ train/ActionL1Loss=0.5250
+ throughput/total_tokens=13,248,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=70/500000]
+ train/ActionL1Loss=0.5484
+ throughput/total_tokens=13,440,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=71/500000]
+ train/ActionL1Loss=0.5563
+ throughput/total_tokens=13,632,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=72/500000]
+ train/ActionL1Loss=0.4984
+ throughput/total_tokens=13,824,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=73/500000]
+ train/ActionL1Loss=0.5155
+ throughput/total_tokens=14,016,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=74/500000]
+ train/ActionL1Loss=0.5304
+ throughput/total_tokens=14,208,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=75/500000]
+ train/ActionL1Loss=0.5154
+ throughput/total_tokens=14,400,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=76/500000]
+ train/ActionL1Loss=0.4780
+ throughput/total_tokens=14,592,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=77/500000]
+ train/ActionL1Loss=0.4628
+ throughput/total_tokens=14,784,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=78/500000]
+ train/ActionL1Loss=0.5457
+ throughput/total_tokens=14,976,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=79/500000]
+ train/ActionL1Loss=0.5861
+ throughput/total_tokens=15,168,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=80/500000]
+ optim/total_grad_norm=22.39
+ train/ActionL1Loss=0.4898
+ throughput/total_tokens=15,360,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=81/500000]
+ train/ActionL1Loss=0.6331
+ throughput/total_tokens=15,552,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=82/500000]
+ train/ActionL1Loss=0.5654
+ throughput/total_tokens=15,744,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=83/500000]
+ train/ActionL1Loss=0.5584
+ throughput/total_tokens=15,936,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=84/500000]
+ train/ActionL1Loss=0.5468
+ throughput/total_tokens=16,128,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=85/500000]
+ train/ActionL1Loss=0.5248
+ throughput/total_tokens=16,320,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=86/500000]
+ train/ActionL1Loss=0.5029
+ throughput/total_tokens=16,512,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=87/500000]
+ train/ActionL1Loss=0.5210
+ throughput/total_tokens=16,704,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=88/500000]
+ train/ActionL1Loss=0.5187
+ throughput/total_tokens=16,896,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=89/500000]
+ train/ActionL1Loss=0.5067
+ throughput/total_tokens=17,088,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=90/500000]
+ train/ActionL1Loss=0.4625
+ throughput/total_tokens=17,280,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=91/500000]
+ train/ActionL1Loss=0.5168
+ throughput/total_tokens=17,472,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=92/500000]
+ train/ActionL1Loss=0.5409
+ throughput/total_tokens=17,664,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=93/500000]
+ train/ActionL1Loss=0.5601
+ throughput/total_tokens=17,856,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=94/500000]
+ train/ActionL1Loss=0.4243
+ throughput/total_tokens=18,048,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=95/500000]
+ train/ActionL1Loss=0.5194
+ throughput/total_tokens=18,240,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=96/500000]
+ train/ActionL1Loss=0.5665
+ throughput/total_tokens=18,432,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=97/500000]
+ train/ActionL1Loss=0.4418
+ throughput/total_tokens=18,624,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=98/500000]
+ train/ActionL1Loss=0.5506
+ throughput/total_tokens=18,816,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=99/500000]
+ train/ActionL1Loss=0.4717
+ throughput/total_tokens=19,008,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=100/500000]
+ optim/total_grad_norm=17.61
+ train/ActionL1Loss=0.5248
+ throughput/total_tokens=19,200,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=101/500000]
+ train/ActionL1Loss=0.5125
+ throughput/total_tokens=19,392,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=102/500000]
+ train/ActionL1Loss=0.5354
+ throughput/total_tokens=19,584,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=103/500000]
+ train/ActionL1Loss=0.5102
+ throughput/total_tokens=19,776,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=104/500000]
+ train/ActionL1Loss=0.5482
+ throughput/total_tokens=19,968,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=105/500000]
+ train/ActionL1Loss=0.5705
+ throughput/total_tokens=20,160,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=106/500000]
+ train/ActionL1Loss=0.4926
+ throughput/total_tokens=20,352,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=107/500000]
+ train/ActionL1Loss=0.5551
+ throughput/total_tokens=20,544,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=108/500000]
+ train/ActionL1Loss=0.5551
+ throughput/total_tokens=20,736,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=109/500000]
+ train/ActionL1Loss=0.4997
+ throughput/total_tokens=20,928,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=110/500000]
+ train/ActionL1Loss=0.5270
+ throughput/total_tokens=21,120,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=111/500000]
+ train/ActionL1Loss=0.5325
+ throughput/total_tokens=21,312,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=112/500000]
+ train/ActionL1Loss=0.4957
+ throughput/total_tokens=21,504,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=113/500000]
+ train/ActionL1Loss=0.5392
+ throughput/total_tokens=21,696,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=114/500000]
+ train/ActionL1Loss=0.5045
+ throughput/total_tokens=21,888,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=115/500000]
+ train/ActionL1Loss=0.5695
+ throughput/total_tokens=22,080,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=116/500000]
+ train/ActionL1Loss=0.5164
+ throughput/total_tokens=22,272,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=117/500000]
+ train/ActionL1Loss=0.5523
+ throughput/total_tokens=22,464,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=118/500000]
+ train/ActionL1Loss=0.5445
+ throughput/total_tokens=22,656,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=119/500000]
+ train/ActionL1Loss=0.4686
+ throughput/total_tokens=22,848,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=120/500000]
+ optim/total_grad_norm=14.19
+ train/ActionL1Loss=0.5430
+ throughput/total_tokens=23,040,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=121/500000]
+ train/ActionL1Loss=0.5473
+ throughput/total_tokens=23,232,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=122/500000]
+ train/ActionL1Loss=0.5987
+ throughput/total_tokens=23,424,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=123/500000]
+ train/ActionL1Loss=0.5515
+ throughput/total_tokens=23,616,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=124/500000]
+ train/ActionL1Loss=0.5459
+ throughput/total_tokens=23,808,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=125/500000]
+ train/ActionL1Loss=0.5786
+ throughput/total_tokens=24,000,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=126/500000]
+ train/ActionL1Loss=0.5208
+ throughput/total_tokens=24,192,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=127/500000]
+ train/ActionL1Loss=0.5503
+ throughput/total_tokens=24,384,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=128/500000]
+ train/ActionL1Loss=0.5833
+ throughput/total_tokens=24,576,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=129/500000]
+ train/ActionL1Loss=0.5338
+ throughput/total_tokens=24,768,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=130/500000]
+ train/ActionL1Loss=0.5428
+ throughput/total_tokens=24,960,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=131/500000]
+ train/ActionL1Loss=0.4783
+ throughput/total_tokens=25,152,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=132/500000]
+ train/ActionL1Loss=0.5188
+ throughput/total_tokens=25,344,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=133/500000]
+ train/ActionL1Loss=0.5806
+ throughput/total_tokens=25,536,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=134/500000]
+ train/ActionL1Loss=0.5368
+ throughput/total_tokens=25,728,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=135/500000]
+ train/ActionL1Loss=0.5088
+ throughput/total_tokens=25,920,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=136/500000]
+ train/ActionL1Loss=0.5313
+ throughput/total_tokens=26,112,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=137/500000]
+ train/ActionL1Loss=0.5029
+ throughput/total_tokens=26,304,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=138/500000]
+ train/ActionL1Loss=0.5067
+ throughput/total_tokens=26,496,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=139/500000]
+ train/ActionL1Loss=0.5152
+ throughput/total_tokens=26,688,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=140/500000]
+ optim/total_grad_norm=23.63
+ train/ActionL1Loss=0.5234
+ throughput/total_tokens=26,880,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=141/500000]
+ train/ActionL1Loss=0.4837
+ throughput/total_tokens=27,072,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=142/500000]
+ train/ActionL1Loss=0.4590
+ throughput/total_tokens=27,264,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=143/500000]
+ train/ActionL1Loss=0.5121
+ throughput/total_tokens=27,456,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=144/500000]
+ train/ActionL1Loss=0.4559
+ throughput/total_tokens=27,648,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=145/500000]
+ train/ActionL1Loss=0.4668
+ throughput/total_tokens=27,840,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=146/500000]
+ train/ActionL1Loss=0.4233
+ throughput/total_tokens=28,032,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=147/500000]
+ train/ActionL1Loss=0.4365
+ throughput/total_tokens=28,224,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=148/500000]
+ train/ActionL1Loss=0.4604
+ throughput/total_tokens=28,416,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=149/500000]
+ train/ActionL1Loss=0.3949
+ throughput/total_tokens=28,608,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=150/500000]
+ train/ActionL1Loss=0.3900
+ throughput/total_tokens=28,800,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=151/500000]
+ train/ActionL1Loss=0.3793
+ throughput/total_tokens=28,992,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=152/500000]
+ train/ActionL1Loss=0.3793
+ throughput/total_tokens=29,184,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=153/500000]
+ train/ActionL1Loss=0.3564
+ throughput/total_tokens=29,376,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=154/500000]
+ train/ActionL1Loss=0.3782
+ throughput/total_tokens=29,568,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=155/500000]
+ train/ActionL1Loss=0.3420
+ throughput/total_tokens=29,760,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=156/500000]
+ train/ActionL1Loss=0.3574
+ throughput/total_tokens=29,952,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=157/500000]
+ train/ActionL1Loss=0.3317
+ throughput/total_tokens=30,144,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=158/500000]
+ train/ActionL1Loss=0.3487
+ throughput/total_tokens=30,336,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=159/500000]
+ train/ActionL1Loss=0.3473
+ throughput/total_tokens=30,528,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=160/500000]
+ optim/total_grad_norm=25.91
+ train/ActionL1Loss=0.3520
+ throughput/total_tokens=30,720,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=161/500000]
+ train/ActionL1Loss=0.3086
+ throughput/total_tokens=30,912,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=162/500000]
+ train/ActionL1Loss=0.3156
+ throughput/total_tokens=31,104,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=163/500000]
+ train/ActionL1Loss=0.3163
+ throughput/total_tokens=31,296,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=164/500000]
+ train/ActionL1Loss=0.3323
+ throughput/total_tokens=31,488,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=165/500000]
+ train/ActionL1Loss=0.3094
+ throughput/total_tokens=31,680,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=166/500000]
+ train/ActionL1Loss=0.3056
+ throughput/total_tokens=31,872,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=167/500000]
+ train/ActionL1Loss=0.3072
+ throughput/total_tokens=32,064,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=168/500000]
+ train/ActionL1Loss=0.3139
+ throughput/total_tokens=32,256,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=169/500000]
+ train/ActionL1Loss=0.2983
+ throughput/total_tokens=32,448,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=170/500000]
+ train/ActionL1Loss=0.3091
+ throughput/total_tokens=32,640,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=171/500000]
+ train/ActionL1Loss=0.3064
+ throughput/total_tokens=32,832,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=172/500000]
+ train/ActionL1Loss=0.2767
+ throughput/total_tokens=33,024,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=173/500000]
+ train/ActionL1Loss=0.2648
+ throughput/total_tokens=33,216,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=174/500000]
+ train/ActionL1Loss=0.2654
+ throughput/total_tokens=33,408,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=175/500000]
+ train/ActionL1Loss=0.3006
+ throughput/total_tokens=33,600,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=176/500000]
+ train/ActionL1Loss=0.2798
+ throughput/total_tokens=33,792,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=177/500000]
+ train/ActionL1Loss=0.2637
+ throughput/total_tokens=33,984,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=178/500000]
+ train/ActionL1Loss=0.2645
+ throughput/total_tokens=34,176,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=179/500000]
+ train/ActionL1Loss=0.2754
+ throughput/total_tokens=34,368,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=180/500000]
+ optim/total_grad_norm=27.13
+ train/ActionL1Loss=0.3134
+ throughput/total_tokens=34,560,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=181/500000]
+ train/ActionL1Loss=0.2563
+ throughput/total_tokens=34,752,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=182/500000]
+ train/ActionL1Loss=0.2556
+ throughput/total_tokens=34,944,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=183/500000]
+ train/ActionL1Loss=0.2431
+ throughput/total_tokens=35,136,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=184/500000]
+ train/ActionL1Loss=0.2556
+ throughput/total_tokens=35,328,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=185/500000]
+ train/ActionL1Loss=0.2454
+ throughput/total_tokens=35,520,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=186/500000]
+ train/ActionL1Loss=0.2586
+ throughput/total_tokens=35,712,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=187/500000]
+ train/ActionL1Loss=0.2259
+ throughput/total_tokens=35,904,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=188/500000]
+ train/ActionL1Loss=0.2343
+ throughput/total_tokens=36,096,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=189/500000]
+ train/ActionL1Loss=0.2712
+ throughput/total_tokens=36,288,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=190/500000]
+ train/ActionL1Loss=0.2490
+ throughput/total_tokens=36,480,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=191/500000]
+ train/ActionL1Loss=0.2473
+ throughput/total_tokens=36,672,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=192/500000]
+ train/ActionL1Loss=0.2364
+ throughput/total_tokens=36,864,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=193/500000]
+ train/ActionL1Loss=0.2644
+ throughput/total_tokens=37,056,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=194/500000]
+ train/ActionL1Loss=0.2711
+ throughput/total_tokens=37,248,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=195/500000]
+ train/ActionL1Loss=0.2709
+ throughput/total_tokens=37,440,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=196/500000]
+ train/ActionL1Loss=0.2424
+ throughput/total_tokens=37,632,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=197/500000]
+ train/ActionL1Loss=0.2658
+ throughput/total_tokens=37,824,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=198/500000]
+ train/ActionL1Loss=0.2573
+ throughput/total_tokens=38,016,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=199/500000]
+ train/ActionL1Loss=0.2192
+ throughput/total_tokens=38,208,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=200/500000]
+ optim/total_grad_norm=23.02
+ train/ActionL1Loss=0.2533
+ throughput/total_tokens=38,400,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=201/500000]
+ train/ActionL1Loss=0.2330
+ throughput/total_tokens=38,592,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=202/500000]
+ train/ActionL1Loss=0.2304
+ throughput/total_tokens=38,784,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=203/500000]
+ train/ActionL1Loss=0.2472
+ throughput/total_tokens=38,976,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=204/500000]
+ train/ActionL1Loss=0.2292
+ throughput/total_tokens=39,168,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=205/500000]
+ train/ActionL1Loss=0.2286
+ throughput/total_tokens=39,360,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=206/500000]
+ train/ActionL1Loss=0.2284
+ throughput/total_tokens=39,552,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=207/500000]
+ train/ActionL1Loss=0.2619
+ throughput/total_tokens=39,744,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=208/500000]
+ train/ActionL1Loss=0.2163
+ throughput/total_tokens=39,936,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=209/500000]
+ train/ActionL1Loss=0.2141
+ throughput/total_tokens=40,128,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=210/500000]
+ train/ActionL1Loss=0.2182
+ throughput/total_tokens=40,320,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=211/500000]
+ train/ActionL1Loss=0.2002
+ throughput/total_tokens=40,512,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=212/500000]
+ train/ActionL1Loss=0.2113
+ throughput/total_tokens=40,704,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=213/500000]
+ train/ActionL1Loss=0.2084
+ throughput/total_tokens=40,896,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=214/500000]
+ train/ActionL1Loss=0.2268
+ throughput/total_tokens=41,088,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=215/500000]
+ train/ActionL1Loss=0.2274
+ throughput/total_tokens=41,280,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=216/500000]
+ train/ActionL1Loss=0.2201
+ throughput/total_tokens=41,472,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=217/500000]
+ train/ActionL1Loss=0.2271
+ throughput/total_tokens=41,664,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=218/500000]
+ train/ActionL1Loss=0.2161
+ throughput/total_tokens=41,856,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=219/500000]
+ train/ActionL1Loss=0.2182
+ throughput/total_tokens=42,048,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=220/500000]
+ optim/total_grad_norm=15.42
+ train/ActionL1Loss=0.2233
+ throughput/total_tokens=42,240,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=221/500000]
+ train/ActionL1Loss=0.2247
+ throughput/total_tokens=42,432,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=222/500000]
+ train/ActionL1Loss=0.2127
+ throughput/total_tokens=42,624,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=223/500000]
+ train/ActionL1Loss=0.2059
+ throughput/total_tokens=42,816,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=224/500000]
+ train/ActionL1Loss=0.2405
+ throughput/total_tokens=43,008,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=225/500000]
+ train/ActionL1Loss=0.2014
+ throughput/total_tokens=43,200,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=226/500000]
+ train/ActionL1Loss=0.2097
+ throughput/total_tokens=43,392,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=227/500000]
+ train/ActionL1Loss=0.2146
+ throughput/total_tokens=43,584,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=228/500000]
+ train/ActionL1Loss=0.2103
+ throughput/total_tokens=43,776,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=229/500000]
+ train/ActionL1Loss=0.2259
+ throughput/total_tokens=43,968,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=230/500000]
+ train/ActionL1Loss=0.2152
+ throughput/total_tokens=44,160,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=231/500000]
+ train/ActionL1Loss=0.2642
+ throughput/total_tokens=44,352,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=232/500000]
+ train/ActionL1Loss=0.2295
+ throughput/total_tokens=44,544,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=233/500000]
+ train/ActionL1Loss=0.2271
+ throughput/total_tokens=44,736,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=234/500000]
+ train/ActionL1Loss=0.2425
+ throughput/total_tokens=44,928,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=235/500000]
+ train/ActionL1Loss=0.2361
+ throughput/total_tokens=45,120,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=236/500000]
+ train/ActionL1Loss=0.2297
+ throughput/total_tokens=45,312,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=237/500000]
+ train/ActionL1Loss=0.2281
+ throughput/total_tokens=45,504,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=238/500000]
+ train/ActionL1Loss=0.2035
+ throughput/total_tokens=45,696,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=239/500000]
+ train/ActionL1Loss=0.2001
+ throughput/total_tokens=45,888,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=240/500000]
+ optim/total_grad_norm=16.29
+ train/ActionL1Loss=0.1758
+ throughput/total_tokens=46,080,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=241/500000]
+ train/ActionL1Loss=0.2035
+ throughput/total_tokens=46,272,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=242/500000]
+ train/ActionL1Loss=0.2020
+ throughput/total_tokens=46,464,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=243/500000]
+ train/ActionL1Loss=0.2050
+ throughput/total_tokens=46,656,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=244/500000]
+ train/ActionL1Loss=0.2206
+ throughput/total_tokens=46,848,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=245/500000]
+ train/ActionL1Loss=0.2011
+ throughput/total_tokens=47,040,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=246/500000]
+ train/ActionL1Loss=0.2121
+ throughput/total_tokens=47,232,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=247/500000]
+ train/ActionL1Loss=0.1984
+ throughput/total_tokens=47,424,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=248/500000]
+ train/ActionL1Loss=0.2219
+ throughput/total_tokens=47,616,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=249/500000]
+ train/ActionL1Loss=0.2219
+ throughput/total_tokens=47,808,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=250/500000]
+ train/ActionL1Loss=0.1928
+ throughput/total_tokens=48,000,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=251/500000]
+ train/ActionL1Loss=0.1916
+ throughput/total_tokens=48,192,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=252/500000]
+ train/ActionL1Loss=0.1974
+ throughput/total_tokens=48,384,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=253/500000]
+ train/ActionL1Loss=0.2183
+ throughput/total_tokens=48,576,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=254/500000]
+ train/ActionL1Loss=0.1849
+ throughput/total_tokens=48,768,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=255/500000]
+ train/ActionL1Loss=0.1801
+ throughput/total_tokens=48,960,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=256/500000]
+ train/ActionL1Loss=0.1961
+ throughput/total_tokens=49,152,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=257/500000]
+ train/ActionL1Loss=0.2249
+ throughput/total_tokens=49,344,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=258/500000]
+ train/ActionL1Loss=0.2318
+ throughput/total_tokens=49,536,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=259/500000]
+ train/ActionL1Loss=0.2273
+ throughput/total_tokens=49,728,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=260/500000]
+ optim/total_grad_norm=23.18
+ train/ActionL1Loss=0.1848
+ throughput/total_tokens=49,920,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=261/500000]
+ train/ActionL1Loss=0.2027
+ throughput/total_tokens=50,112,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=262/500000]
+ train/ActionL1Loss=0.2094
+ throughput/total_tokens=50,304,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=263/500000]
+ train/ActionL1Loss=0.1992
+ throughput/total_tokens=50,496,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=264/500000]
+ train/ActionL1Loss=0.1909
+ throughput/total_tokens=50,688,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=265/500000]
+ train/ActionL1Loss=0.2196
+ throughput/total_tokens=50,880,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=266/500000]
+ train/ActionL1Loss=0.2268
+ throughput/total_tokens=51,072,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=267/500000]
+ train/ActionL1Loss=0.2233
+ throughput/total_tokens=51,264,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=268/500000]
+ train/ActionL1Loss=0.2066
+ throughput/total_tokens=51,456,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=269/500000]
+ train/ActionL1Loss=0.2016
+ throughput/total_tokens=51,648,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=270/500000]
+ train/ActionL1Loss=0.2015
+ throughput/total_tokens=51,840,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=271/500000]
+ train/ActionL1Loss=0.2055
+ throughput/total_tokens=52,032,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=272/500000]
+ train/ActionL1Loss=0.2198
+ throughput/total_tokens=52,224,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=273/500000]
+ train/ActionL1Loss=0.1757
+ throughput/total_tokens=52,416,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=274/500000]
+ train/ActionL1Loss=0.1993
+ throughput/total_tokens=52,608,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=275/500000]
+ train/ActionL1Loss=0.1859
+ throughput/total_tokens=52,800,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=276/500000]
+ train/ActionL1Loss=0.2143
+ throughput/total_tokens=52,992,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=277/500000]
+ train/ActionL1Loss=0.1780
+ throughput/total_tokens=53,184,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=278/500000]
+ train/ActionL1Loss=0.1863
+ throughput/total_tokens=53,376,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=279/500000]
+ train/ActionL1Loss=0.2004
+ throughput/total_tokens=53,568,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=280/500000]
+ optim/total_grad_norm=17.31
+ train/ActionL1Loss=0.2156
+ throughput/total_tokens=53,760,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=281/500000]
+ train/ActionL1Loss=0.1881
+ throughput/total_tokens=53,952,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=282/500000]
+ train/ActionL1Loss=0.1926
+ throughput/total_tokens=54,144,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=283/500000]
+ train/ActionL1Loss=0.1987
+ throughput/total_tokens=54,336,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=284/500000]
+ train/ActionL1Loss=0.1991
+ throughput/total_tokens=54,528,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=285/500000]
+ train/ActionL1Loss=0.2128
+ throughput/total_tokens=54,720,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=286/500000]
+ train/ActionL1Loss=0.2022
+ throughput/total_tokens=54,912,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=287/500000]
+ train/ActionL1Loss=0.2106
+ throughput/total_tokens=55,104,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=288/500000]
+ train/ActionL1Loss=0.1919
+ throughput/total_tokens=55,296,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=289/500000]
+ train/ActionL1Loss=0.2025
+ throughput/total_tokens=55,488,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=290/500000]
+ train/ActionL1Loss=0.2146
+ throughput/total_tokens=55,680,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=291/500000]
+ train/ActionL1Loss=0.1919
+ throughput/total_tokens=55,872,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=292/500000]
+ train/ActionL1Loss=0.1888
+ throughput/total_tokens=56,064,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=293/500000]
+ train/ActionL1Loss=0.2064
+ throughput/total_tokens=56,256,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=294/500000]
+ train/ActionL1Loss=0.2027
+ throughput/total_tokens=56,448,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=295/500000]
+ train/ActionL1Loss=0.1783
+ throughput/total_tokens=56,640,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=296/500000]
+ train/ActionL1Loss=0.1850
+ throughput/total_tokens=56,832,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=297/500000]
+ train/ActionL1Loss=0.2083
+ throughput/total_tokens=57,024,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=298/500000]
+ train/ActionL1Loss=0.2021
+ throughput/total_tokens=57,216,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=299/500000]
+ train/ActionL1Loss=0.1845
+ throughput/total_tokens=57,408,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=300/500000]
+ optim/total_grad_norm=16.60
+ train/ActionL1Loss=0.1845
+ throughput/total_tokens=57,600,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=301/500000]
+ train/ActionL1Loss=0.2018
+ throughput/total_tokens=57,792,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=302/500000]
+ train/ActionL1Loss=0.1957
+ throughput/total_tokens=57,984,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=303/500000]
+ train/ActionL1Loss=0.1652
+ throughput/total_tokens=58,176,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=304/500000]
+ train/ActionL1Loss=0.2014
+ throughput/total_tokens=58,368,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=305/500000]
+ train/ActionL1Loss=0.1828
+ throughput/total_tokens=58,560,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=306/500000]
+ train/ActionL1Loss=0.1950
+ throughput/total_tokens=58,752,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=307/500000]
+ train/ActionL1Loss=0.2061
+ throughput/total_tokens=58,944,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=308/500000]
+ train/ActionL1Loss=0.1981
+ throughput/total_tokens=59,136,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=309/500000]
+ train/ActionL1Loss=0.1972
+ throughput/total_tokens=59,328,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=310/500000]
+ train/ActionL1Loss=0.1936
+ throughput/total_tokens=59,520,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=311/500000]
+ train/ActionL1Loss=0.1842
+ throughput/total_tokens=59,712,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=312/500000]
+ train/ActionL1Loss=0.1736
+ throughput/total_tokens=59,904,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=313/500000]
+ train/ActionL1Loss=0.2145
+ throughput/total_tokens=60,096,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=314/500000]
+ train/ActionL1Loss=0.1929
+ throughput/total_tokens=60,288,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=315/500000]
+ train/ActionL1Loss=0.1741
+ throughput/total_tokens=60,480,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=316/500000]
+ train/ActionL1Loss=0.1897
+ throughput/total_tokens=60,672,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=317/500000]
+ train/ActionL1Loss=0.1783
+ throughput/total_tokens=60,864,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=318/500000]
+ train/ActionL1Loss=0.2116
+ throughput/total_tokens=61,056,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=319/500000]
+ train/ActionL1Loss=0.1984
+ throughput/total_tokens=61,248,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=320/500000]
+ optim/total_grad_norm=18.84
+ train/ActionL1Loss=0.1753
+ throughput/total_tokens=61,440,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=321/500000]
+ train/ActionL1Loss=0.1826
+ throughput/total_tokens=61,632,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=322/500000]
+ train/ActionL1Loss=0.1789
+ throughput/total_tokens=61,824,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=323/500000]
+ train/ActionL1Loss=0.2173
+ throughput/total_tokens=62,016,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=324/500000]
+ train/ActionL1Loss=0.2000
+ throughput/total_tokens=62,208,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=325/500000]
+ train/ActionL1Loss=0.1843
+ throughput/total_tokens=62,400,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=326/500000]
+ train/ActionL1Loss=0.1770
+ throughput/total_tokens=62,592,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=327/500000]
+ train/ActionL1Loss=0.1927
+ throughput/total_tokens=62,784,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=328/500000]
+ train/ActionL1Loss=0.2250
+ throughput/total_tokens=62,976,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=329/500000]
+ train/ActionL1Loss=0.2190
+ throughput/total_tokens=63,168,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=330/500000]
+ train/ActionL1Loss=0.2269
+ throughput/total_tokens=63,360,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=331/500000]
+ train/ActionL1Loss=0.1855
+ throughput/total_tokens=63,552,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=332/500000]
+ train/ActionL1Loss=0.1776
+ throughput/total_tokens=63,744,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=333/500000]
+ train/ActionL1Loss=0.1907
+ throughput/total_tokens=63,936,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=334/500000]
+ train/ActionL1Loss=0.2260
+ throughput/total_tokens=64,128,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=335/500000]
+ train/ActionL1Loss=0.2014
+ throughput/total_tokens=64,320,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=336/500000]
+ train/ActionL1Loss=0.2003
+ throughput/total_tokens=64,512,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=337/500000]
+ train/ActionL1Loss=0.1959
+ throughput/total_tokens=64,704,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=338/500000]
+ train/ActionL1Loss=0.1735
+ throughput/total_tokens=64,896,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=339/500000]
+ train/ActionL1Loss=0.1894
+ throughput/total_tokens=65,088,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=340/500000]
+ optim/total_grad_norm=21.32
+ train/ActionL1Loss=0.1724
+ throughput/total_tokens=65,280,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=341/500000]
+ train/ActionL1Loss=0.1899
+ throughput/total_tokens=65,472,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=342/500000]
+ train/ActionL1Loss=0.1680
+ throughput/total_tokens=65,664,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=343/500000]
+ train/ActionL1Loss=0.1867
+ throughput/total_tokens=65,856,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=344/500000]
+ train/ActionL1Loss=0.1897
+ throughput/total_tokens=66,048,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=345/500000]
+ train/ActionL1Loss=0.1672
+ throughput/total_tokens=66,240,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=346/500000]
+ train/ActionL1Loss=0.1895
+ throughput/total_tokens=66,432,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=347/500000]
+ train/ActionL1Loss=0.1931
+ throughput/total_tokens=66,624,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=348/500000]
+ train/ActionL1Loss=0.1768
+ throughput/total_tokens=66,816,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=349/500000]
+ train/ActionL1Loss=0.1995
+ throughput/total_tokens=67,008,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=350/500000]
+ train/ActionL1Loss=0.1850
+ throughput/total_tokens=67,200,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=351/500000]
+ train/ActionL1Loss=0.1660
+ throughput/total_tokens=67,392,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=352/500000]
+ train/ActionL1Loss=0.1564
+ throughput/total_tokens=67,584,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=353/500000]
+ train/ActionL1Loss=0.2194
+ throughput/total_tokens=67,776,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=354/500000]
+ train/ActionL1Loss=0.1869
+ throughput/total_tokens=67,968,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=355/500000]
+ train/ActionL1Loss=0.1860
+ throughput/total_tokens=68,160,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=356/500000]
+ train/ActionL1Loss=0.1781
+ throughput/total_tokens=68,352,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=357/500000]
+ train/ActionL1Loss=0.1937
+ throughput/total_tokens=68,544,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=358/500000]
+ train/ActionL1Loss=0.2198
+ throughput/total_tokens=68,736,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=359/500000]
+ train/ActionL1Loss=0.2007
+ throughput/total_tokens=68,928,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=360/500000]
+ optim/total_grad_norm=14.89
+ train/ActionL1Loss=0.1978
+ throughput/total_tokens=69,120,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=361/500000]
+ train/ActionL1Loss=0.1887
+ throughput/total_tokens=69,312,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=362/500000]
+ train/ActionL1Loss=0.2224
+ throughput/total_tokens=69,504,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=363/500000]
+ train/ActionL1Loss=0.2033
+ throughput/total_tokens=69,696,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=364/500000]
+ train/ActionL1Loss=0.1799
+ throughput/total_tokens=69,888,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=365/500000]
+ train/ActionL1Loss=0.1892
+ throughput/total_tokens=70,080,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=366/500000]
+ train/ActionL1Loss=0.2160
+ throughput/total_tokens=70,272,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=367/500000]
+ train/ActionL1Loss=0.1924
+ throughput/total_tokens=70,464,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=368/500000]
+ train/ActionL1Loss=0.1888
+ throughput/total_tokens=70,656,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=369/500000]
+ train/ActionL1Loss=0.1720
+ throughput/total_tokens=70,848,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=370/500000]
+ train/ActionL1Loss=0.1896
+ throughput/total_tokens=71,040,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=371/500000]
+ train/ActionL1Loss=0.1847
+ throughput/total_tokens=71,232,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=372/500000]
+ train/ActionL1Loss=0.2005
+ throughput/total_tokens=71,424,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=373/500000]
+ train/ActionL1Loss=0.1846
+ throughput/total_tokens=71,616,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=374/500000]
+ train/ActionL1Loss=0.1679
+ throughput/total_tokens=71,808,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=375/500000]
+ train/ActionL1Loss=0.2013
+ throughput/total_tokens=72,000,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=376/500000]
+ train/ActionL1Loss=0.2209
+ throughput/total_tokens=72,192,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=377/500000]
+ train/ActionL1Loss=0.1855
+ throughput/total_tokens=72,384,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=378/500000]
+ train/ActionL1Loss=0.1956
+ throughput/total_tokens=72,576,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=379/500000]
+ train/ActionL1Loss=0.1849
+ throughput/total_tokens=72,768,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=380/500000]
+ optim/total_grad_norm=15.45
+ train/ActionL1Loss=0.1742
+ throughput/total_tokens=72,960,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=381/500000]
+ train/ActionL1Loss=0.1951
+ throughput/total_tokens=73,152,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=382/500000]
+ train/ActionL1Loss=0.1919
+ throughput/total_tokens=73,344,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=383/500000]
+ train/ActionL1Loss=0.1954
+ throughput/total_tokens=73,536,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=384/500000]
+ train/ActionL1Loss=0.1812
+ throughput/total_tokens=73,728,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=385/500000]
+ train/ActionL1Loss=0.2024
+ throughput/total_tokens=73,920,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=386/500000]
+ train/ActionL1Loss=0.1881
+ throughput/total_tokens=74,112,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=387/500000]
+ train/ActionL1Loss=0.1644
+ throughput/total_tokens=74,304,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=388/500000]
+ train/ActionL1Loss=0.1779
+ throughput/total_tokens=74,496,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=389/500000]
+ train/ActionL1Loss=0.1972
+ throughput/total_tokens=74,688,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=390/500000]
+ train/ActionL1Loss=0.1736
+ throughput/total_tokens=74,880,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=391/500000]
+ train/ActionL1Loss=0.1827
+ throughput/total_tokens=75,072,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=392/500000]
+ train/ActionL1Loss=0.1778
+ throughput/total_tokens=75,264,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=393/500000]
+ train/ActionL1Loss=0.1618
+ throughput/total_tokens=75,456,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=394/500000]
+ train/ActionL1Loss=0.1876
+ throughput/total_tokens=75,648,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=395/500000]
+ train/ActionL1Loss=0.1848
+ throughput/total_tokens=75,840,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=396/500000]
+ train/ActionL1Loss=0.2213
+ throughput/total_tokens=76,032,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=397/500000]
+ train/ActionL1Loss=0.2074
+ throughput/total_tokens=76,224,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=398/500000]
+ train/ActionL1Loss=0.2003
+ throughput/total_tokens=76,416,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=399/500000]
+ train/ActionL1Loss=0.1818
+ throughput/total_tokens=76,608,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=400/500000]
+ optim/total_grad_norm=17.86
+ train/ActionL1Loss=0.2047
+ throughput/total_tokens=76,800,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=401/500000]
+ train/ActionL1Loss=0.1926
+ throughput/total_tokens=76,992,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=402/500000]
+ train/ActionL1Loss=0.1891
+ throughput/total_tokens=77,184,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=403/500000]
+ train/ActionL1Loss=0.1927
+ throughput/total_tokens=77,376,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=404/500000]
+ train/ActionL1Loss=0.1880
+ throughput/total_tokens=77,568,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=405/500000]
+ train/ActionL1Loss=0.1822
+ throughput/total_tokens=77,760,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=406/500000]
+ train/ActionL1Loss=0.1853
+ throughput/total_tokens=77,952,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=407/500000]
+ train/ActionL1Loss=0.1951
+ throughput/total_tokens=78,144,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=408/500000]
+ train/ActionL1Loss=0.1735
+ throughput/total_tokens=78,336,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=409/500000]
+ train/ActionL1Loss=0.1942
+ throughput/total_tokens=78,528,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=410/500000]
+ train/ActionL1Loss=0.1782
+ throughput/total_tokens=78,720,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=411/500000]
+ train/ActionL1Loss=0.1767
+ throughput/total_tokens=78,912,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=412/500000]
+ train/ActionL1Loss=0.1702
+ throughput/total_tokens=79,104,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=413/500000]
+ train/ActionL1Loss=0.1537
+ throughput/total_tokens=79,296,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=414/500000]
+ train/ActionL1Loss=0.1630
+ throughput/total_tokens=79,488,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=415/500000]
+ train/ActionL1Loss=0.1654
+ throughput/total_tokens=79,680,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=416/500000]
+ train/ActionL1Loss=0.2000
+ throughput/total_tokens=79,872,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=417/500000]
+ train/ActionL1Loss=0.2001
+ throughput/total_tokens=80,064,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=418/500000]
+ train/ActionL1Loss=0.1751
+ throughput/total_tokens=80,256,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=419/500000]
+ train/ActionL1Loss=0.1956
+ throughput/total_tokens=80,448,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=420/500000]
+ optim/total_grad_norm=14.71
+ train/ActionL1Loss=0.1834
+ throughput/total_tokens=80,640,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=421/500000]
+ train/ActionL1Loss=0.1955
+ throughput/total_tokens=80,832,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=422/500000]
+ train/ActionL1Loss=0.1663
+ throughput/total_tokens=81,024,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=423/500000]
+ train/ActionL1Loss=0.1624
+ throughput/total_tokens=81,216,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=424/500000]
+ train/ActionL1Loss=0.1808
+ throughput/total_tokens=81,408,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=425/500000]
+ train/ActionL1Loss=0.1633
+ throughput/total_tokens=81,600,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=426/500000]
+ train/ActionL1Loss=0.1948
+ throughput/total_tokens=81,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=427/500000]
+ train/ActionL1Loss=0.1933
+ throughput/total_tokens=81,984,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=428/500000]
+ train/ActionL1Loss=0.2128
+ throughput/total_tokens=82,176,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=429/500000]
+ train/ActionL1Loss=0.1875
+ throughput/total_tokens=82,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=430/500000]
+ train/ActionL1Loss=0.1616
+ throughput/total_tokens=82,560,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=431/500000]
+ train/ActionL1Loss=0.1755
+ throughput/total_tokens=82,752,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=432/500000]
+ train/ActionL1Loss=0.1929
+ throughput/total_tokens=82,944,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=433/500000]
+ train/ActionL1Loss=0.1802
+ throughput/total_tokens=83,136,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=434/500000]
+ train/ActionL1Loss=0.1632
+ throughput/total_tokens=83,328,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=435/500000]
+ train/ActionL1Loss=0.1780
+ throughput/total_tokens=83,520,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=436/500000]
+ train/ActionL1Loss=0.1602
+ throughput/total_tokens=83,712,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=437/500000]
+ train/ActionL1Loss=0.2114
+ throughput/total_tokens=83,904,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=438/500000]
+ train/ActionL1Loss=0.1750
+ throughput/total_tokens=84,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=439/500000]
+ train/ActionL1Loss=0.1717
+ throughput/total_tokens=84,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=440/500000]
+ optim/total_grad_norm=15.10
+ train/ActionL1Loss=0.1774
+ throughput/total_tokens=84,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=441/500000]
+ train/ActionL1Loss=0.1699
+ throughput/total_tokens=84,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=442/500000]
+ train/ActionL1Loss=0.1903
+ throughput/total_tokens=84,864,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=443/500000]
+ train/ActionL1Loss=0.1792
+ throughput/total_tokens=85,056,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=444/500000]
+ train/ActionL1Loss=0.1757
+ throughput/total_tokens=85,248,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=445/500000]
+ train/ActionL1Loss=0.1856
+ throughput/total_tokens=85,440,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=446/500000]
+ train/ActionL1Loss=0.1619
+ throughput/total_tokens=85,632,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=447/500000]
+ train/ActionL1Loss=0.1671
+ throughput/total_tokens=85,824,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=448/500000]
+ train/ActionL1Loss=0.1778
+ throughput/total_tokens=86,016,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=449/500000]
+ train/ActionL1Loss=0.1607
+ throughput/total_tokens=86,208,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=450/500000]
+ train/ActionL1Loss=0.1515
+ throughput/total_tokens=86,400,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=451/500000]
+ train/ActionL1Loss=0.1881
+ throughput/total_tokens=86,592,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=452/500000]
+ train/ActionL1Loss=0.1799
+ throughput/total_tokens=86,784,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=453/500000]
+ train/ActionL1Loss=0.1364
+ throughput/total_tokens=86,976,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=454/500000]
+ train/ActionL1Loss=0.1777
+ throughput/total_tokens=87,168,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=455/500000]
+ train/ActionL1Loss=0.1706
+ throughput/total_tokens=87,360,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=456/500000]
+ train/ActionL1Loss=0.1875
+ throughput/total_tokens=87,552,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=457/500000]
+ train/ActionL1Loss=0.1717
+ throughput/total_tokens=87,744,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=458/500000]
+ train/ActionL1Loss=0.1822
+ throughput/total_tokens=87,936,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=459/500000]
+ train/ActionL1Loss=0.1582
+ throughput/total_tokens=88,128,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=460/500000]
+ optim/total_grad_norm=13.39
+ train/ActionL1Loss=0.1967
+ throughput/total_tokens=88,320,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=461/500000]
+ train/ActionL1Loss=0.1721
+ throughput/total_tokens=88,512,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=462/500000]
+ train/ActionL1Loss=0.1768
+ throughput/total_tokens=88,704,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=463/500000]
+ train/ActionL1Loss=0.1719
+ throughput/total_tokens=88,896,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=464/500000]
+ train/ActionL1Loss=0.1579
+ throughput/total_tokens=89,088,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=465/500000]
+ train/ActionL1Loss=0.1571
+ throughput/total_tokens=89,280,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=466/500000]
+ train/ActionL1Loss=0.1580
+ throughput/total_tokens=89,472,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=467/500000]
+ train/ActionL1Loss=0.1895
+ throughput/total_tokens=89,664,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=468/500000]
+ train/ActionL1Loss=0.1769
+ throughput/total_tokens=89,856,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=469/500000]
+ train/ActionL1Loss=0.1762
+ throughput/total_tokens=90,048,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=470/500000]
+ train/ActionL1Loss=0.1711
+ throughput/total_tokens=90,240,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=471/500000]
+ train/ActionL1Loss=0.2014
+ throughput/total_tokens=90,432,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=472/500000]
+ train/ActionL1Loss=0.1827
+ throughput/total_tokens=90,624,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=473/500000]
+ train/ActionL1Loss=0.1709
+ throughput/total_tokens=90,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=474/500000]
+ train/ActionL1Loss=0.1947
+ throughput/total_tokens=91,008,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=475/500000]
+ train/ActionL1Loss=0.1680
+ throughput/total_tokens=91,200,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=476/500000]
+ train/ActionL1Loss=0.1825
+ throughput/total_tokens=91,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=477/500000]
+ train/ActionL1Loss=0.1651
+ throughput/total_tokens=91,584,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=478/500000]
+ train/ActionL1Loss=0.1814
+ throughput/total_tokens=91,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=479/500000]
+ train/ActionL1Loss=0.1909
+ throughput/total_tokens=91,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=480/500000]
+ optim/total_grad_norm=13.61
+ train/ActionL1Loss=0.1864
+ throughput/total_tokens=92,160,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=481/500000]
+ train/ActionL1Loss=0.1661
+ throughput/total_tokens=92,352,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=482/500000]
+ train/ActionL1Loss=0.2058
+ throughput/total_tokens=92,544,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=483/500000]
+ train/ActionL1Loss=0.1784
+ throughput/total_tokens=92,736,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=484/500000]
+ train/ActionL1Loss=0.1447
+ throughput/total_tokens=92,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=485/500000]
+ train/ActionL1Loss=0.1619
+ throughput/total_tokens=93,120,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=486/500000]
+ train/ActionL1Loss=0.2042
+ throughput/total_tokens=93,312,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=487/500000]
+ train/ActionL1Loss=0.1800
+ throughput/total_tokens=93,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=488/500000]
+ train/ActionL1Loss=0.1888
+ throughput/total_tokens=93,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=489/500000]
+ train/ActionL1Loss=0.1719
+ throughput/total_tokens=93,888,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=490/500000]
+ train/ActionL1Loss=0.1775
+ throughput/total_tokens=94,080,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=491/500000]
+ train/ActionL1Loss=0.1873
+ throughput/total_tokens=94,272,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=492/500000]
+ train/ActionL1Loss=0.1884
+ throughput/total_tokens=94,464,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=493/500000]
+ train/ActionL1Loss=0.1794
+ throughput/total_tokens=94,656,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=494/500000]
+ train/ActionL1Loss=0.1571
+ throughput/total_tokens=94,848,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=495/500000]
+ train/ActionL1Loss=0.2027
+ throughput/total_tokens=95,040,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=496/500000]
+ train/ActionL1Loss=0.1813
+ throughput/total_tokens=95,232,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=497/500000]
+ train/ActionL1Loss=0.1860
+ throughput/total_tokens=95,424,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=498/500000]
+ train/ActionL1Loss=0.1742
+ throughput/total_tokens=95,616,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=499/500000]
+ train/ActionL1Loss=0.1614
+ throughput/total_tokens=95,808,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=500/500000]
+ optim/total_grad_norm=13.84
+ train/ActionL1Loss=0.1861
+ throughput/total_tokens=96,000,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+09/30 [16:07:44] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:678: warnings.py:109
+ FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and
+ set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc:
+ https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial:
+ https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
+ warnings.warn(
+
+ WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py:860: UserWarning: warnings.py:109
+ `_get_pg_default_device` will be deprecated, it only stays for backward-compatiblity reason. If you need to find a device for object
+ collectives, please use `_get_object_coll_device`. If you need to query the device types supported by group, please use
+ `_device_capability(group)`.
+ warnings.warn(
+
+09/30 [16:07:47] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/fsdp/_state_dict_utils.py:722: FutureWarning: warnings.py:109
+ Please use DTensor instead and we are deprecating ShardedTensor.
+ local_shape = tensor.shape
+
+ WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/fsdp/_state_dict_utils.py:739: FutureWarning: warnings.py:109
+ Please use DTensor instead and we are deprecating ShardedTensor.
+ tensor.shape,
+
+ WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/distributed/fsdp/_state_dict_utils.py:741: FutureWarning: warnings.py:109
+ Please use DTensor instead and we are deprecating ShardedTensor.
+ tensor.dtype,
+
+09/30 [16:07:51] WARNING | >> /vast/users/xiaodan/miniconda3/envs/a1/lib/python3.10/site-packages/torch/overrides.py:1747: FutureWarning: Please use DTensor warnings.py:109
+ instead and we are deprecating ShardedTensor.
+ result = torch_func_method(public_api, types, args, kwargs)
+
+09/30 [16:08:23] INFO | >> Saving config... checkpoint.py:608
+09/30 [16:08:54] INFO | >> Saving model state... checkpoint.py:796
+09/30 [16:10:00] INFO | >> Saving optim state... checkpoint.py:811
+09/30 [16:11:31] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=501/500000]
+ train/ActionL1Loss=0.1892
+ throughput/total_tokens=96,192,000
+ throughput/device/tokens_per_second=1,207
+ throughput/device/batches_per_second=0.0503
+[step=502/500000]
+ train/ActionL1Loss=0.1546
+ throughput/total_tokens=96,384,000
+ throughput/device/tokens_per_second=1,181
+ throughput/device/batches_per_second=0.0492
+[step=503/500000]
+ train/ActionL1Loss=0.1637
+ throughput/total_tokens=96,576,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=504/500000]
+ train/ActionL1Loss=0.1788
+ throughput/total_tokens=96,768,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=505/500000]
+ train/ActionL1Loss=0.1868
+ throughput/total_tokens=96,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=506/500000]
+ train/ActionL1Loss=0.1487
+ throughput/total_tokens=97,152,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=507/500000]
+ train/ActionL1Loss=0.1739
+ throughput/total_tokens=97,344,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=508/500000]
+ train/ActionL1Loss=0.1846
+ throughput/total_tokens=97,536,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=509/500000]
+ train/ActionL1Loss=0.1693
+ throughput/total_tokens=97,728,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=510/500000]
+ train/ActionL1Loss=0.1609
+ throughput/total_tokens=97,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=511/500000]
+ train/ActionL1Loss=0.1648
+ throughput/total_tokens=98,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=512/500000]
+ train/ActionL1Loss=0.1716
+ throughput/total_tokens=98,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=513/500000]
+ train/ActionL1Loss=0.1814
+ throughput/total_tokens=98,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=514/500000]
+ train/ActionL1Loss=0.1770
+ throughput/total_tokens=98,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=515/500000]
+ train/ActionL1Loss=0.1785
+ throughput/total_tokens=98,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=516/500000]
+ train/ActionL1Loss=0.1820
+ throughput/total_tokens=99,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=517/500000]
+ train/ActionL1Loss=0.1843
+ throughput/total_tokens=99,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=518/500000]
+ train/ActionL1Loss=0.1524
+ throughput/total_tokens=99,456,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=519/500000]
+ train/ActionL1Loss=0.1808
+ throughput/total_tokens=99,648,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=520/500000]
+ optim/total_grad_norm=14.66
+ train/ActionL1Loss=0.1712
+ throughput/total_tokens=99,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=521/500000]
+ train/ActionL1Loss=0.1766
+ throughput/total_tokens=100,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=522/500000]
+ train/ActionL1Loss=0.1539
+ throughput/total_tokens=100,224,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=523/500000]
+ train/ActionL1Loss=0.1631
+ throughput/total_tokens=100,416,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=524/500000]
+ train/ActionL1Loss=0.1721
+ throughput/total_tokens=100,608,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=525/500000]
+ train/ActionL1Loss=0.1662
+ throughput/total_tokens=100,800,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=526/500000]
+ train/ActionL1Loss=0.1434
+ throughput/total_tokens=100,992,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=527/500000]
+ train/ActionL1Loss=0.1556
+ throughput/total_tokens=101,184,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=528/500000]
+ train/ActionL1Loss=0.1482
+ throughput/total_tokens=101,376,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=529/500000]
+ train/ActionL1Loss=0.1653
+ throughput/total_tokens=101,568,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=530/500000]
+ train/ActionL1Loss=0.2038
+ throughput/total_tokens=101,760,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=531/500000]
+ train/ActionL1Loss=0.1562
+ throughput/total_tokens=101,952,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=532/500000]
+ train/ActionL1Loss=0.1583
+ throughput/total_tokens=102,144,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=533/500000]
+ train/ActionL1Loss=0.1652
+ throughput/total_tokens=102,336,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=534/500000]
+ train/ActionL1Loss=0.1821
+ throughput/total_tokens=102,528,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=535/500000]
+ train/ActionL1Loss=0.1932
+ throughput/total_tokens=102,720,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=536/500000]
+ train/ActionL1Loss=0.1625
+ throughput/total_tokens=102,912,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=537/500000]
+ train/ActionL1Loss=0.1979
+ throughput/total_tokens=103,104,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=538/500000]
+ train/ActionL1Loss=0.1715
+ throughput/total_tokens=103,296,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=539/500000]
+ train/ActionL1Loss=0.1873
+ throughput/total_tokens=103,488,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=540/500000]
+ optim/total_grad_norm=10.94
+ train/ActionL1Loss=0.1753
+ throughput/total_tokens=103,680,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=541/500000]
+ train/ActionL1Loss=0.1665
+ throughput/total_tokens=103,872,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=542/500000]
+ train/ActionL1Loss=0.1859
+ throughput/total_tokens=104,064,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=543/500000]
+ train/ActionL1Loss=0.1495
+ throughput/total_tokens=104,256,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=544/500000]
+ train/ActionL1Loss=0.1803
+ throughput/total_tokens=104,448,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=545/500000]
+ train/ActionL1Loss=0.1802
+ throughput/total_tokens=104,640,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=546/500000]
+ train/ActionL1Loss=0.1907
+ throughput/total_tokens=104,832,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=547/500000]
+ train/ActionL1Loss=0.1633
+ throughput/total_tokens=105,024,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=548/500000]
+ train/ActionL1Loss=0.1828
+ throughput/total_tokens=105,216,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=549/500000]
+ train/ActionL1Loss=0.1400
+ throughput/total_tokens=105,408,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=550/500000]
+ train/ActionL1Loss=0.1469
+ throughput/total_tokens=105,600,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=551/500000]
+ train/ActionL1Loss=0.1730
+ throughput/total_tokens=105,792,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=552/500000]
+ train/ActionL1Loss=0.1472
+ throughput/total_tokens=105,984,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=553/500000]
+ train/ActionL1Loss=0.1625
+ throughput/total_tokens=106,176,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=554/500000]
+ train/ActionL1Loss=0.1470
+ throughput/total_tokens=106,368,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=555/500000]
+ train/ActionL1Loss=0.1795
+ throughput/total_tokens=106,560,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=556/500000]
+ train/ActionL1Loss=0.1644
+ throughput/total_tokens=106,752,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=557/500000]
+ train/ActionL1Loss=0.1542
+ throughput/total_tokens=106,944,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=558/500000]
+ train/ActionL1Loss=0.1752
+ throughput/total_tokens=107,136,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=559/500000]
+ train/ActionL1Loss=0.1490
+ throughput/total_tokens=107,328,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=560/500000]
+ optim/total_grad_norm=12.44
+ train/ActionL1Loss=0.1595
+ throughput/total_tokens=107,520,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=561/500000]
+ train/ActionL1Loss=0.1708
+ throughput/total_tokens=107,712,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=562/500000]
+ train/ActionL1Loss=0.1758
+ throughput/total_tokens=107,904,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=563/500000]
+ train/ActionL1Loss=0.1604
+ throughput/total_tokens=108,096,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=564/500000]
+ train/ActionL1Loss=0.1665
+ throughput/total_tokens=108,288,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=565/500000]
+ train/ActionL1Loss=0.1680
+ throughput/total_tokens=108,480,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=566/500000]
+ train/ActionL1Loss=0.1650
+ throughput/total_tokens=108,672,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=567/500000]
+ train/ActionL1Loss=0.1549
+ throughput/total_tokens=108,864,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=568/500000]
+ train/ActionL1Loss=0.1581
+ throughput/total_tokens=109,056,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=569/500000]
+ train/ActionL1Loss=0.1455
+ throughput/total_tokens=109,248,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=570/500000]
+ train/ActionL1Loss=0.1426
+ throughput/total_tokens=109,440,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=571/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=109,632,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=572/500000]
+ train/ActionL1Loss=0.1557
+ throughput/total_tokens=109,824,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=573/500000]
+ train/ActionL1Loss=0.1850
+ throughput/total_tokens=110,016,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=574/500000]
+ train/ActionL1Loss=0.1582
+ throughput/total_tokens=110,208,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=575/500000]
+ train/ActionL1Loss=0.1534
+ throughput/total_tokens=110,400,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=576/500000]
+ train/ActionL1Loss=0.1990
+ throughput/total_tokens=110,592,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=577/500000]
+ train/ActionL1Loss=0.1684
+ throughput/total_tokens=110,784,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=578/500000]
+ train/ActionL1Loss=0.1667
+ throughput/total_tokens=110,976,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=579/500000]
+ train/ActionL1Loss=0.1699
+ throughput/total_tokens=111,168,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=580/500000]
+ optim/total_grad_norm=12.20
+ train/ActionL1Loss=0.1562
+ throughput/total_tokens=111,360,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=581/500000]
+ train/ActionL1Loss=0.1721
+ throughput/total_tokens=111,552,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=582/500000]
+ train/ActionL1Loss=0.1681
+ throughput/total_tokens=111,744,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=583/500000]
+ train/ActionL1Loss=0.1900
+ throughput/total_tokens=111,936,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=584/500000]
+ train/ActionL1Loss=0.1779
+ throughput/total_tokens=112,128,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=585/500000]
+ train/ActionL1Loss=0.1568
+ throughput/total_tokens=112,320,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=586/500000]
+ train/ActionL1Loss=0.1544
+ throughput/total_tokens=112,512,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=587/500000]
+ train/ActionL1Loss=0.1512
+ throughput/total_tokens=112,704,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=588/500000]
+ train/ActionL1Loss=0.1533
+ throughput/total_tokens=112,896,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=589/500000]
+ train/ActionL1Loss=0.1479
+ throughput/total_tokens=113,088,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=590/500000]
+ train/ActionL1Loss=0.1514
+ throughput/total_tokens=113,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=591/500000]
+ train/ActionL1Loss=0.1670
+ throughput/total_tokens=113,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=592/500000]
+ train/ActionL1Loss=0.1521
+ throughput/total_tokens=113,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=593/500000]
+ train/ActionL1Loss=0.1696
+ throughput/total_tokens=113,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=594/500000]
+ train/ActionL1Loss=0.1888
+ throughput/total_tokens=114,048,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=595/500000]
+ train/ActionL1Loss=0.1999
+ throughput/total_tokens=114,240,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=596/500000]
+ train/ActionL1Loss=0.1522
+ throughput/total_tokens=114,432,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=597/500000]
+ train/ActionL1Loss=0.1696
+ throughput/total_tokens=114,624,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=598/500000]
+ train/ActionL1Loss=0.1681
+ throughput/total_tokens=114,816,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=599/500000]
+ train/ActionL1Loss=0.1478
+ throughput/total_tokens=115,008,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=600/500000]
+ optim/total_grad_norm=11.63
+ train/ActionL1Loss=0.1499
+ throughput/total_tokens=115,200,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=601/500000]
+ train/ActionL1Loss=0.1504
+ throughput/total_tokens=115,392,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=602/500000]
+ train/ActionL1Loss=0.1367
+ throughput/total_tokens=115,584,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=603/500000]
+ train/ActionL1Loss=0.1357
+ throughput/total_tokens=115,776,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=604/500000]
+ train/ActionL1Loss=0.1414
+ throughput/total_tokens=115,968,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=605/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=116,160,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=606/500000]
+ train/ActionL1Loss=0.1550
+ throughput/total_tokens=116,352,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=607/500000]
+ train/ActionL1Loss=0.1435
+ throughput/total_tokens=116,544,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=608/500000]
+ train/ActionL1Loss=0.1646
+ throughput/total_tokens=116,736,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=609/500000]
+ train/ActionL1Loss=0.1828
+ throughput/total_tokens=116,928,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=610/500000]
+ train/ActionL1Loss=0.1604
+ throughput/total_tokens=117,120,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=611/500000]
+ train/ActionL1Loss=0.1711
+ throughput/total_tokens=117,312,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=612/500000]
+ train/ActionL1Loss=0.1788
+ throughput/total_tokens=117,504,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=613/500000]
+ train/ActionL1Loss=0.1762
+ throughput/total_tokens=117,696,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=614/500000]
+ train/ActionL1Loss=0.1835
+ throughput/total_tokens=117,888,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=615/500000]
+ train/ActionL1Loss=0.1961
+ throughput/total_tokens=118,080,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=616/500000]
+ train/ActionL1Loss=0.1737
+ throughput/total_tokens=118,272,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=617/500000]
+ train/ActionL1Loss=0.1824
+ throughput/total_tokens=118,464,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=618/500000]
+ train/ActionL1Loss=0.1693
+ throughput/total_tokens=118,656,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=619/500000]
+ train/ActionL1Loss=0.1639
+ throughput/total_tokens=118,848,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=620/500000]
+ optim/total_grad_norm=10.83
+ train/ActionL1Loss=0.1511
+ throughput/total_tokens=119,040,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=621/500000]
+ train/ActionL1Loss=0.1893
+ throughput/total_tokens=119,232,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=622/500000]
+ train/ActionL1Loss=0.1579
+ throughput/total_tokens=119,424,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=623/500000]
+ train/ActionL1Loss=0.1715
+ throughput/total_tokens=119,616,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=624/500000]
+ train/ActionL1Loss=0.1778
+ throughput/total_tokens=119,808,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=625/500000]
+ train/ActionL1Loss=0.1698
+ throughput/total_tokens=120,000,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=626/500000]
+ train/ActionL1Loss=0.1776
+ throughput/total_tokens=120,192,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=627/500000]
+ train/ActionL1Loss=0.1676
+ throughput/total_tokens=120,384,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=628/500000]
+ train/ActionL1Loss=0.1607
+ throughput/total_tokens=120,576,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=629/500000]
+ train/ActionL1Loss=0.1709
+ throughput/total_tokens=120,768,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=630/500000]
+ train/ActionL1Loss=0.1916
+ throughput/total_tokens=120,960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=631/500000]
+ train/ActionL1Loss=0.1805
+ throughput/total_tokens=121,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=632/500000]
+ train/ActionL1Loss=0.1624
+ throughput/total_tokens=121,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=633/500000]
+ train/ActionL1Loss=0.1753
+ throughput/total_tokens=121,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=634/500000]
+ train/ActionL1Loss=0.1582
+ throughput/total_tokens=121,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=635/500000]
+ train/ActionL1Loss=0.1804
+ throughput/total_tokens=121,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=636/500000]
+ train/ActionL1Loss=0.1661
+ throughput/total_tokens=122,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=637/500000]
+ train/ActionL1Loss=0.1705
+ throughput/total_tokens=122,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=638/500000]
+ train/ActionL1Loss=0.1667
+ throughput/total_tokens=122,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=639/500000]
+ train/ActionL1Loss=0.1588
+ throughput/total_tokens=122,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=640/500000]
+ optim/total_grad_norm=7.792
+ train/ActionL1Loss=0.1705
+ throughput/total_tokens=122,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=641/500000]
+ train/ActionL1Loss=0.1604
+ throughput/total_tokens=123,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=642/500000]
+ train/ActionL1Loss=0.1535
+ throughput/total_tokens=123,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=643/500000]
+ train/ActionL1Loss=0.1611
+ throughput/total_tokens=123,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=644/500000]
+ train/ActionL1Loss=0.1560
+ throughput/total_tokens=123,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=645/500000]
+ train/ActionL1Loss=0.1524
+ throughput/total_tokens=123,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=646/500000]
+ train/ActionL1Loss=0.1444
+ throughput/total_tokens=124,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=647/500000]
+ train/ActionL1Loss=0.1762
+ throughput/total_tokens=124,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=648/500000]
+ train/ActionL1Loss=0.1676
+ throughput/total_tokens=124,416,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=649/500000]
+ train/ActionL1Loss=0.1642
+ throughput/total_tokens=124,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=650/500000]
+ train/ActionL1Loss=0.1733
+ throughput/total_tokens=124,800,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=651/500000]
+ train/ActionL1Loss=0.1750
+ throughput/total_tokens=124,992,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=652/500000]
+ train/ActionL1Loss=0.1695
+ throughput/total_tokens=125,184,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=653/500000]
+ train/ActionL1Loss=0.1622
+ throughput/total_tokens=125,376,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=654/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=125,568,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=655/500000]
+ train/ActionL1Loss=0.1591
+ throughput/total_tokens=125,760,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=656/500000]
+ train/ActionL1Loss=0.1495
+ throughput/total_tokens=125,952,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=657/500000]
+ train/ActionL1Loss=0.1560
+ throughput/total_tokens=126,144,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=658/500000]
+ train/ActionL1Loss=0.1567
+ throughput/total_tokens=126,336,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=659/500000]
+ train/ActionL1Loss=0.1450
+ throughput/total_tokens=126,528,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=660/500000]
+ optim/total_grad_norm=10.12
+ train/ActionL1Loss=0.1396
+ throughput/total_tokens=126,720,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=661/500000]
+ train/ActionL1Loss=0.1337
+ throughput/total_tokens=126,912,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=662/500000]
+ train/ActionL1Loss=0.1720
+ throughput/total_tokens=127,104,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=663/500000]
+ train/ActionL1Loss=0.1641
+ throughput/total_tokens=127,296,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=664/500000]
+ train/ActionL1Loss=0.1631
+ throughput/total_tokens=127,488,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=665/500000]
+ train/ActionL1Loss=0.1716
+ throughput/total_tokens=127,680,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=666/500000]
+ train/ActionL1Loss=0.1773
+ throughput/total_tokens=127,872,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=667/500000]
+ train/ActionL1Loss=0.1498
+ throughput/total_tokens=128,064,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=668/500000]
+ train/ActionL1Loss=0.1626
+ throughput/total_tokens=128,256,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=669/500000]
+ train/ActionL1Loss=0.1574
+ throughput/total_tokens=128,448,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=670/500000]
+ train/ActionL1Loss=0.1628
+ throughput/total_tokens=128,640,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=671/500000]
+ train/ActionL1Loss=0.1662
+ throughput/total_tokens=128,832,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=672/500000]
+ train/ActionL1Loss=0.1703
+ throughput/total_tokens=129,024,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=673/500000]
+ train/ActionL1Loss=0.1387
+ throughput/total_tokens=129,216,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=674/500000]
+ train/ActionL1Loss=0.1716
+ throughput/total_tokens=129,408,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=675/500000]
+ train/ActionL1Loss=0.1512
+ throughput/total_tokens=129,600,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=676/500000]
+ train/ActionL1Loss=0.1651
+ throughput/total_tokens=129,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=677/500000]
+ train/ActionL1Loss=0.1614
+ throughput/total_tokens=129,984,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=678/500000]
+ train/ActionL1Loss=0.1721
+ throughput/total_tokens=130,176,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=679/500000]
+ train/ActionL1Loss=0.1599
+ throughput/total_tokens=130,368,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=680/500000]
+ optim/total_grad_norm=9.022
+ train/ActionL1Loss=0.1375
+ throughput/total_tokens=130,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=681/500000]
+ train/ActionL1Loss=0.1611
+ throughput/total_tokens=130,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=682/500000]
+ train/ActionL1Loss=0.1788
+ throughput/total_tokens=130,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=683/500000]
+ train/ActionL1Loss=0.1680
+ throughput/total_tokens=131,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=684/500000]
+ train/ActionL1Loss=0.1723
+ throughput/total_tokens=131,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=685/500000]
+ train/ActionL1Loss=0.1567
+ throughput/total_tokens=131,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=686/500000]
+ train/ActionL1Loss=0.1299
+ throughput/total_tokens=131,712,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=687/500000]
+ train/ActionL1Loss=0.1533
+ throughput/total_tokens=131,904,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=688/500000]
+ train/ActionL1Loss=0.1442
+ throughput/total_tokens=132,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=689/500000]
+ train/ActionL1Loss=0.1522
+ throughput/total_tokens=132,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=690/500000]
+ train/ActionL1Loss=0.1421
+ throughput/total_tokens=132,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=691/500000]
+ train/ActionL1Loss=0.1593
+ throughput/total_tokens=132,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=692/500000]
+ train/ActionL1Loss=0.1645
+ throughput/total_tokens=132,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=693/500000]
+ train/ActionL1Loss=0.1529
+ throughput/total_tokens=133,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=694/500000]
+ train/ActionL1Loss=0.1509
+ throughput/total_tokens=133,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=695/500000]
+ train/ActionL1Loss=0.1389
+ throughput/total_tokens=133,440,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=696/500000]
+ train/ActionL1Loss=0.1402
+ throughput/total_tokens=133,632,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=697/500000]
+ train/ActionL1Loss=0.1487
+ throughput/total_tokens=133,824,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=698/500000]
+ train/ActionL1Loss=0.1517
+ throughput/total_tokens=134,016,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=699/500000]
+ train/ActionL1Loss=0.1461
+ throughput/total_tokens=134,208,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=700/500000]
+ optim/total_grad_norm=9.758
+ train/ActionL1Loss=0.1691
+ throughput/total_tokens=134,400,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=701/500000]
+ train/ActionL1Loss=0.1614
+ throughput/total_tokens=134,592,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=702/500000]
+ train/ActionL1Loss=0.1607
+ throughput/total_tokens=134,784,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=703/500000]
+ train/ActionL1Loss=0.1637
+ throughput/total_tokens=134,976,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=704/500000]
+ train/ActionL1Loss=0.1802
+ throughput/total_tokens=135,168,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=705/500000]
+ train/ActionL1Loss=0.1469
+ throughput/total_tokens=135,360,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=706/500000]
+ train/ActionL1Loss=0.1766
+ throughput/total_tokens=135,552,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=707/500000]
+ train/ActionL1Loss=0.1778
+ throughput/total_tokens=135,744,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=708/500000]
+ train/ActionL1Loss=0.1418
+ throughput/total_tokens=135,936,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=709/500000]
+ train/ActionL1Loss=0.1731
+ throughput/total_tokens=136,128,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=710/500000]
+ train/ActionL1Loss=0.1640
+ throughput/total_tokens=136,320,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=711/500000]
+ train/ActionL1Loss=0.1540
+ throughput/total_tokens=136,512,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=712/500000]
+ train/ActionL1Loss=0.1474
+ throughput/total_tokens=136,704,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=713/500000]
+ train/ActionL1Loss=0.1622
+ throughput/total_tokens=136,896,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=714/500000]
+ train/ActionL1Loss=0.1737
+ throughput/total_tokens=137,088,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=715/500000]
+ train/ActionL1Loss=0.1523
+ throughput/total_tokens=137,280,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=716/500000]
+ train/ActionL1Loss=0.1387
+ throughput/total_tokens=137,472,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=717/500000]
+ train/ActionL1Loss=0.1657
+ throughput/total_tokens=137,664,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=718/500000]
+ train/ActionL1Loss=0.1606
+ throughput/total_tokens=137,856,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=719/500000]
+ train/ActionL1Loss=0.1646
+ throughput/total_tokens=138,048,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=720/500000]
+ optim/total_grad_norm=10.56
+ train/ActionL1Loss=0.1626
+ throughput/total_tokens=138,240,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=721/500000]
+ train/ActionL1Loss=0.1555
+ throughput/total_tokens=138,432,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=722/500000]
+ train/ActionL1Loss=0.1369
+ throughput/total_tokens=138,624,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=723/500000]
+ train/ActionL1Loss=0.1379
+ throughput/total_tokens=138,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=724/500000]
+ train/ActionL1Loss=0.1367
+ throughput/total_tokens=139,008,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=725/500000]
+ train/ActionL1Loss=0.1612
+ throughput/total_tokens=139,200,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=726/500000]
+ train/ActionL1Loss=0.1397
+ throughput/total_tokens=139,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=727/500000]
+ train/ActionL1Loss=0.1375
+ throughput/total_tokens=139,584,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=728/500000]
+ train/ActionL1Loss=0.1399
+ throughput/total_tokens=139,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=729/500000]
+ train/ActionL1Loss=0.1435
+ throughput/total_tokens=139,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=730/500000]
+ train/ActionL1Loss=0.1410
+ throughput/total_tokens=140,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=731/500000]
+ train/ActionL1Loss=0.1403
+ throughput/total_tokens=140,352,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=732/500000]
+ train/ActionL1Loss=0.1429
+ throughput/total_tokens=140,544,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=733/500000]
+ train/ActionL1Loss=0.1438
+ throughput/total_tokens=140,736,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=734/500000]
+ train/ActionL1Loss=0.1489
+ throughput/total_tokens=140,928,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=735/500000]
+ train/ActionL1Loss=0.1561
+ throughput/total_tokens=141,120,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=736/500000]
+ train/ActionL1Loss=0.1653
+ throughput/total_tokens=141,312,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=737/500000]
+ train/ActionL1Loss=0.1463
+ throughput/total_tokens=141,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=738/500000]
+ train/ActionL1Loss=0.1842
+ throughput/total_tokens=141,696,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=739/500000]
+ train/ActionL1Loss=0.1796
+ throughput/total_tokens=141,888,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=740/500000]
+ optim/total_grad_norm=13.07
+ train/ActionL1Loss=0.1747
+ throughput/total_tokens=142,080,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=741/500000]
+ train/ActionL1Loss=0.1499
+ throughput/total_tokens=142,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=742/500000]
+ train/ActionL1Loss=0.1509
+ throughput/total_tokens=142,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=743/500000]
+ train/ActionL1Loss=0.1874
+ throughput/total_tokens=142,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=744/500000]
+ train/ActionL1Loss=0.1866
+ throughput/total_tokens=142,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=745/500000]
+ train/ActionL1Loss=0.1692
+ throughput/total_tokens=143,040,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=746/500000]
+ train/ActionL1Loss=0.1704
+ throughput/total_tokens=143,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=747/500000]
+ train/ActionL1Loss=0.1718
+ throughput/total_tokens=143,424,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=748/500000]
+ train/ActionL1Loss=0.1846
+ throughput/total_tokens=143,616,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=749/500000]
+ train/ActionL1Loss=0.1342
+ throughput/total_tokens=143,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=750/500000]
+ train/ActionL1Loss=0.1687
+ throughput/total_tokens=144,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=751/500000]
+ train/ActionL1Loss=0.1737
+ throughput/total_tokens=144,192,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=752/500000]
+ train/ActionL1Loss=0.1428
+ throughput/total_tokens=144,384,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=753/500000]
+ train/ActionL1Loss=0.1427
+ throughput/total_tokens=144,576,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=754/500000]
+ train/ActionL1Loss=0.1580
+ throughput/total_tokens=144,768,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=755/500000]
+ train/ActionL1Loss=0.1741
+ throughput/total_tokens=144,960,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=756/500000]
+ train/ActionL1Loss=0.1640
+ throughput/total_tokens=145,152,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=757/500000]
+ train/ActionL1Loss=0.1556
+ throughput/total_tokens=145,344,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=758/500000]
+ train/ActionL1Loss=0.1626
+ throughput/total_tokens=145,536,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=759/500000]
+ train/ActionL1Loss=0.1660
+ throughput/total_tokens=145,728,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=760/500000]
+ optim/total_grad_norm=11.49
+ train/ActionL1Loss=0.1739
+ throughput/total_tokens=145,920,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=761/500000]
+ train/ActionL1Loss=0.1826
+ throughput/total_tokens=146,112,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=762/500000]
+ train/ActionL1Loss=0.1863
+ throughput/total_tokens=146,304,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=763/500000]
+ train/ActionL1Loss=0.1606
+ throughput/total_tokens=146,496,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=764/500000]
+ train/ActionL1Loss=0.1537
+ throughput/total_tokens=146,688,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=765/500000]
+ train/ActionL1Loss=0.1648
+ throughput/total_tokens=146,880,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=766/500000]
+ train/ActionL1Loss=0.1854
+ throughput/total_tokens=147,072,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=767/500000]
+ train/ActionL1Loss=0.1783
+ throughput/total_tokens=147,264,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=768/500000]
+ train/ActionL1Loss=0.1668
+ throughput/total_tokens=147,456,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=769/500000]
+ train/ActionL1Loss=0.1766
+ throughput/total_tokens=147,648,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=770/500000]
+ train/ActionL1Loss=0.1656
+ throughput/total_tokens=147,840,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=771/500000]
+ train/ActionL1Loss=0.1889
+ throughput/total_tokens=148,032,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=772/500000]
+ train/ActionL1Loss=0.1883
+ throughput/total_tokens=148,224,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=773/500000]
+ train/ActionL1Loss=0.1834
+ throughput/total_tokens=148,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=774/500000]
+ train/ActionL1Loss=0.1763
+ throughput/total_tokens=148,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=775/500000]
+ train/ActionL1Loss=0.1466
+ throughput/total_tokens=148,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=776/500000]
+ train/ActionL1Loss=0.1513
+ throughput/total_tokens=148,992,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=777/500000]
+ train/ActionL1Loss=0.1714
+ throughput/total_tokens=149,184,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=778/500000]
+ train/ActionL1Loss=0.1584
+ throughput/total_tokens=149,376,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=779/500000]
+ train/ActionL1Loss=0.1744
+ throughput/total_tokens=149,568,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=780/500000]
+ optim/total_grad_norm=11.31
+ train/ActionL1Loss=0.1472
+ throughput/total_tokens=149,760,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=781/500000]
+ train/ActionL1Loss=0.1600
+ throughput/total_tokens=149,952,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=782/500000]
+ train/ActionL1Loss=0.1603
+ throughput/total_tokens=150,144,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=783/500000]
+ train/ActionL1Loss=0.1467
+ throughput/total_tokens=150,336,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=784/500000]
+ train/ActionL1Loss=0.1657
+ throughput/total_tokens=150,528,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=785/500000]
+ train/ActionL1Loss=0.1410
+ throughput/total_tokens=150,720,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=786/500000]
+ train/ActionL1Loss=0.1562
+ throughput/total_tokens=150,912,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=787/500000]
+ train/ActionL1Loss=0.1957
+ throughput/total_tokens=151,104,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=788/500000]
+ train/ActionL1Loss=0.1814
+ throughput/total_tokens=151,296,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=789/500000]
+ train/ActionL1Loss=0.1674
+ throughput/total_tokens=151,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=790/500000]
+ train/ActionL1Loss=0.1786
+ throughput/total_tokens=151,680,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=791/500000]
+ train/ActionL1Loss=0.1749
+ throughput/total_tokens=151,872,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=792/500000]
+ train/ActionL1Loss=0.1912
+ throughput/total_tokens=152,064,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=793/500000]
+ train/ActionL1Loss=0.1536
+ throughput/total_tokens=152,256,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=794/500000]
+ train/ActionL1Loss=0.1580
+ throughput/total_tokens=152,448,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=795/500000]
+ train/ActionL1Loss=0.1824
+ throughput/total_tokens=152,640,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=796/500000]
+ train/ActionL1Loss=0.1493
+ throughput/total_tokens=152,832,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=797/500000]
+ train/ActionL1Loss=0.1488
+ throughput/total_tokens=153,024,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=798/500000]
+ train/ActionL1Loss=0.1645
+ throughput/total_tokens=153,216,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=799/500000]
+ train/ActionL1Loss=0.1973
+ throughput/total_tokens=153,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=800/500000]
+ optim/total_grad_norm=9.145
+ train/ActionL1Loss=0.1505
+ throughput/total_tokens=153,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=801/500000]
+ train/ActionL1Loss=0.1653
+ throughput/total_tokens=153,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=802/500000]
+ train/ActionL1Loss=0.1586
+ throughput/total_tokens=153,984,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=803/500000]
+ train/ActionL1Loss=0.1435
+ throughput/total_tokens=154,176,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=804/500000]
+ train/ActionL1Loss=0.1600
+ throughput/total_tokens=154,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=805/500000]
+ train/ActionL1Loss=0.1432
+ throughput/total_tokens=154,560,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=806/500000]
+ train/ActionL1Loss=0.1667
+ throughput/total_tokens=154,752,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=807/500000]
+ train/ActionL1Loss=0.1732
+ throughput/total_tokens=154,944,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=808/500000]
+ train/ActionL1Loss=0.1419
+ throughput/total_tokens=155,136,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=809/500000]
+ train/ActionL1Loss=0.1734
+ throughput/total_tokens=155,328,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=810/500000]
+ train/ActionL1Loss=0.1556
+ throughput/total_tokens=155,520,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=811/500000]
+ train/ActionL1Loss=0.1434
+ throughput/total_tokens=155,712,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=812/500000]
+ train/ActionL1Loss=0.1686
+ throughput/total_tokens=155,904,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=813/500000]
+ train/ActionL1Loss=0.1577
+ throughput/total_tokens=156,096,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=814/500000]
+ train/ActionL1Loss=0.1399
+ throughput/total_tokens=156,288,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=815/500000]
+ train/ActionL1Loss=0.1553
+ throughput/total_tokens=156,480,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=816/500000]
+ train/ActionL1Loss=0.1439
+ throughput/total_tokens=156,672,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=817/500000]
+ train/ActionL1Loss=0.1585
+ throughput/total_tokens=156,864,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=818/500000]
+ train/ActionL1Loss=0.1449
+ throughput/total_tokens=157,056,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=819/500000]
+ train/ActionL1Loss=0.1586
+ throughput/total_tokens=157,248,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=820/500000]
+ optim/total_grad_norm=7.629
+ train/ActionL1Loss=0.1517
+ throughput/total_tokens=157,440,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=821/500000]
+ train/ActionL1Loss=0.1475
+ throughput/total_tokens=157,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=822/500000]
+ train/ActionL1Loss=0.1590
+ throughput/total_tokens=157,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=823/500000]
+ train/ActionL1Loss=0.1694
+ throughput/total_tokens=158,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=824/500000]
+ train/ActionL1Loss=0.1549
+ throughput/total_tokens=158,208,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=825/500000]
+ train/ActionL1Loss=0.1279
+ throughput/total_tokens=158,400,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=826/500000]
+ train/ActionL1Loss=0.1445
+ throughput/total_tokens=158,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=827/500000]
+ train/ActionL1Loss=0.1516
+ throughput/total_tokens=158,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=828/500000]
+ train/ActionL1Loss=0.1690
+ throughput/total_tokens=158,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=829/500000]
+ train/ActionL1Loss=0.1370
+ throughput/total_tokens=159,168,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=830/500000]
+ train/ActionL1Loss=0.1352
+ throughput/total_tokens=159,360,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=831/500000]
+ train/ActionL1Loss=0.1387
+ throughput/total_tokens=159,552,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=832/500000]
+ train/ActionL1Loss=0.1737
+ throughput/total_tokens=159,744,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=833/500000]
+ train/ActionL1Loss=0.1650
+ throughput/total_tokens=159,936,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=834/500000]
+ train/ActionL1Loss=0.1371
+ throughput/total_tokens=160,128,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=835/500000]
+ train/ActionL1Loss=0.1424
+ throughput/total_tokens=160,320,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=836/500000]
+ train/ActionL1Loss=0.1396
+ throughput/total_tokens=160,512,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=837/500000]
+ train/ActionL1Loss=0.1503
+ throughput/total_tokens=160,704,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=838/500000]
+ train/ActionL1Loss=0.1433
+ throughput/total_tokens=160,896,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=839/500000]
+ train/ActionL1Loss=0.1452
+ throughput/total_tokens=161,088,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=840/500000]
+ optim/total_grad_norm=9.636
+ train/ActionL1Loss=0.1341
+ throughput/total_tokens=161,280,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=841/500000]
+ train/ActionL1Loss=0.1695
+ throughput/total_tokens=161,472,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=842/500000]
+ train/ActionL1Loss=0.1436
+ throughput/total_tokens=161,664,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=843/500000]
+ train/ActionL1Loss=0.1424
+ throughput/total_tokens=161,856,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=844/500000]
+ train/ActionL1Loss=0.1423
+ throughput/total_tokens=162,048,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=845/500000]
+ train/ActionL1Loss=0.1476
+ throughput/total_tokens=162,240,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=846/500000]
+ train/ActionL1Loss=0.1308
+ throughput/total_tokens=162,432,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=847/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=162,624,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=848/500000]
+ train/ActionL1Loss=0.1466
+ throughput/total_tokens=162,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=849/500000]
+ train/ActionL1Loss=0.1697
+ throughput/total_tokens=163,008,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=850/500000]
+ train/ActionL1Loss=0.1445
+ throughput/total_tokens=163,200,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=851/500000]
+ train/ActionL1Loss=0.1282
+ throughput/total_tokens=163,392,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=852/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=163,584,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=853/500000]
+ train/ActionL1Loss=0.1740
+ throughput/total_tokens=163,776,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=854/500000]
+ train/ActionL1Loss=0.1745
+ throughput/total_tokens=163,968,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=855/500000]
+ train/ActionL1Loss=0.1307
+ throughput/total_tokens=164,160,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=856/500000]
+ train/ActionL1Loss=0.1560
+ throughput/total_tokens=164,352,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=857/500000]
+ train/ActionL1Loss=0.1933
+ throughput/total_tokens=164,544,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=858/500000]
+ train/ActionL1Loss=0.1667
+ throughput/total_tokens=164,736,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=859/500000]
+ train/ActionL1Loss=0.1447
+ throughput/total_tokens=164,928,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=860/500000]
+ optim/total_grad_norm=10.16
+ train/ActionL1Loss=0.1464
+ throughput/total_tokens=165,120,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=861/500000]
+ train/ActionL1Loss=0.1788
+ throughput/total_tokens=165,312,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=862/500000]
+ train/ActionL1Loss=0.1575
+ throughput/total_tokens=165,504,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=863/500000]
+ train/ActionL1Loss=0.1543
+ throughput/total_tokens=165,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=864/500000]
+ train/ActionL1Loss=0.1463
+ throughput/total_tokens=165,888,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=865/500000]
+ train/ActionL1Loss=0.1722
+ throughput/total_tokens=166,080,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=866/500000]
+ train/ActionL1Loss=0.1714
+ throughput/total_tokens=166,272,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=867/500000]
+ train/ActionL1Loss=0.1578
+ throughput/total_tokens=166,464,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=868/500000]
+ train/ActionL1Loss=0.1680
+ throughput/total_tokens=166,656,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=869/500000]
+ train/ActionL1Loss=0.1692
+ throughput/total_tokens=166,848,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=870/500000]
+ train/ActionL1Loss=0.1468
+ throughput/total_tokens=167,040,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=871/500000]
+ train/ActionL1Loss=0.1555
+ throughput/total_tokens=167,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=872/500000]
+ train/ActionL1Loss=0.1717
+ throughput/total_tokens=167,424,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=873/500000]
+ train/ActionL1Loss=0.1743
+ throughput/total_tokens=167,616,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=874/500000]
+ train/ActionL1Loss=0.1614
+ throughput/total_tokens=167,808,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=875/500000]
+ train/ActionL1Loss=0.1628
+ throughput/total_tokens=168,000,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=876/500000]
+ train/ActionL1Loss=0.1586
+ throughput/total_tokens=168,192,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=877/500000]
+ train/ActionL1Loss=0.1576
+ throughput/total_tokens=168,384,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=878/500000]
+ train/ActionL1Loss=0.1539
+ throughput/total_tokens=168,576,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=879/500000]
+ train/ActionL1Loss=0.1627
+ throughput/total_tokens=168,768,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=880/500000]
+ optim/total_grad_norm=11.79
+ train/ActionL1Loss=0.1584
+ throughput/total_tokens=168,960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=881/500000]
+ train/ActionL1Loss=0.1448
+ throughput/total_tokens=169,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=882/500000]
+ train/ActionL1Loss=0.1452
+ throughput/total_tokens=169,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=883/500000]
+ train/ActionL1Loss=0.1612
+ throughput/total_tokens=169,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=884/500000]
+ train/ActionL1Loss=0.1500
+ throughput/total_tokens=169,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=885/500000]
+ train/ActionL1Loss=0.1677
+ throughput/total_tokens=169,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=886/500000]
+ train/ActionL1Loss=0.1593
+ throughput/total_tokens=170,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=887/500000]
+ train/ActionL1Loss=0.1499
+ throughput/total_tokens=170,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=888/500000]
+ train/ActionL1Loss=0.1639
+ throughput/total_tokens=170,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=889/500000]
+ train/ActionL1Loss=0.1532
+ throughput/total_tokens=170,688,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=890/500000]
+ train/ActionL1Loss=0.1922
+ throughput/total_tokens=170,880,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=891/500000]
+ train/ActionL1Loss=0.1461
+ throughput/total_tokens=171,072,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=892/500000]
+ train/ActionL1Loss=0.1343
+ throughput/total_tokens=171,264,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=893/500000]
+ train/ActionL1Loss=0.1486
+ throughput/total_tokens=171,456,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=894/500000]
+ train/ActionL1Loss=0.1501
+ throughput/total_tokens=171,648,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=895/500000]
+ train/ActionL1Loss=0.1493
+ throughput/total_tokens=171,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=896/500000]
+ train/ActionL1Loss=0.1372
+ throughput/total_tokens=172,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=897/500000]
+ train/ActionL1Loss=0.1474
+ throughput/total_tokens=172,224,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=898/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=172,416,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=899/500000]
+ train/ActionL1Loss=0.1513
+ throughput/total_tokens=172,608,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=900/500000]
+ optim/total_grad_norm=10.51
+ train/ActionL1Loss=0.1317
+ throughput/total_tokens=172,800,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=901/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=172,992,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=902/500000]
+ train/ActionL1Loss=0.1654
+ throughput/total_tokens=173,184,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=903/500000]
+ train/ActionL1Loss=0.1364
+ throughput/total_tokens=173,376,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0482
+[step=904/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=173,568,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=905/500000]
+ train/ActionL1Loss=0.1406
+ throughput/total_tokens=173,760,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=906/500000]
+ train/ActionL1Loss=0.1363
+ throughput/total_tokens=173,952,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=907/500000]
+ train/ActionL1Loss=0.1503
+ throughput/total_tokens=174,144,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=908/500000]
+ train/ActionL1Loss=0.1787
+ throughput/total_tokens=174,336,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=909/500000]
+ train/ActionL1Loss=0.1686
+ throughput/total_tokens=174,528,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=910/500000]
+ train/ActionL1Loss=0.1576
+ throughput/total_tokens=174,720,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=911/500000]
+ train/ActionL1Loss=0.1643
+ throughput/total_tokens=174,912,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=912/500000]
+ train/ActionL1Loss=0.1362
+ throughput/total_tokens=175,104,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=913/500000]
+ train/ActionL1Loss=0.1598
+ throughput/total_tokens=175,296,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=914/500000]
+ train/ActionL1Loss=0.1930
+ throughput/total_tokens=175,488,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=915/500000]
+ train/ActionL1Loss=0.1680
+ throughput/total_tokens=175,680,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=916/500000]
+ train/ActionL1Loss=0.1669
+ throughput/total_tokens=175,872,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=917/500000]
+ train/ActionL1Loss=0.1588
+ throughput/total_tokens=176,064,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=918/500000]
+ train/ActionL1Loss=0.1490
+ throughput/total_tokens=176,256,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=919/500000]
+ train/ActionL1Loss=0.1690
+ throughput/total_tokens=176,448,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=920/500000]
+ optim/total_grad_norm=9.688
+ train/ActionL1Loss=0.1555
+ throughput/total_tokens=176,640,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=921/500000]
+ train/ActionL1Loss=0.1434
+ throughput/total_tokens=176,832,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=922/500000]
+ train/ActionL1Loss=0.1468
+ throughput/total_tokens=177,024,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=923/500000]
+ train/ActionL1Loss=0.1595
+ throughput/total_tokens=177,216,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=924/500000]
+ train/ActionL1Loss=0.1475
+ throughput/total_tokens=177,408,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=925/500000]
+ train/ActionL1Loss=0.1507
+ throughput/total_tokens=177,600,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=926/500000]
+ train/ActionL1Loss=0.1558
+ throughput/total_tokens=177,792,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=927/500000]
+ train/ActionL1Loss=0.1657
+ throughput/total_tokens=177,984,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=928/500000]
+ train/ActionL1Loss=0.1596
+ throughput/total_tokens=178,176,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=929/500000]
+ train/ActionL1Loss=0.1429
+ throughput/total_tokens=178,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=930/500000]
+ train/ActionL1Loss=0.1374
+ throughput/total_tokens=178,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=931/500000]
+ train/ActionL1Loss=0.1604
+ throughput/total_tokens=178,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=932/500000]
+ train/ActionL1Loss=0.1634
+ throughput/total_tokens=178,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=933/500000]
+ train/ActionL1Loss=0.1583
+ throughput/total_tokens=179,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=934/500000]
+ train/ActionL1Loss=0.1590
+ throughput/total_tokens=179,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=935/500000]
+ train/ActionL1Loss=0.1566
+ throughput/total_tokens=179,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=936/500000]
+ train/ActionL1Loss=0.1686
+ throughput/total_tokens=179,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=937/500000]
+ train/ActionL1Loss=0.1293
+ throughput/total_tokens=179,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=938/500000]
+ train/ActionL1Loss=0.1465
+ throughput/total_tokens=180,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=939/500000]
+ train/ActionL1Loss=0.1619
+ throughput/total_tokens=180,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=940/500000]
+ optim/total_grad_norm=10.06
+ train/ActionL1Loss=0.1491
+ throughput/total_tokens=180,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=941/500000]
+ train/ActionL1Loss=0.1804
+ throughput/total_tokens=180,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=942/500000]
+ train/ActionL1Loss=0.1496
+ throughput/total_tokens=180,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=943/500000]
+ train/ActionL1Loss=0.1239
+ throughput/total_tokens=181,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=944/500000]
+ train/ActionL1Loss=0.1376
+ throughput/total_tokens=181,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=945/500000]
+ train/ActionL1Loss=0.1449
+ throughput/total_tokens=181,440,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=946/500000]
+ train/ActionL1Loss=0.1448
+ throughput/total_tokens=181,632,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=947/500000]
+ train/ActionL1Loss=0.1421
+ throughput/total_tokens=181,824,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=948/500000]
+ train/ActionL1Loss=0.1656
+ throughput/total_tokens=182,016,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=949/500000]
+ train/ActionL1Loss=0.1326
+ throughput/total_tokens=182,208,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=950/500000]
+ train/ActionL1Loss=0.1490
+ throughput/total_tokens=182,400,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=951/500000]
+ train/ActionL1Loss=0.1549
+ throughput/total_tokens=182,592,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=952/500000]
+ train/ActionL1Loss=0.1402
+ throughput/total_tokens=182,784,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=953/500000]
+ train/ActionL1Loss=0.1353
+ throughput/total_tokens=182,976,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=954/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=183,168,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=955/500000]
+ train/ActionL1Loss=0.1315
+ throughput/total_tokens=183,360,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=956/500000]
+ train/ActionL1Loss=0.1447
+ throughput/total_tokens=183,552,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=957/500000]
+ train/ActionL1Loss=0.1222
+ throughput/total_tokens=183,744,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=958/500000]
+ train/ActionL1Loss=0.1395
+ throughput/total_tokens=183,936,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=959/500000]
+ train/ActionL1Loss=0.1518
+ throughput/total_tokens=184,128,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=960/500000]
+ optim/total_grad_norm=8.135
+ train/ActionL1Loss=0.1476
+ throughput/total_tokens=184,320,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=961/500000]
+ train/ActionL1Loss=0.1299
+ throughput/total_tokens=184,512,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=962/500000]
+ train/ActionL1Loss=0.1476
+ throughput/total_tokens=184,704,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=963/500000]
+ train/ActionL1Loss=0.1244
+ throughput/total_tokens=184,896,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=964/500000]
+ train/ActionL1Loss=0.1293
+ throughput/total_tokens=185,088,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=965/500000]
+ train/ActionL1Loss=0.1355
+ throughput/total_tokens=185,280,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=966/500000]
+ train/ActionL1Loss=0.1192
+ throughput/total_tokens=185,472,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=967/500000]
+ train/ActionL1Loss=0.1280
+ throughput/total_tokens=185,664,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=968/500000]
+ train/ActionL1Loss=0.1541
+ throughput/total_tokens=185,856,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=969/500000]
+ train/ActionL1Loss=0.1724
+ throughput/total_tokens=186,048,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=970/500000]
+ train/ActionL1Loss=0.1471
+ throughput/total_tokens=186,240,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=971/500000]
+ train/ActionL1Loss=0.1426
+ throughput/total_tokens=186,432,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=972/500000]
+ train/ActionL1Loss=0.1510
+ throughput/total_tokens=186,624,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=973/500000]
+ train/ActionL1Loss=0.1675
+ throughput/total_tokens=186,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=974/500000]
+ train/ActionL1Loss=0.1430
+ throughput/total_tokens=187,008,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=975/500000]
+ train/ActionL1Loss=0.1444
+ throughput/total_tokens=187,200,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=976/500000]
+ train/ActionL1Loss=0.1458
+ throughput/total_tokens=187,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=977/500000]
+ train/ActionL1Loss=0.1407
+ throughput/total_tokens=187,584,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=978/500000]
+ train/ActionL1Loss=0.1449
+ throughput/total_tokens=187,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=979/500000]
+ train/ActionL1Loss=0.1397
+ throughput/total_tokens=187,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=980/500000]
+ optim/total_grad_norm=9.094
+ train/ActionL1Loss=0.1425
+ throughput/total_tokens=188,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=981/500000]
+ train/ActionL1Loss=0.1390
+ throughput/total_tokens=188,352,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=982/500000]
+ train/ActionL1Loss=0.1411
+ throughput/total_tokens=188,544,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=983/500000]
+ train/ActionL1Loss=0.1551
+ throughput/total_tokens=188,736,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=984/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=188,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=985/500000]
+ train/ActionL1Loss=0.1313
+ throughput/total_tokens=189,120,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=986/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=189,312,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=987/500000]
+ train/ActionL1Loss=0.1339
+ throughput/total_tokens=189,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=988/500000]
+ train/ActionL1Loss=0.1336
+ throughput/total_tokens=189,696,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=989/500000]
+ train/ActionL1Loss=0.1321
+ throughput/total_tokens=189,888,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=990/500000]
+ train/ActionL1Loss=0.1378
+ throughput/total_tokens=190,080,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=991/500000]
+ train/ActionL1Loss=0.1521
+ throughput/total_tokens=190,272,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=992/500000]
+ train/ActionL1Loss=0.1443
+ throughput/total_tokens=190,464,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=993/500000]
+ train/ActionL1Loss=0.1345
+ throughput/total_tokens=190,656,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=994/500000]
+ train/ActionL1Loss=0.1419
+ throughput/total_tokens=190,848,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=995/500000]
+ train/ActionL1Loss=0.1351
+ throughput/total_tokens=191,040,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=996/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=191,232,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=997/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=191,424,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=998/500000]
+ train/ActionL1Loss=0.1528
+ throughput/total_tokens=191,616,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=999/500000]
+ train/ActionL1Loss=0.1400
+ throughput/total_tokens=191,808,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1000/500000]
+ optim/total_grad_norm=8.571
+ train/ActionL1Loss=0.1402
+ throughput/total_tokens=192,000,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+09/30 [19:05:19] INFO | >> Saving config... checkpoint.py:608
+09/30 [19:06:02] INFO | >> Saving model state... checkpoint.py:796
+09/30 [19:07:10] INFO | >> Saving optim state... checkpoint.py:811
+09/30 [19:08:42] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=1001/500000]
+ train/ActionL1Loss=0.1380
+ throughput/total_tokens=192,192,000
+ throughput/device/tokens_per_second=1,208
+ throughput/device/batches_per_second=0.0504
+[step=1002/500000]
+ train/ActionL1Loss=0.1510
+ throughput/total_tokens=192,384,000
+ throughput/device/tokens_per_second=1,181
+ throughput/device/batches_per_second=0.0492
+[step=1003/500000]
+ train/ActionL1Loss=0.1533
+ throughput/total_tokens=192,576,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=1004/500000]
+ train/ActionL1Loss=0.1215
+ throughput/total_tokens=192,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1005/500000]
+ train/ActionL1Loss=0.1401
+ throughput/total_tokens=192,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1006/500000]
+ train/ActionL1Loss=0.1456
+ throughput/total_tokens=193,152,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1007/500000]
+ train/ActionL1Loss=0.1613
+ throughput/total_tokens=193,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1008/500000]
+ train/ActionL1Loss=0.1520
+ throughput/total_tokens=193,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1009/500000]
+ train/ActionL1Loss=0.1174
+ throughput/total_tokens=193,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1010/500000]
+ train/ActionL1Loss=0.1458
+ throughput/total_tokens=193,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1011/500000]
+ train/ActionL1Loss=0.1642
+ throughput/total_tokens=194,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1012/500000]
+ train/ActionL1Loss=0.1559
+ throughput/total_tokens=194,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1013/500000]
+ train/ActionL1Loss=0.1421
+ throughput/total_tokens=194,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1014/500000]
+ train/ActionL1Loss=0.1449
+ throughput/total_tokens=194,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1015/500000]
+ train/ActionL1Loss=0.1318
+ throughput/total_tokens=194,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1016/500000]
+ train/ActionL1Loss=0.1640
+ throughput/total_tokens=195,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1017/500000]
+ train/ActionL1Loss=0.1886
+ throughput/total_tokens=195,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1018/500000]
+ train/ActionL1Loss=0.1359
+ throughput/total_tokens=195,456,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1019/500000]
+ train/ActionL1Loss=0.1465
+ throughput/total_tokens=195,648,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1020/500000]
+ optim/total_grad_norm=12.15
+ train/ActionL1Loss=0.1839
+ throughput/total_tokens=195,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1021/500000]
+ train/ActionL1Loss=0.1961
+ throughput/total_tokens=196,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1022/500000]
+ train/ActionL1Loss=0.1845
+ throughput/total_tokens=196,224,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1023/500000]
+ train/ActionL1Loss=0.1471
+ throughput/total_tokens=196,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1024/500000]
+ train/ActionL1Loss=0.1432
+ throughput/total_tokens=196,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1025/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=196,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1026/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=196,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1027/500000]
+ train/ActionL1Loss=0.1532
+ throughput/total_tokens=197,184,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1028/500000]
+ train/ActionL1Loss=0.1339
+ throughput/total_tokens=197,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1029/500000]
+ train/ActionL1Loss=0.1344
+ throughput/total_tokens=197,568,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1030/500000]
+ train/ActionL1Loss=0.1538
+ throughput/total_tokens=197,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1031/500000]
+ train/ActionL1Loss=0.1487
+ throughput/total_tokens=197,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1032/500000]
+ train/ActionL1Loss=0.1556
+ throughput/total_tokens=198,144,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1033/500000]
+ train/ActionL1Loss=0.1649
+ throughput/total_tokens=198,336,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1034/500000]
+ train/ActionL1Loss=0.1547
+ throughput/total_tokens=198,528,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1035/500000]
+ train/ActionL1Loss=0.1602
+ throughput/total_tokens=198,720,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1036/500000]
+ train/ActionL1Loss=0.1372
+ throughput/total_tokens=198,912,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1037/500000]
+ train/ActionL1Loss=0.1455
+ throughput/total_tokens=199,104,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1038/500000]
+ train/ActionL1Loss=0.1631
+ throughput/total_tokens=199,296,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1039/500000]
+ train/ActionL1Loss=0.1459
+ throughput/total_tokens=199,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1040/500000]
+ optim/total_grad_norm=7.662
+ train/ActionL1Loss=0.1411
+ throughput/total_tokens=199,680,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1041/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=199,872,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1042/500000]
+ train/ActionL1Loss=0.1367
+ throughput/total_tokens=200,064,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1043/500000]
+ train/ActionL1Loss=0.1429
+ throughput/total_tokens=200,256,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1044/500000]
+ train/ActionL1Loss=0.1520
+ throughput/total_tokens=200,448,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1045/500000]
+ train/ActionL1Loss=0.1299
+ throughput/total_tokens=200,640,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1046/500000]
+ train/ActionL1Loss=0.1317
+ throughput/total_tokens=200,832,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1047/500000]
+ train/ActionL1Loss=0.1491
+ throughput/total_tokens=201,024,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1048/500000]
+ train/ActionL1Loss=0.1373
+ throughput/total_tokens=201,216,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1049/500000]
+ train/ActionL1Loss=0.1405
+ throughput/total_tokens=201,408,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1050/500000]
+ train/ActionL1Loss=0.1661
+ throughput/total_tokens=201,600,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1051/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=201,792,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1052/500000]
+ train/ActionL1Loss=0.1347
+ throughput/total_tokens=201,984,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1053/500000]
+ train/ActionL1Loss=0.1521
+ throughput/total_tokens=202,176,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1054/500000]
+ train/ActionL1Loss=0.1615
+ throughput/total_tokens=202,368,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1055/500000]
+ train/ActionL1Loss=0.1477
+ throughput/total_tokens=202,560,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1056/500000]
+ train/ActionL1Loss=0.1579
+ throughput/total_tokens=202,752,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1057/500000]
+ train/ActionL1Loss=0.1211
+ throughput/total_tokens=202,944,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1058/500000]
+ train/ActionL1Loss=0.1509
+ throughput/total_tokens=203,136,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1059/500000]
+ train/ActionL1Loss=0.1479
+ throughput/total_tokens=203,328,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1060/500000]
+ optim/total_grad_norm=8.221
+ train/ActionL1Loss=0.1295
+ throughput/total_tokens=203,520,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1061/500000]
+ train/ActionL1Loss=0.1427
+ throughput/total_tokens=203,712,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1062/500000]
+ train/ActionL1Loss=0.1454
+ throughput/total_tokens=203,904,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1063/500000]
+ train/ActionL1Loss=0.1301
+ throughput/total_tokens=204,096,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1064/500000]
+ train/ActionL1Loss=0.1516
+ throughput/total_tokens=204,288,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1065/500000]
+ train/ActionL1Loss=0.1365
+ throughput/total_tokens=204,480,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1066/500000]
+ train/ActionL1Loss=0.1634
+ throughput/total_tokens=204,672,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1067/500000]
+ train/ActionL1Loss=0.1326
+ throughput/total_tokens=204,864,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1068/500000]
+ train/ActionL1Loss=0.1410
+ throughput/total_tokens=205,056,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1069/500000]
+ train/ActionL1Loss=0.1340
+ throughput/total_tokens=205,248,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1070/500000]
+ train/ActionL1Loss=0.1576
+ throughput/total_tokens=205,440,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1071/500000]
+ train/ActionL1Loss=0.1407
+ throughput/total_tokens=205,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1072/500000]
+ train/ActionL1Loss=0.1353
+ throughput/total_tokens=205,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1073/500000]
+ train/ActionL1Loss=0.1584
+ throughput/total_tokens=206,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1074/500000]
+ train/ActionL1Loss=0.1409
+ throughput/total_tokens=206,208,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1075/500000]
+ train/ActionL1Loss=0.1333
+ throughput/total_tokens=206,400,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1076/500000]
+ train/ActionL1Loss=0.1377
+ throughput/total_tokens=206,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1077/500000]
+ train/ActionL1Loss=0.1546
+ throughput/total_tokens=206,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1078/500000]
+ train/ActionL1Loss=0.1409
+ throughput/total_tokens=206,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1079/500000]
+ train/ActionL1Loss=0.1694
+ throughput/total_tokens=207,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1080/500000]
+ optim/total_grad_norm=8.767
+ train/ActionL1Loss=0.1425
+ throughput/total_tokens=207,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1081/500000]
+ train/ActionL1Loss=0.1441
+ throughput/total_tokens=207,552,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1082/500000]
+ train/ActionL1Loss=0.1550
+ throughput/total_tokens=207,744,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1083/500000]
+ train/ActionL1Loss=0.1369
+ throughput/total_tokens=207,936,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1084/500000]
+ train/ActionL1Loss=0.1423
+ throughput/total_tokens=208,128,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1085/500000]
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=208,320,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1086/500000]
+ train/ActionL1Loss=0.1557
+ throughput/total_tokens=208,512,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1087/500000]
+ train/ActionL1Loss=0.1539
+ throughput/total_tokens=208,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1088/500000]
+ train/ActionL1Loss=0.1349
+ throughput/total_tokens=208,896,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1089/500000]
+ train/ActionL1Loss=0.1569
+ throughput/total_tokens=209,088,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1090/500000]
+ train/ActionL1Loss=0.1452
+ throughput/total_tokens=209,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1091/500000]
+ train/ActionL1Loss=0.1225
+ throughput/total_tokens=209,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1092/500000]
+ train/ActionL1Loss=0.1646
+ throughput/total_tokens=209,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1093/500000]
+ train/ActionL1Loss=0.1394
+ throughput/total_tokens=209,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1094/500000]
+ train/ActionL1Loss=0.1264
+ throughput/total_tokens=210,048,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1095/500000]
+ train/ActionL1Loss=0.1501
+ throughput/total_tokens=210,240,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1096/500000]
+ train/ActionL1Loss=0.1453
+ throughput/total_tokens=210,432,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1097/500000]
+ train/ActionL1Loss=0.1411
+ throughput/total_tokens=210,624,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1098/500000]
+ train/ActionL1Loss=0.1420
+ throughput/total_tokens=210,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1099/500000]
+ train/ActionL1Loss=0.1357
+ throughput/total_tokens=211,008,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1100/500000]
+ optim/total_grad_norm=8.492
+ train/ActionL1Loss=0.1488
+ throughput/total_tokens=211,200,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1101/500000]
+ train/ActionL1Loss=0.1638
+ throughput/total_tokens=211,392,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1102/500000]
+ train/ActionL1Loss=0.1508
+ throughput/total_tokens=211,584,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1103/500000]
+ train/ActionL1Loss=0.1441
+ throughput/total_tokens=211,776,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1104/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=211,968,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1105/500000]
+ train/ActionL1Loss=0.1574
+ throughput/total_tokens=212,160,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1106/500000]
+ train/ActionL1Loss=0.1525
+ throughput/total_tokens=212,352,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1107/500000]
+ train/ActionL1Loss=0.1423
+ throughput/total_tokens=212,544,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1108/500000]
+ train/ActionL1Loss=0.1139
+ throughput/total_tokens=212,736,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1109/500000]
+ train/ActionL1Loss=0.1586
+ throughput/total_tokens=212,928,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1110/500000]
+ train/ActionL1Loss=0.1805
+ throughput/total_tokens=213,120,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1111/500000]
+ train/ActionL1Loss=0.1477
+ throughput/total_tokens=213,312,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1112/500000]
+ train/ActionL1Loss=0.1570
+ throughput/total_tokens=213,504,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1113/500000]
+ train/ActionL1Loss=0.1554
+ throughput/total_tokens=213,696,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1114/500000]
+ train/ActionL1Loss=0.1416
+ throughput/total_tokens=213,888,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1115/500000]
+ train/ActionL1Loss=0.1719
+ throughput/total_tokens=214,080,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1116/500000]
+ train/ActionL1Loss=0.1594
+ throughput/total_tokens=214,272,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1117/500000]
+ train/ActionL1Loss=0.1447
+ throughput/total_tokens=214,464,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1118/500000]
+ train/ActionL1Loss=0.1488
+ throughput/total_tokens=214,656,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1119/500000]
+ train/ActionL1Loss=0.1463
+ throughput/total_tokens=214,848,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1120/500000]
+ optim/total_grad_norm=7.752
+ train/ActionL1Loss=0.1292
+ throughput/total_tokens=215,040,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1121/500000]
+ train/ActionL1Loss=0.1561
+ throughput/total_tokens=215,232,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1122/500000]
+ train/ActionL1Loss=0.1302
+ throughput/total_tokens=215,424,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1123/500000]
+ train/ActionL1Loss=0.1409
+ throughput/total_tokens=215,616,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1124/500000]
+ train/ActionL1Loss=0.1380
+ throughput/total_tokens=215,808,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1125/500000]
+ train/ActionL1Loss=0.1431
+ throughput/total_tokens=216,000,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1126/500000]
+ train/ActionL1Loss=0.1293
+ throughput/total_tokens=216,192,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1127/500000]
+ train/ActionL1Loss=0.1570
+ throughput/total_tokens=216,384,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1128/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=216,576,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1129/500000]
+ train/ActionL1Loss=0.1595
+ throughput/total_tokens=216,768,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1130/500000]
+ train/ActionL1Loss=0.1438
+ throughput/total_tokens=216,960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1131/500000]
+ train/ActionL1Loss=0.1298
+ throughput/total_tokens=217,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1132/500000]
+ train/ActionL1Loss=0.1372
+ throughput/total_tokens=217,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1133/500000]
+ train/ActionL1Loss=0.1329
+ throughput/total_tokens=217,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1134/500000]
+ train/ActionL1Loss=0.1549
+ throughput/total_tokens=217,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1135/500000]
+ train/ActionL1Loss=0.1307
+ throughput/total_tokens=217,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1136/500000]
+ train/ActionL1Loss=0.1560
+ throughput/total_tokens=218,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1137/500000]
+ train/ActionL1Loss=0.1163
+ throughput/total_tokens=218,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1138/500000]
+ train/ActionL1Loss=0.1054
+ throughput/total_tokens=218,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1139/500000]
+ train/ActionL1Loss=0.1550
+ throughput/total_tokens=218,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1140/500000]
+ optim/total_grad_norm=8.216
+ train/ActionL1Loss=0.1487
+ throughput/total_tokens=218,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1141/500000]
+ train/ActionL1Loss=0.1262
+ throughput/total_tokens=219,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1142/500000]
+ train/ActionL1Loss=0.1087
+ throughput/total_tokens=219,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1143/500000]
+ train/ActionL1Loss=0.1503
+ throughput/total_tokens=219,456,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1144/500000]
+ train/ActionL1Loss=0.1453
+ throughput/total_tokens=219,648,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1145/500000]
+ train/ActionL1Loss=0.1532
+ throughput/total_tokens=219,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1146/500000]
+ train/ActionL1Loss=0.1213
+ throughput/total_tokens=220,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1147/500000]
+ train/ActionL1Loss=0.1415
+ throughput/total_tokens=220,224,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1148/500000]
+ train/ActionL1Loss=0.1548
+ throughput/total_tokens=220,416,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1149/500000]
+ train/ActionL1Loss=0.1579
+ throughput/total_tokens=220,608,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1150/500000]
+ train/ActionL1Loss=0.1539
+ throughput/total_tokens=220,800,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1151/500000]
+ train/ActionL1Loss=0.1339
+ throughput/total_tokens=220,992,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1152/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=221,184,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1153/500000]
+ train/ActionL1Loss=0.1384
+ throughput/total_tokens=221,376,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1154/500000]
+ train/ActionL1Loss=0.1318
+ throughput/total_tokens=221,568,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1155/500000]
+ train/ActionL1Loss=0.1365
+ throughput/total_tokens=221,760,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1156/500000]
+ train/ActionL1Loss=0.1225
+ throughput/total_tokens=221,952,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1157/500000]
+ train/ActionL1Loss=0.1310
+ throughput/total_tokens=222,144,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1158/500000]
+ train/ActionL1Loss=0.1213
+ throughput/total_tokens=222,336,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1159/500000]
+ train/ActionL1Loss=0.1269
+ throughput/total_tokens=222,528,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1160/500000]
+ optim/total_grad_norm=5.809
+ train/ActionL1Loss=0.1305
+ throughput/total_tokens=222,720,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=1161/500000]
+ train/ActionL1Loss=0.1192
+ throughput/total_tokens=222,912,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1162/500000]
+ train/ActionL1Loss=0.1218
+ throughput/total_tokens=223,104,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1163/500000]
+ train/ActionL1Loss=0.1256
+ throughput/total_tokens=223,296,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1164/500000]
+ train/ActionL1Loss=0.1264
+ throughput/total_tokens=223,488,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1165/500000]
+ train/ActionL1Loss=0.1271
+ throughput/total_tokens=223,680,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1166/500000]
+ train/ActionL1Loss=0.1281
+ throughput/total_tokens=223,872,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1167/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=224,064,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1168/500000]
+ train/ActionL1Loss=0.1420
+ throughput/total_tokens=224,256,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1169/500000]
+ train/ActionL1Loss=0.1473
+ throughput/total_tokens=224,448,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1170/500000]
+ train/ActionL1Loss=0.1551
+ throughput/total_tokens=224,640,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1171/500000]
+ train/ActionL1Loss=0.1700
+ throughput/total_tokens=224,832,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1172/500000]
+ train/ActionL1Loss=0.1373
+ throughput/total_tokens=225,024,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1173/500000]
+ train/ActionL1Loss=0.1642
+ throughput/total_tokens=225,216,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1174/500000]
+ train/ActionL1Loss=0.1725
+ throughput/total_tokens=225,408,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1175/500000]
+ train/ActionL1Loss=0.1602
+ throughput/total_tokens=225,600,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1176/500000]
+ train/ActionL1Loss=0.1536
+ throughput/total_tokens=225,792,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1177/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=225,984,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1178/500000]
+ train/ActionL1Loss=0.1451
+ throughput/total_tokens=226,176,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1179/500000]
+ train/ActionL1Loss=0.1668
+ throughput/total_tokens=226,368,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1180/500000]
+ optim/total_grad_norm=7.688
+ train/ActionL1Loss=0.1467
+ throughput/total_tokens=226,560,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1181/500000]
+ train/ActionL1Loss=0.1458
+ throughput/total_tokens=226,752,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1182/500000]
+ train/ActionL1Loss=0.1621
+ throughput/total_tokens=226,944,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1183/500000]
+ train/ActionL1Loss=0.1484
+ throughput/total_tokens=227,136,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1184/500000]
+ train/ActionL1Loss=0.1393
+ throughput/total_tokens=227,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1185/500000]
+ train/ActionL1Loss=0.1420
+ throughput/total_tokens=227,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1186/500000]
+ train/ActionL1Loss=0.1262
+ throughput/total_tokens=227,712,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1187/500000]
+ train/ActionL1Loss=0.1204
+ throughput/total_tokens=227,904,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1188/500000]
+ train/ActionL1Loss=0.1081
+ throughput/total_tokens=228,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1189/500000]
+ train/ActionL1Loss=0.1262
+ throughput/total_tokens=228,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1190/500000]
+ train/ActionL1Loss=0.1213
+ throughput/total_tokens=228,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1191/500000]
+ train/ActionL1Loss=0.1316
+ throughput/total_tokens=228,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1192/500000]
+ train/ActionL1Loss=0.1308
+ throughput/total_tokens=228,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1193/500000]
+ train/ActionL1Loss=0.1589
+ throughput/total_tokens=229,056,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1194/500000]
+ train/ActionL1Loss=0.1537
+ throughput/total_tokens=229,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1195/500000]
+ train/ActionL1Loss=0.1541
+ throughput/total_tokens=229,440,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1196/500000]
+ train/ActionL1Loss=0.1420
+ throughput/total_tokens=229,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1197/500000]
+ train/ActionL1Loss=0.1552
+ throughput/total_tokens=229,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1198/500000]
+ train/ActionL1Loss=0.1492
+ throughput/total_tokens=230,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1199/500000]
+ train/ActionL1Loss=0.1464
+ throughput/total_tokens=230,208,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1200/500000]
+ optim/total_grad_norm=7.543
+ train/ActionL1Loss=0.1461
+ throughput/total_tokens=230,400,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1201/500000]
+ train/ActionL1Loss=0.1505
+ throughput/total_tokens=230,592,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1202/500000]
+ train/ActionL1Loss=0.1520
+ throughput/total_tokens=230,784,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1203/500000]
+ train/ActionL1Loss=0.1386
+ throughput/total_tokens=230,976,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1204/500000]
+ train/ActionL1Loss=0.1367
+ throughput/total_tokens=231,168,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1205/500000]
+ train/ActionL1Loss=0.1413
+ throughput/total_tokens=231,360,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1206/500000]
+ train/ActionL1Loss=0.1485
+ throughput/total_tokens=231,552,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1207/500000]
+ train/ActionL1Loss=0.1570
+ throughput/total_tokens=231,744,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1208/500000]
+ train/ActionL1Loss=0.1298
+ throughput/total_tokens=231,936,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1209/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=232,128,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1210/500000]
+ train/ActionL1Loss=0.1354
+ throughput/total_tokens=232,320,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=1211/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=232,512,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1212/500000]
+ train/ActionL1Loss=0.1693
+ throughput/total_tokens=232,704,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1213/500000]
+ train/ActionL1Loss=0.1594
+ throughput/total_tokens=232,896,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1214/500000]
+ train/ActionL1Loss=0.1445
+ throughput/total_tokens=233,088,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1215/500000]
+ train/ActionL1Loss=0.1590
+ throughput/total_tokens=233,280,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1216/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=233,472,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1217/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=233,664,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1218/500000]
+ train/ActionL1Loss=0.1079
+ throughput/total_tokens=233,856,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1219/500000]
+ train/ActionL1Loss=0.1408
+ throughput/total_tokens=234,048,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1220/500000]
+ optim/total_grad_norm=9.819
+ train/ActionL1Loss=0.1333
+ throughput/total_tokens=234,240,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1221/500000]
+ train/ActionL1Loss=0.1422
+ throughput/total_tokens=234,432,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1222/500000]
+ train/ActionL1Loss=0.1601
+ throughput/total_tokens=234,624,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1223/500000]
+ train/ActionL1Loss=0.1318
+ throughput/total_tokens=234,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1224/500000]
+ train/ActionL1Loss=0.1388
+ throughput/total_tokens=235,008,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1225/500000]
+ train/ActionL1Loss=0.1417
+ throughput/total_tokens=235,200,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1226/500000]
+ train/ActionL1Loss=0.1340
+ throughput/total_tokens=235,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1227/500000]
+ train/ActionL1Loss=0.1388
+ throughput/total_tokens=235,584,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1228/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=235,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1229/500000]
+ train/ActionL1Loss=0.1389
+ throughput/total_tokens=235,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1230/500000]
+ train/ActionL1Loss=0.1448
+ throughput/total_tokens=236,160,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1231/500000]
+ train/ActionL1Loss=0.1559
+ throughput/total_tokens=236,352,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1232/500000]
+ train/ActionL1Loss=0.1377
+ throughput/total_tokens=236,544,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1233/500000]
+ train/ActionL1Loss=0.1346
+ throughput/total_tokens=236,736,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1234/500000]
+ train/ActionL1Loss=0.1393
+ throughput/total_tokens=236,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1235/500000]
+ train/ActionL1Loss=0.1631
+ throughput/total_tokens=237,120,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1236/500000]
+ train/ActionL1Loss=0.1488
+ throughput/total_tokens=237,312,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1237/500000]
+ train/ActionL1Loss=0.1288
+ throughput/total_tokens=237,504,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1238/500000]
+ train/ActionL1Loss=0.1342
+ throughput/total_tokens=237,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1239/500000]
+ train/ActionL1Loss=0.1295
+ throughput/total_tokens=237,888,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1240/500000]
+ optim/total_grad_norm=7.144
+ train/ActionL1Loss=0.1241
+ throughput/total_tokens=238,080,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1241/500000]
+ train/ActionL1Loss=0.1370
+ throughput/total_tokens=238,272,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1242/500000]
+ train/ActionL1Loss=0.1356
+ throughput/total_tokens=238,464,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1243/500000]
+ train/ActionL1Loss=0.1256
+ throughput/total_tokens=238,656,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1244/500000]
+ train/ActionL1Loss=0.1181
+ throughput/total_tokens=238,848,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1245/500000]
+ train/ActionL1Loss=0.1372
+ throughput/total_tokens=239,040,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1246/500000]
+ train/ActionL1Loss=0.1440
+ throughput/total_tokens=239,232,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1247/500000]
+ train/ActionL1Loss=0.1233
+ throughput/total_tokens=239,424,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1248/500000]
+ train/ActionL1Loss=0.1295
+ throughput/total_tokens=239,616,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1249/500000]
+ train/ActionL1Loss=0.1121
+ throughput/total_tokens=239,808,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1250/500000]
+ train/ActionL1Loss=0.1231
+ throughput/total_tokens=240,000,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1251/500000]
+ train/ActionL1Loss=0.1248
+ throughput/total_tokens=240,192,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1252/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=240,384,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1253/500000]
+ train/ActionL1Loss=0.1307
+ throughput/total_tokens=240,576,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1254/500000]
+ train/ActionL1Loss=0.1256
+ throughput/total_tokens=240,768,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1255/500000]
+ train/ActionL1Loss=0.1470
+ throughput/total_tokens=240,960,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1256/500000]
+ train/ActionL1Loss=0.1349
+ throughput/total_tokens=241,152,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1257/500000]
+ train/ActionL1Loss=0.1329
+ throughput/total_tokens=241,344,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1258/500000]
+ train/ActionL1Loss=0.1419
+ throughput/total_tokens=241,536,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1259/500000]
+ train/ActionL1Loss=0.1498
+ throughput/total_tokens=241,728,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1260/500000]
+ optim/total_grad_norm=6.370
+ train/ActionL1Loss=0.1325
+ throughput/total_tokens=241,920,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1261/500000]
+ train/ActionL1Loss=0.1290
+ throughput/total_tokens=242,112,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1262/500000]
+ train/ActionL1Loss=0.1184
+ throughput/total_tokens=242,304,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1263/500000]
+ train/ActionL1Loss=0.1175
+ throughput/total_tokens=242,496,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1264/500000]
+ train/ActionL1Loss=0.1442
+ throughput/total_tokens=242,688,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1265/500000]
+ train/ActionL1Loss=0.1388
+ throughput/total_tokens=242,880,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1266/500000]
+ train/ActionL1Loss=0.1236
+ throughput/total_tokens=243,072,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1267/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=243,264,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1268/500000]
+ train/ActionL1Loss=0.1291
+ throughput/total_tokens=243,456,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1269/500000]
+ train/ActionL1Loss=0.1389
+ throughput/total_tokens=243,648,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1270/500000]
+ train/ActionL1Loss=0.1507
+ throughput/total_tokens=243,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1271/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=244,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1272/500000]
+ train/ActionL1Loss=0.1334
+ throughput/total_tokens=244,224,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1273/500000]
+ train/ActionL1Loss=0.1335
+ throughput/total_tokens=244,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1274/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=244,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1275/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=244,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1276/500000]
+ train/ActionL1Loss=0.1335
+ throughput/total_tokens=244,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1277/500000]
+ train/ActionL1Loss=0.1279
+ throughput/total_tokens=245,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1278/500000]
+ train/ActionL1Loss=0.1314
+ throughput/total_tokens=245,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1279/500000]
+ train/ActionL1Loss=0.1412
+ throughput/total_tokens=245,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1280/500000]
+ optim/total_grad_norm=6.727
+ train/ActionL1Loss=0.1450
+ throughput/total_tokens=245,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1281/500000]
+ train/ActionL1Loss=0.1652
+ throughput/total_tokens=245,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1282/500000]
+ train/ActionL1Loss=0.1230
+ throughput/total_tokens=246,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1283/500000]
+ train/ActionL1Loss=0.1575
+ throughput/total_tokens=246,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1284/500000]
+ train/ActionL1Loss=0.1443
+ throughput/total_tokens=246,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1285/500000]
+ train/ActionL1Loss=0.1574
+ throughput/total_tokens=246,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1286/500000]
+ train/ActionL1Loss=0.1436
+ throughput/total_tokens=246,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1287/500000]
+ train/ActionL1Loss=0.1505
+ throughput/total_tokens=247,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1288/500000]
+ train/ActionL1Loss=0.1489
+ throughput/total_tokens=247,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1289/500000]
+ train/ActionL1Loss=0.1390
+ throughput/total_tokens=247,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1290/500000]
+ train/ActionL1Loss=0.1546
+ throughput/total_tokens=247,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1291/500000]
+ train/ActionL1Loss=0.1592
+ throughput/total_tokens=247,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1292/500000]
+ train/ActionL1Loss=0.1238
+ throughput/total_tokens=248,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1293/500000]
+ train/ActionL1Loss=0.1311
+ throughput/total_tokens=248,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1294/500000]
+ train/ActionL1Loss=0.1376
+ throughput/total_tokens=248,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1295/500000]
+ train/ActionL1Loss=0.1221
+ throughput/total_tokens=248,640,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1296/500000]
+ train/ActionL1Loss=0.1473
+ throughput/total_tokens=248,832,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1297/500000]
+ train/ActionL1Loss=0.1518
+ throughput/total_tokens=249,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1298/500000]
+ train/ActionL1Loss=0.1385
+ throughput/total_tokens=249,216,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1299/500000]
+ train/ActionL1Loss=0.1161
+ throughput/total_tokens=249,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1300/500000]
+ optim/total_grad_norm=5.362
+ train/ActionL1Loss=0.1453
+ throughput/total_tokens=249,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1301/500000]
+ train/ActionL1Loss=0.1219
+ throughput/total_tokens=249,792,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1302/500000]
+ train/ActionL1Loss=0.1241
+ throughput/total_tokens=249,984,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1303/500000]
+ train/ActionL1Loss=0.1376
+ throughput/total_tokens=250,176,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1304/500000]
+ train/ActionL1Loss=0.1133
+ throughput/total_tokens=250,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1305/500000]
+ train/ActionL1Loss=0.1359
+ throughput/total_tokens=250,560,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1306/500000]
+ train/ActionL1Loss=0.1425
+ throughput/total_tokens=250,752,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1307/500000]
+ train/ActionL1Loss=0.1427
+ throughput/total_tokens=250,944,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1308/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=251,136,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1309/500000]
+ train/ActionL1Loss=0.1393
+ throughput/total_tokens=251,328,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1310/500000]
+ train/ActionL1Loss=0.1246
+ throughput/total_tokens=251,520,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1311/500000]
+ train/ActionL1Loss=0.1249
+ throughput/total_tokens=251,712,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1312/500000]
+ train/ActionL1Loss=0.1281
+ throughput/total_tokens=251,904,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1313/500000]
+ train/ActionL1Loss=0.1425
+ throughput/total_tokens=252,096,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1314/500000]
+ train/ActionL1Loss=0.1314
+ throughput/total_tokens=252,288,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1315/500000]
+ train/ActionL1Loss=0.1308
+ throughput/total_tokens=252,480,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1316/500000]
+ train/ActionL1Loss=0.1361
+ throughput/total_tokens=252,672,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1317/500000]
+ train/ActionL1Loss=0.1119
+ throughput/total_tokens=252,864,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1318/500000]
+ train/ActionL1Loss=0.1155
+ throughput/total_tokens=253,056,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1319/500000]
+ train/ActionL1Loss=0.1346
+ throughput/total_tokens=253,248,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1320/500000]
+ optim/total_grad_norm=7.233
+ train/ActionL1Loss=0.1375
+ throughput/total_tokens=253,440,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1321/500000]
+ train/ActionL1Loss=0.1251
+ throughput/total_tokens=253,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1322/500000]
+ train/ActionL1Loss=0.1199
+ throughput/total_tokens=253,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1323/500000]
+ train/ActionL1Loss=0.1376
+ throughput/total_tokens=254,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1324/500000]
+ train/ActionL1Loss=0.1478
+ throughput/total_tokens=254,208,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1325/500000]
+ train/ActionL1Loss=0.1874
+ throughput/total_tokens=254,400,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1326/500000]
+ train/ActionL1Loss=0.1538
+ throughput/total_tokens=254,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1327/500000]
+ train/ActionL1Loss=0.1445
+ throughput/total_tokens=254,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1328/500000]
+ train/ActionL1Loss=0.1431
+ throughput/total_tokens=254,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1329/500000]
+ train/ActionL1Loss=0.1451
+ throughput/total_tokens=255,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1330/500000]
+ train/ActionL1Loss=0.1582
+ throughput/total_tokens=255,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1331/500000]
+ train/ActionL1Loss=0.1758
+ throughput/total_tokens=255,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1332/500000]
+ train/ActionL1Loss=0.1479
+ throughput/total_tokens=255,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1333/500000]
+ train/ActionL1Loss=0.1345
+ throughput/total_tokens=255,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1334/500000]
+ train/ActionL1Loss=0.1419
+ throughput/total_tokens=256,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1335/500000]
+ train/ActionL1Loss=0.1356
+ throughput/total_tokens=256,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1336/500000]
+ train/ActionL1Loss=0.1475
+ throughput/total_tokens=256,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1337/500000]
+ train/ActionL1Loss=0.1760
+ throughput/total_tokens=256,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1338/500000]
+ train/ActionL1Loss=0.1463
+ throughput/total_tokens=256,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1339/500000]
+ train/ActionL1Loss=0.1332
+ throughput/total_tokens=257,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1340/500000]
+ optim/total_grad_norm=7.615
+ train/ActionL1Loss=0.1370
+ throughput/total_tokens=257,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1341/500000]
+ train/ActionL1Loss=0.1413
+ throughput/total_tokens=257,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1342/500000]
+ train/ActionL1Loss=0.1650
+ throughput/total_tokens=257,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1343/500000]
+ train/ActionL1Loss=0.1446
+ throughput/total_tokens=257,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1344/500000]
+ train/ActionL1Loss=0.1375
+ throughput/total_tokens=258,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1345/500000]
+ train/ActionL1Loss=0.1129
+ throughput/total_tokens=258,240,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1346/500000]
+ train/ActionL1Loss=0.1465
+ throughput/total_tokens=258,432,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1347/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=258,624,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1348/500000]
+ train/ActionL1Loss=0.1335
+ throughput/total_tokens=258,816,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1349/500000]
+ train/ActionL1Loss=0.1213
+ throughput/total_tokens=259,008,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1350/500000]
+ train/ActionL1Loss=0.1201
+ throughput/total_tokens=259,200,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1351/500000]
+ train/ActionL1Loss=0.1236
+ throughput/total_tokens=259,392,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1352/500000]
+ train/ActionL1Loss=0.1600
+ throughput/total_tokens=259,584,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1353/500000]
+ train/ActionL1Loss=0.1190
+ throughput/total_tokens=259,776,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1354/500000]
+ train/ActionL1Loss=0.1382
+ throughput/total_tokens=259,968,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1355/500000]
+ train/ActionL1Loss=0.1104
+ throughput/total_tokens=260,160,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1356/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=260,352,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1357/500000]
+ train/ActionL1Loss=0.1223
+ throughput/total_tokens=260,544,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1358/500000]
+ train/ActionL1Loss=0.1110
+ throughput/total_tokens=260,736,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1359/500000]
+ train/ActionL1Loss=0.1124
+ throughput/total_tokens=260,928,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1360/500000]
+ optim/total_grad_norm=6.653
+ train/ActionL1Loss=0.1322
+ throughput/total_tokens=261,120,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1361/500000]
+ train/ActionL1Loss=0.1203
+ throughput/total_tokens=261,312,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1362/500000]
+ train/ActionL1Loss=0.1239
+ throughput/total_tokens=261,504,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1363/500000]
+ train/ActionL1Loss=0.1267
+ throughput/total_tokens=261,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1364/500000]
+ train/ActionL1Loss=0.1272
+ throughput/total_tokens=261,888,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1365/500000]
+ train/ActionL1Loss=0.1025
+ throughput/total_tokens=262,080,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1366/500000]
+ train/ActionL1Loss=0.1544
+ throughput/total_tokens=262,272,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1367/500000]
+ train/ActionL1Loss=0.1352
+ throughput/total_tokens=262,464,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1368/500000]
+ train/ActionL1Loss=0.1276
+ throughput/total_tokens=262,656,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1369/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=262,848,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1370/500000]
+ train/ActionL1Loss=0.1242
+ throughput/total_tokens=263,040,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1371/500000]
+ train/ActionL1Loss=0.1276
+ throughput/total_tokens=263,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1372/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=263,424,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1373/500000]
+ train/ActionL1Loss=0.1293
+ throughput/total_tokens=263,616,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1374/500000]
+ train/ActionL1Loss=0.1309
+ throughput/total_tokens=263,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1375/500000]
+ train/ActionL1Loss=0.1538
+ throughput/total_tokens=264,000,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1376/500000]
+ train/ActionL1Loss=0.1135
+ throughput/total_tokens=264,192,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1377/500000]
+ train/ActionL1Loss=0.1442
+ throughput/total_tokens=264,384,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1378/500000]
+ train/ActionL1Loss=0.1341
+ throughput/total_tokens=264,576,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1379/500000]
+ train/ActionL1Loss=0.1256
+ throughput/total_tokens=264,768,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1380/500000]
+ optim/total_grad_norm=7.195
+ train/ActionL1Loss=0.1380
+ throughput/total_tokens=264,960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1381/500000]
+ train/ActionL1Loss=0.1316
+ throughput/total_tokens=265,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1382/500000]
+ train/ActionL1Loss=0.1433
+ throughput/total_tokens=265,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1383/500000]
+ train/ActionL1Loss=0.1162
+ throughput/total_tokens=265,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1384/500000]
+ train/ActionL1Loss=0.1250
+ throughput/total_tokens=265,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1385/500000]
+ train/ActionL1Loss=0.1152
+ throughput/total_tokens=265,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1386/500000]
+ train/ActionL1Loss=0.1170
+ throughput/total_tokens=266,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1387/500000]
+ train/ActionL1Loss=0.1304
+ throughput/total_tokens=266,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1388/500000]
+ train/ActionL1Loss=0.1353
+ throughput/total_tokens=266,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1389/500000]
+ train/ActionL1Loss=0.1452
+ throughput/total_tokens=266,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1390/500000]
+ train/ActionL1Loss=0.1180
+ throughput/total_tokens=266,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1391/500000]
+ train/ActionL1Loss=0.1324
+ throughput/total_tokens=267,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1392/500000]
+ train/ActionL1Loss=0.1401
+ throughput/total_tokens=267,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1393/500000]
+ train/ActionL1Loss=0.1613
+ throughput/total_tokens=267,456,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1394/500000]
+ train/ActionL1Loss=0.1478
+ throughput/total_tokens=267,648,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1395/500000]
+ train/ActionL1Loss=0.1490
+ throughput/total_tokens=267,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1396/500000]
+ train/ActionL1Loss=0.1517
+ throughput/total_tokens=268,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1397/500000]
+ train/ActionL1Loss=0.1322
+ throughput/total_tokens=268,224,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1398/500000]
+ train/ActionL1Loss=0.1342
+ throughput/total_tokens=268,416,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1399/500000]
+ train/ActionL1Loss=0.1569
+ throughput/total_tokens=268,608,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1400/500000]
+ optim/total_grad_norm=8.248
+ train/ActionL1Loss=0.1506
+ throughput/total_tokens=268,800,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1401/500000]
+ train/ActionL1Loss=0.1506
+ throughput/total_tokens=268,992,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1402/500000]
+ train/ActionL1Loss=0.1223
+ throughput/total_tokens=269,184,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1403/500000]
+ train/ActionL1Loss=0.1229
+ throughput/total_tokens=269,376,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1404/500000]
+ train/ActionL1Loss=0.1155
+ throughput/total_tokens=269,568,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1405/500000]
+ train/ActionL1Loss=0.1210
+ throughput/total_tokens=269,760,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1406/500000]
+ train/ActionL1Loss=0.1142
+ throughput/total_tokens=269,952,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1407/500000]
+ train/ActionL1Loss=0.1230
+ throughput/total_tokens=270,144,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1408/500000]
+ train/ActionL1Loss=0.1266
+ throughput/total_tokens=270,336,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1409/500000]
+ train/ActionL1Loss=0.1407
+ throughput/total_tokens=270,528,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1410/500000]
+ train/ActionL1Loss=0.1410
+ throughput/total_tokens=270,720,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1411/500000]
+ train/ActionL1Loss=0.1475
+ throughput/total_tokens=270,912,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1412/500000]
+ train/ActionL1Loss=0.1205
+ throughput/total_tokens=271,104,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1413/500000]
+ train/ActionL1Loss=0.1080
+ throughput/total_tokens=271,296,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1414/500000]
+ train/ActionL1Loss=0.1421
+ throughput/total_tokens=271,488,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1415/500000]
+ train/ActionL1Loss=0.1344
+ throughput/total_tokens=271,680,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1416/500000]
+ train/ActionL1Loss=0.1256
+ throughput/total_tokens=271,872,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1417/500000]
+ train/ActionL1Loss=0.1341
+ throughput/total_tokens=272,064,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1418/500000]
+ train/ActionL1Loss=0.1571
+ throughput/total_tokens=272,256,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1419/500000]
+ train/ActionL1Loss=0.1310
+ throughput/total_tokens=272,448,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1420/500000]
+ optim/total_grad_norm=8.794
+ train/ActionL1Loss=0.1353
+ throughput/total_tokens=272,640,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1421/500000]
+ train/ActionL1Loss=0.1305
+ throughput/total_tokens=272,832,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1422/500000]
+ train/ActionL1Loss=0.1373
+ throughput/total_tokens=273,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1423/500000]
+ train/ActionL1Loss=0.1286
+ throughput/total_tokens=273,216,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1424/500000]
+ train/ActionL1Loss=0.1239
+ throughput/total_tokens=273,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1425/500000]
+ train/ActionL1Loss=0.1555
+ throughput/total_tokens=273,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1426/500000]
+ train/ActionL1Loss=0.1343
+ throughput/total_tokens=273,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1427/500000]
+ train/ActionL1Loss=0.1250
+ throughput/total_tokens=273,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1428/500000]
+ train/ActionL1Loss=0.1308
+ throughput/total_tokens=274,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1429/500000]
+ train/ActionL1Loss=0.1693
+ throughput/total_tokens=274,368,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1430/500000]
+ train/ActionL1Loss=0.1576
+ throughput/total_tokens=274,560,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1431/500000]
+ train/ActionL1Loss=0.1274
+ throughput/total_tokens=274,752,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1432/500000]
+ train/ActionL1Loss=0.1520
+ throughput/total_tokens=274,944,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1433/500000]
+ train/ActionL1Loss=0.1565
+ throughput/total_tokens=275,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1434/500000]
+ train/ActionL1Loss=0.1439
+ throughput/total_tokens=275,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1435/500000]
+ train/ActionL1Loss=0.1450
+ throughput/total_tokens=275,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1436/500000]
+ train/ActionL1Loss=0.1111
+ throughput/total_tokens=275,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1437/500000]
+ train/ActionL1Loss=0.1274
+ throughput/total_tokens=275,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1438/500000]
+ train/ActionL1Loss=0.1375
+ throughput/total_tokens=276,096,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1439/500000]
+ train/ActionL1Loss=0.1310
+ throughput/total_tokens=276,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1440/500000]
+ optim/total_grad_norm=7.740
+ train/ActionL1Loss=0.1322
+ throughput/total_tokens=276,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1441/500000]
+ train/ActionL1Loss=0.1491
+ throughput/total_tokens=276,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1442/500000]
+ train/ActionL1Loss=0.1312
+ throughput/total_tokens=276,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1443/500000]
+ train/ActionL1Loss=0.1204
+ throughput/total_tokens=277,056,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1444/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=277,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1445/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=277,440,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1446/500000]
+ train/ActionL1Loss=0.1351
+ throughput/total_tokens=277,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1447/500000]
+ train/ActionL1Loss=0.1472
+ throughput/total_tokens=277,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1448/500000]
+ train/ActionL1Loss=0.1221
+ throughput/total_tokens=278,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1449/500000]
+ train/ActionL1Loss=0.1403
+ throughput/total_tokens=278,208,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1450/500000]
+ train/ActionL1Loss=0.1352
+ throughput/total_tokens=278,400,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1451/500000]
+ train/ActionL1Loss=0.1163
+ throughput/total_tokens=278,592,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1452/500000]
+ train/ActionL1Loss=0.1503
+ throughput/total_tokens=278,784,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1453/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=278,976,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1454/500000]
+ train/ActionL1Loss=0.1385
+ throughput/total_tokens=279,168,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1455/500000]
+ train/ActionL1Loss=0.1417
+ throughput/total_tokens=279,360,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1456/500000]
+ train/ActionL1Loss=0.1407
+ throughput/total_tokens=279,552,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1457/500000]
+ train/ActionL1Loss=0.1180
+ throughput/total_tokens=279,744,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1458/500000]
+ train/ActionL1Loss=0.1304
+ throughput/total_tokens=279,936,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1459/500000]
+ train/ActionL1Loss=0.1085
+ throughput/total_tokens=280,128,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1460/500000]
+ optim/total_grad_norm=7.945
+ train/ActionL1Loss=0.1312
+ throughput/total_tokens=280,320,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=1461/500000]
+ train/ActionL1Loss=0.1311
+ throughput/total_tokens=280,512,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1462/500000]
+ train/ActionL1Loss=0.1316
+ throughput/total_tokens=280,704,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1463/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=280,896,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1464/500000]
+ train/ActionL1Loss=0.1173
+ throughput/total_tokens=281,088,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1465/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=281,280,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1466/500000]
+ train/ActionL1Loss=0.1223
+ throughput/total_tokens=281,472,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1467/500000]
+ train/ActionL1Loss=0.1157
+ throughput/total_tokens=281,664,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1468/500000]
+ train/ActionL1Loss=0.1444
+ throughput/total_tokens=281,856,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1469/500000]
+ train/ActionL1Loss=0.1195
+ throughput/total_tokens=282,048,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1470/500000]
+ train/ActionL1Loss=0.1511
+ throughput/total_tokens=282,240,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1471/500000]
+ train/ActionL1Loss=0.1294
+ throughput/total_tokens=282,432,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1472/500000]
+ train/ActionL1Loss=0.1206
+ throughput/total_tokens=282,624,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1473/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=282,816,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1474/500000]
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=283,008,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1475/500000]
+ train/ActionL1Loss=0.1174
+ throughput/total_tokens=283,200,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1476/500000]
+ train/ActionL1Loss=0.1187
+ throughput/total_tokens=283,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1477/500000]
+ train/ActionL1Loss=0.1548
+ throughput/total_tokens=283,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1478/500000]
+ train/ActionL1Loss=0.1351
+ throughput/total_tokens=283,776,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1479/500000]
+ train/ActionL1Loss=0.1338
+ throughput/total_tokens=283,968,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1480/500000]
+ optim/total_grad_norm=8.052
+ train/ActionL1Loss=0.1208
+ throughput/total_tokens=284,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1481/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=284,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1482/500000]
+ train/ActionL1Loss=0.1101
+ throughput/total_tokens=284,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1483/500000]
+ train/ActionL1Loss=0.1473
+ throughput/total_tokens=284,736,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1484/500000]
+ train/ActionL1Loss=0.1344
+ throughput/total_tokens=284,928,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1485/500000]
+ train/ActionL1Loss=0.1562
+ throughput/total_tokens=285,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1486/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=285,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1487/500000]
+ train/ActionL1Loss=0.1424
+ throughput/total_tokens=285,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1488/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=285,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1489/500000]
+ train/ActionL1Loss=0.1318
+ throughput/total_tokens=285,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1490/500000]
+ train/ActionL1Loss=0.1576
+ throughput/total_tokens=286,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1491/500000]
+ train/ActionL1Loss=0.1440
+ throughput/total_tokens=286,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1492/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=286,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1493/500000]
+ train/ActionL1Loss=0.1582
+ throughput/total_tokens=286,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1494/500000]
+ train/ActionL1Loss=0.1376
+ throughput/total_tokens=286,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1495/500000]
+ train/ActionL1Loss=0.1422
+ throughput/total_tokens=287,040,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1496/500000]
+ train/ActionL1Loss=0.1451
+ throughput/total_tokens=287,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1497/500000]
+ train/ActionL1Loss=0.1462
+ throughput/total_tokens=287,424,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1498/500000]
+ train/ActionL1Loss=0.1181
+ throughput/total_tokens=287,616,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1499/500000]
+ train/ActionL1Loss=0.1207
+ throughput/total_tokens=287,808,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1500/500000]
+ optim/total_grad_norm=7.636
+ train/ActionL1Loss=0.1383
+ throughput/total_tokens=288,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+09/30 [22:02:43] INFO | >> Saving config... checkpoint.py:608
+09/30 [22:03:22] INFO | >> Saving model state... checkpoint.py:796
+09/30 [22:04:29] INFO | >> Saving optim state... checkpoint.py:811
+09/30 [22:05:55] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=1501/500000]
+ train/ActionL1Loss=0.1275
+ throughput/total_tokens=288,192,000
+ throughput/device/tokens_per_second=1,202
+ throughput/device/batches_per_second=0.0501
+[step=1502/500000]
+ train/ActionL1Loss=0.1225
+ throughput/total_tokens=288,384,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=1503/500000]
+ train/ActionL1Loss=0.1112
+ throughput/total_tokens=288,576,000
+ throughput/device/tokens_per_second=1,137
+ throughput/device/batches_per_second=0.0474
+[step=1504/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=288,768,000
+ throughput/device/tokens_per_second=1,137
+ throughput/device/batches_per_second=0.0474
+[step=1505/500000]
+ train/ActionL1Loss=0.1246
+ throughput/total_tokens=288,960,000
+ throughput/device/tokens_per_second=1,142
+ throughput/device/batches_per_second=0.0476
+[step=1506/500000]
+ train/ActionL1Loss=0.1115
+ throughput/total_tokens=289,152,000
+ throughput/device/tokens_per_second=1,145
+ throughput/device/batches_per_second=0.0477
+[step=1507/500000]
+ train/ActionL1Loss=0.1246
+ throughput/total_tokens=289,344,000
+ throughput/device/tokens_per_second=1,147
+ throughput/device/batches_per_second=0.0478
+[step=1508/500000]
+ train/ActionL1Loss=0.1303
+ throughput/total_tokens=289,536,000
+ throughput/device/tokens_per_second=1,149
+ throughput/device/batches_per_second=0.0479
+[step=1509/500000]
+ train/ActionL1Loss=0.1425
+ throughput/total_tokens=289,728,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0480
+[step=1510/500000]
+ train/ActionL1Loss=0.1164
+ throughput/total_tokens=289,920,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+ System/Peak GPU Memory (MB)=46,917
+[step=1511/500000]
+ train/ActionL1Loss=0.1368
+ throughput/total_tokens=290,112,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=1512/500000]
+ train/ActionL1Loss=0.1081
+ throughput/total_tokens=290,304,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=1513/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=290,496,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=1514/500000]
+ train/ActionL1Loss=0.1435
+ throughput/total_tokens=290,688,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=1515/500000]
+ train/ActionL1Loss=0.1187
+ throughput/total_tokens=290,880,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=1516/500000]
+ train/ActionL1Loss=0.1186
+ throughput/total_tokens=291,072,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=1517/500000]
+ train/ActionL1Loss=0.1336
+ throughput/total_tokens=291,264,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=1518/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=291,456,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=1519/500000]
+ train/ActionL1Loss=0.1327
+ throughput/total_tokens=291,648,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=1520/500000]
+ optim/total_grad_norm=6.378
+ train/ActionL1Loss=0.1259
+ throughput/total_tokens=291,840,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+ System/Peak GPU Memory (MB)=46,917
+[step=1521/500000]
+ train/ActionL1Loss=0.1149
+ throughput/total_tokens=292,032,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=1522/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=292,224,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=1523/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=292,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1524/500000]
+ train/ActionL1Loss=0.1238
+ throughput/total_tokens=292,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1525/500000]
+ train/ActionL1Loss=0.1023
+ throughput/total_tokens=292,800,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1526/500000]
+ train/ActionL1Loss=0.1133
+ throughput/total_tokens=292,992,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1527/500000]
+ train/ActionL1Loss=0.1385
+ throughput/total_tokens=293,184,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1528/500000]
+ train/ActionL1Loss=0.1301
+ throughput/total_tokens=293,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1529/500000]
+ train/ActionL1Loss=0.1440
+ throughput/total_tokens=293,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1530/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=293,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1531/500000]
+ train/ActionL1Loss=0.1357
+ throughput/total_tokens=293,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1532/500000]
+ train/ActionL1Loss=0.1379
+ throughput/total_tokens=294,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1533/500000]
+ train/ActionL1Loss=0.1625
+ throughput/total_tokens=294,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1534/500000]
+ train/ActionL1Loss=0.1576
+ throughput/total_tokens=294,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1535/500000]
+ train/ActionL1Loss=0.1402
+ throughput/total_tokens=294,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1536/500000]
+ train/ActionL1Loss=0.1493
+ throughput/total_tokens=294,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1537/500000]
+ train/ActionL1Loss=0.1385
+ throughput/total_tokens=295,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1538/500000]
+ train/ActionL1Loss=0.1422
+ throughput/total_tokens=295,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1539/500000]
+ train/ActionL1Loss=0.1394
+ throughput/total_tokens=295,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1540/500000]
+ optim/total_grad_norm=7.708
+ train/ActionL1Loss=0.1152
+ throughput/total_tokens=295,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1541/500000]
+ train/ActionL1Loss=0.1606
+ throughput/total_tokens=295,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1542/500000]
+ train/ActionL1Loss=0.1394
+ throughput/total_tokens=296,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1543/500000]
+ train/ActionL1Loss=0.1365
+ throughput/total_tokens=296,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1544/500000]
+ train/ActionL1Loss=0.1441
+ throughput/total_tokens=296,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1545/500000]
+ train/ActionL1Loss=0.1407
+ throughput/total_tokens=296,640,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1546/500000]
+ train/ActionL1Loss=0.1312
+ throughput/total_tokens=296,832,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1547/500000]
+ train/ActionL1Loss=0.1302
+ throughput/total_tokens=297,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1548/500000]
+ train/ActionL1Loss=0.1332
+ throughput/total_tokens=297,216,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1549/500000]
+ train/ActionL1Loss=0.1054
+ throughput/total_tokens=297,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1550/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=297,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1551/500000]
+ train/ActionL1Loss=0.1185
+ throughput/total_tokens=297,792,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1552/500000]
+ train/ActionL1Loss=0.1445
+ throughput/total_tokens=297,984,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1553/500000]
+ train/ActionL1Loss=0.1412
+ throughput/total_tokens=298,176,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1554/500000]
+ train/ActionL1Loss=0.1318
+ throughput/total_tokens=298,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1555/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=298,560,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1556/500000]
+ train/ActionL1Loss=0.1449
+ throughput/total_tokens=298,752,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1557/500000]
+ train/ActionL1Loss=0.1372
+ throughput/total_tokens=298,944,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1558/500000]
+ train/ActionL1Loss=0.1174
+ throughput/total_tokens=299,136,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1559/500000]
+ train/ActionL1Loss=0.1312
+ throughput/total_tokens=299,328,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1560/500000]
+ optim/total_grad_norm=9.160
+ train/ActionL1Loss=0.1198
+ throughput/total_tokens=299,520,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1561/500000]
+ train/ActionL1Loss=0.1346
+ throughput/total_tokens=299,712,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1562/500000]
+ train/ActionL1Loss=0.1180
+ throughput/total_tokens=299,904,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1563/500000]
+ train/ActionL1Loss=0.1159
+ throughput/total_tokens=300,096,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1564/500000]
+ train/ActionL1Loss=0.1128
+ throughput/total_tokens=300,288,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1565/500000]
+ train/ActionL1Loss=0.1182
+ throughput/total_tokens=300,480,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1566/500000]
+ train/ActionL1Loss=0.1590
+ throughput/total_tokens=300,672,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1567/500000]
+ train/ActionL1Loss=0.1340
+ throughput/total_tokens=300,864,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1568/500000]
+ train/ActionL1Loss=0.1243
+ throughput/total_tokens=301,056,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1569/500000]
+ train/ActionL1Loss=0.1207
+ throughput/total_tokens=301,248,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1570/500000]
+ train/ActionL1Loss=0.1243
+ throughput/total_tokens=301,440,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1571/500000]
+ train/ActionL1Loss=0.1343
+ throughput/total_tokens=301,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1572/500000]
+ train/ActionL1Loss=0.1457
+ throughput/total_tokens=301,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1573/500000]
+ train/ActionL1Loss=0.1323
+ throughput/total_tokens=302,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1574/500000]
+ train/ActionL1Loss=0.1356
+ throughput/total_tokens=302,208,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1575/500000]
+ train/ActionL1Loss=0.1278
+ throughput/total_tokens=302,400,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1576/500000]
+ train/ActionL1Loss=0.1389
+ throughput/total_tokens=302,592,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1577/500000]
+ train/ActionL1Loss=0.1212
+ throughput/total_tokens=302,784,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1578/500000]
+ train/ActionL1Loss=0.1527
+ throughput/total_tokens=302,976,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1579/500000]
+ train/ActionL1Loss=0.1575
+ throughput/total_tokens=303,168,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1580/500000]
+ optim/total_grad_norm=7.087
+ train/ActionL1Loss=0.1428
+ throughput/total_tokens=303,360,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1581/500000]
+ train/ActionL1Loss=0.1164
+ throughput/total_tokens=303,552,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1582/500000]
+ train/ActionL1Loss=0.1548
+ throughput/total_tokens=303,744,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1583/500000]
+ train/ActionL1Loss=0.1475
+ throughput/total_tokens=303,936,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1584/500000]
+ train/ActionL1Loss=0.1291
+ throughput/total_tokens=304,128,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1585/500000]
+ train/ActionL1Loss=0.1186
+ throughput/total_tokens=304,320,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1586/500000]
+ train/ActionL1Loss=0.1250
+ throughput/total_tokens=304,512,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1587/500000]
+ train/ActionL1Loss=0.1141
+ throughput/total_tokens=304,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1588/500000]
+ train/ActionL1Loss=0.1434
+ throughput/total_tokens=304,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1589/500000]
+ train/ActionL1Loss=0.1331
+ throughput/total_tokens=305,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1590/500000]
+ train/ActionL1Loss=0.1447
+ throughput/total_tokens=305,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1591/500000]
+ train/ActionL1Loss=0.1083
+ throughput/total_tokens=305,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1592/500000]
+ train/ActionL1Loss=0.1345
+ throughput/total_tokens=305,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1593/500000]
+ train/ActionL1Loss=0.1257
+ throughput/total_tokens=305,856,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1594/500000]
+ train/ActionL1Loss=0.1289
+ throughput/total_tokens=306,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1595/500000]
+ train/ActionL1Loss=0.1292
+ throughput/total_tokens=306,240,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1596/500000]
+ train/ActionL1Loss=0.1333
+ throughput/total_tokens=306,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1597/500000]
+ train/ActionL1Loss=0.1198
+ throughput/total_tokens=306,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1598/500000]
+ train/ActionL1Loss=0.1436
+ throughput/total_tokens=306,816,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1599/500000]
+ train/ActionL1Loss=0.1276
+ throughput/total_tokens=307,008,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1600/500000]
+ optim/total_grad_norm=7.182
+ train/ActionL1Loss=0.1245
+ throughput/total_tokens=307,200,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1601/500000]
+ train/ActionL1Loss=0.1173
+ throughput/total_tokens=307,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1602/500000]
+ train/ActionL1Loss=0.1426
+ throughput/total_tokens=307,584,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1603/500000]
+ train/ActionL1Loss=0.1151
+ throughput/total_tokens=307,776,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1604/500000]
+ train/ActionL1Loss=0.1190
+ throughput/total_tokens=307,968,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1605/500000]
+ train/ActionL1Loss=0.1216
+ throughput/total_tokens=308,160,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1606/500000]
+ train/ActionL1Loss=0.1111
+ throughput/total_tokens=308,352,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1607/500000]
+ train/ActionL1Loss=0.1425
+ throughput/total_tokens=308,544,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1608/500000]
+ train/ActionL1Loss=0.1232
+ throughput/total_tokens=308,736,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1609/500000]
+ train/ActionL1Loss=0.1208
+ throughput/total_tokens=308,928,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1610/500000]
+ train/ActionL1Loss=0.1207
+ throughput/total_tokens=309,120,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1611/500000]
+ train/ActionL1Loss=0.1188
+ throughput/total_tokens=309,312,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1612/500000]
+ train/ActionL1Loss=0.1502
+ throughput/total_tokens=309,504,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1613/500000]
+ train/ActionL1Loss=0.1303
+ throughput/total_tokens=309,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1614/500000]
+ train/ActionL1Loss=0.1202
+ throughput/total_tokens=309,888,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1615/500000]
+ train/ActionL1Loss=0.0989
+ throughput/total_tokens=310,080,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1616/500000]
+ train/ActionL1Loss=0.1282
+ throughput/total_tokens=310,272,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1617/500000]
+ train/ActionL1Loss=0.1178
+ throughput/total_tokens=310,464,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1618/500000]
+ train/ActionL1Loss=0.1190
+ throughput/total_tokens=310,656,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1619/500000]
+ train/ActionL1Loss=0.1342
+ throughput/total_tokens=310,848,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1620/500000]
+ optim/total_grad_norm=8.399
+ train/ActionL1Loss=0.1457
+ throughput/total_tokens=311,040,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1621/500000]
+ train/ActionL1Loss=0.1322
+ throughput/total_tokens=311,232,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1622/500000]
+ train/ActionL1Loss=0.1429
+ throughput/total_tokens=311,424,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1623/500000]
+ train/ActionL1Loss=0.1247
+ throughput/total_tokens=311,616,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1624/500000]
+ train/ActionL1Loss=0.1304
+ throughput/total_tokens=311,808,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1625/500000]
+ train/ActionL1Loss=0.1297
+ throughput/total_tokens=312,000,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1626/500000]
+ train/ActionL1Loss=0.1488
+ throughput/total_tokens=312,192,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1627/500000]
+ train/ActionL1Loss=0.1551
+ throughput/total_tokens=312,384,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1628/500000]
+ train/ActionL1Loss=0.1441
+ throughput/total_tokens=312,576,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1629/500000]
+ train/ActionL1Loss=0.1423
+ throughput/total_tokens=312,768,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1630/500000]
+ train/ActionL1Loss=0.1343
+ throughput/total_tokens=312,960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1631/500000]
+ train/ActionL1Loss=0.1471
+ throughput/total_tokens=313,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1632/500000]
+ train/ActionL1Loss=0.1151
+ throughput/total_tokens=313,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1633/500000]
+ train/ActionL1Loss=0.1617
+ throughput/total_tokens=313,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1634/500000]
+ train/ActionL1Loss=0.1292
+ throughput/total_tokens=313,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1635/500000]
+ train/ActionL1Loss=0.1429
+ throughput/total_tokens=313,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1636/500000]
+ train/ActionL1Loss=0.1340
+ throughput/total_tokens=314,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1637/500000]
+ train/ActionL1Loss=0.1241
+ throughput/total_tokens=314,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1638/500000]
+ train/ActionL1Loss=0.1405
+ throughput/total_tokens=314,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1639/500000]
+ train/ActionL1Loss=0.1406
+ throughput/total_tokens=314,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1640/500000]
+ optim/total_grad_norm=7.664
+ train/ActionL1Loss=0.1350
+ throughput/total_tokens=314,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1641/500000]
+ train/ActionL1Loss=0.1454
+ throughput/total_tokens=315,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1642/500000]
+ train/ActionL1Loss=0.1199
+ throughput/total_tokens=315,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1643/500000]
+ train/ActionL1Loss=0.1347
+ throughput/total_tokens=315,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1644/500000]
+ train/ActionL1Loss=0.1134
+ throughput/total_tokens=315,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1645/500000]
+ train/ActionL1Loss=0.1228
+ throughput/total_tokens=315,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1646/500000]
+ train/ActionL1Loss=0.1242
+ throughput/total_tokens=316,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1647/500000]
+ train/ActionL1Loss=0.1443
+ throughput/total_tokens=316,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1648/500000]
+ train/ActionL1Loss=0.1228
+ throughput/total_tokens=316,416,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1649/500000]
+ train/ActionL1Loss=0.1387
+ throughput/total_tokens=316,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1650/500000]
+ train/ActionL1Loss=0.1187
+ throughput/total_tokens=316,800,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1651/500000]
+ train/ActionL1Loss=0.1230
+ throughput/total_tokens=316,992,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1652/500000]
+ train/ActionL1Loss=0.1342
+ throughput/total_tokens=317,184,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1653/500000]
+ train/ActionL1Loss=0.1217
+ throughput/total_tokens=317,376,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1654/500000]
+ train/ActionL1Loss=0.1194
+ throughput/total_tokens=317,568,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1655/500000]
+ train/ActionL1Loss=0.1397
+ throughput/total_tokens=317,760,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1656/500000]
+ train/ActionL1Loss=0.1094
+ throughput/total_tokens=317,952,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1657/500000]
+ train/ActionL1Loss=0.1173
+ throughput/total_tokens=318,144,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1658/500000]
+ train/ActionL1Loss=0.1162
+ throughput/total_tokens=318,336,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1659/500000]
+ train/ActionL1Loss=0.1228
+ throughput/total_tokens=318,528,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1660/500000]
+ optim/total_grad_norm=7.771
+ train/ActionL1Loss=0.1096
+ throughput/total_tokens=318,720,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1661/500000]
+ train/ActionL1Loss=0.1266
+ throughput/total_tokens=318,912,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1662/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=319,104,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1663/500000]
+ train/ActionL1Loss=0.1125
+ throughput/total_tokens=319,296,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1664/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=319,488,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1665/500000]
+ train/ActionL1Loss=0.1133
+ throughput/total_tokens=319,680,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1666/500000]
+ train/ActionL1Loss=0.1233
+ throughput/total_tokens=319,872,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1667/500000]
+ train/ActionL1Loss=0.1262
+ throughput/total_tokens=320,064,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1668/500000]
+ train/ActionL1Loss=0.1235
+ throughput/total_tokens=320,256,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1669/500000]
+ train/ActionL1Loss=0.1206
+ throughput/total_tokens=320,448,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1670/500000]
+ train/ActionL1Loss=0.1331
+ throughput/total_tokens=320,640,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1671/500000]
+ train/ActionL1Loss=0.1418
+ throughput/total_tokens=320,832,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1672/500000]
+ train/ActionL1Loss=0.1077
+ throughput/total_tokens=321,024,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1673/500000]
+ train/ActionL1Loss=0.1443
+ throughput/total_tokens=321,216,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1674/500000]
+ train/ActionL1Loss=0.1393
+ throughput/total_tokens=321,408,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1675/500000]
+ train/ActionL1Loss=0.1443
+ throughput/total_tokens=321,600,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1676/500000]
+ train/ActionL1Loss=0.1182
+ throughput/total_tokens=321,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1677/500000]
+ train/ActionL1Loss=0.1537
+ throughput/total_tokens=321,984,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1678/500000]
+ train/ActionL1Loss=0.1378
+ throughput/total_tokens=322,176,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1679/500000]
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=322,368,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1680/500000]
+ optim/total_grad_norm=6.686
+ train/ActionL1Loss=0.1159
+ throughput/total_tokens=322,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1681/500000]
+ train/ActionL1Loss=0.1433
+ throughput/total_tokens=322,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1682/500000]
+ train/ActionL1Loss=0.1179
+ throughput/total_tokens=322,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1683/500000]
+ train/ActionL1Loss=0.1247
+ throughput/total_tokens=323,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1684/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=323,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1685/500000]
+ train/ActionL1Loss=0.1365
+ throughput/total_tokens=323,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1686/500000]
+ train/ActionL1Loss=0.1274
+ throughput/total_tokens=323,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1687/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=323,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1688/500000]
+ train/ActionL1Loss=0.1273
+ throughput/total_tokens=324,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1689/500000]
+ train/ActionL1Loss=0.1295
+ throughput/total_tokens=324,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1690/500000]
+ train/ActionL1Loss=0.1281
+ throughput/total_tokens=324,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1691/500000]
+ train/ActionL1Loss=0.1116
+ throughput/total_tokens=324,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1692/500000]
+ train/ActionL1Loss=0.1566
+ throughput/total_tokens=324,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1693/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=325,056,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1694/500000]
+ train/ActionL1Loss=0.1264
+ throughput/total_tokens=325,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1695/500000]
+ train/ActionL1Loss=0.1124
+ throughput/total_tokens=325,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1696/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=325,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1697/500000]
+ train/ActionL1Loss=0.1233
+ throughput/total_tokens=325,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1698/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=326,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1699/500000]
+ train/ActionL1Loss=0.1218
+ throughput/total_tokens=326,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1700/500000]
+ optim/total_grad_norm=6.495
+ train/ActionL1Loss=0.1355
+ throughput/total_tokens=326,400,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1701/500000]
+ train/ActionL1Loss=0.1130
+ throughput/total_tokens=326,592,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1702/500000]
+ train/ActionL1Loss=0.1392
+ throughput/total_tokens=326,784,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1703/500000]
+ train/ActionL1Loss=0.1317
+ throughput/total_tokens=326,976,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1704/500000]
+ train/ActionL1Loss=0.1250
+ throughput/total_tokens=327,168,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1705/500000]
+ train/ActionL1Loss=0.1324
+ throughput/total_tokens=327,360,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1706/500000]
+ train/ActionL1Loss=0.1277
+ throughput/total_tokens=327,552,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1707/500000]
+ train/ActionL1Loss=0.1442
+ throughput/total_tokens=327,744,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1708/500000]
+ train/ActionL1Loss=0.1314
+ throughput/total_tokens=327,936,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1709/500000]
+ train/ActionL1Loss=0.1377
+ throughput/total_tokens=328,128,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1710/500000]
+ train/ActionL1Loss=0.1163
+ throughput/total_tokens=328,320,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1711/500000]
+ train/ActionL1Loss=0.1341
+ throughput/total_tokens=328,512,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1712/500000]
+ train/ActionL1Loss=0.1191
+ throughput/total_tokens=328,704,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1713/500000]
+ train/ActionL1Loss=0.1181
+ throughput/total_tokens=328,896,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1714/500000]
+ train/ActionL1Loss=0.1404
+ throughput/total_tokens=329,088,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1715/500000]
+ train/ActionL1Loss=0.1318
+ throughput/total_tokens=329,280,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1716/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=329,472,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1717/500000]
+ train/ActionL1Loss=0.1245
+ throughput/total_tokens=329,664,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1718/500000]
+ train/ActionL1Loss=0.1339
+ throughput/total_tokens=329,856,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1719/500000]
+ train/ActionL1Loss=0.1253
+ throughput/total_tokens=330,048,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1720/500000]
+ optim/total_grad_norm=6.059
+ train/ActionL1Loss=0.1084
+ throughput/total_tokens=330,240,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1721/500000]
+ train/ActionL1Loss=0.1316
+ throughput/total_tokens=330,432,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1722/500000]
+ train/ActionL1Loss=0.1212
+ throughput/total_tokens=330,624,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1723/500000]
+ train/ActionL1Loss=0.1053
+ throughput/total_tokens=330,816,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1724/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=331,008,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1725/500000]
+ train/ActionL1Loss=0.1203
+ throughput/total_tokens=331,200,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1726/500000]
+ train/ActionL1Loss=0.1227
+ throughput/total_tokens=331,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1727/500000]
+ train/ActionL1Loss=0.1202
+ throughput/total_tokens=331,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1728/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=331,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1729/500000]
+ train/ActionL1Loss=0.1231
+ throughput/total_tokens=331,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1730/500000]
+ train/ActionL1Loss=0.1167
+ throughput/total_tokens=332,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1731/500000]
+ train/ActionL1Loss=0.1076
+ throughput/total_tokens=332,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1732/500000]
+ train/ActionL1Loss=0.1055
+ throughput/total_tokens=332,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1733/500000]
+ train/ActionL1Loss=0.1218
+ throughput/total_tokens=332,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1734/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=332,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1735/500000]
+ train/ActionL1Loss=0.1349
+ throughput/total_tokens=333,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1736/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=333,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1737/500000]
+ train/ActionL1Loss=0.1332
+ throughput/total_tokens=333,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1738/500000]
+ train/ActionL1Loss=0.1352
+ throughput/total_tokens=333,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1739/500000]
+ train/ActionL1Loss=0.1285
+ throughput/total_tokens=333,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1740/500000]
+ optim/total_grad_norm=7.816
+ train/ActionL1Loss=0.1246
+ throughput/total_tokens=334,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1741/500000]
+ train/ActionL1Loss=0.1276
+ throughput/total_tokens=334,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1742/500000]
+ train/ActionL1Loss=0.1122
+ throughput/total_tokens=334,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1743/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=334,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1744/500000]
+ train/ActionL1Loss=0.1154
+ throughput/total_tokens=334,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1745/500000]
+ train/ActionL1Loss=0.1228
+ throughput/total_tokens=335,040,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1746/500000]
+ train/ActionL1Loss=0.1136
+ throughput/total_tokens=335,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1747/500000]
+ train/ActionL1Loss=0.1121
+ throughput/total_tokens=335,424,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1748/500000]
+ train/ActionL1Loss=0.1299
+ throughput/total_tokens=335,616,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1749/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=335,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1750/500000]
+ train/ActionL1Loss=0.1173
+ throughput/total_tokens=336,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1751/500000]
+ train/ActionL1Loss=0.1010
+ throughput/total_tokens=336,192,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1752/500000]
+ train/ActionL1Loss=0.1000
+ throughput/total_tokens=336,384,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1753/500000]
+ train/ActionL1Loss=0.1306
+ throughput/total_tokens=336,576,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1754/500000]
+ train/ActionL1Loss=0.1382
+ throughput/total_tokens=336,768,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1755/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=336,960,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1756/500000]
+ train/ActionL1Loss=0.1240
+ throughput/total_tokens=337,152,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1757/500000]
+ train/ActionL1Loss=0.1212
+ throughput/total_tokens=337,344,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1758/500000]
+ train/ActionL1Loss=0.1504
+ throughput/total_tokens=337,536,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1759/500000]
+ train/ActionL1Loss=0.1195
+ throughput/total_tokens=337,728,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1760/500000]
+ optim/total_grad_norm=8.062
+ train/ActionL1Loss=0.1374
+ throughput/total_tokens=337,920,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1761/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=338,112,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1762/500000]
+ train/ActionL1Loss=0.1278
+ throughput/total_tokens=338,304,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1763/500000]
+ train/ActionL1Loss=0.1236
+ throughput/total_tokens=338,496,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1764/500000]
+ train/ActionL1Loss=0.1278
+ throughput/total_tokens=338,688,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1765/500000]
+ train/ActionL1Loss=0.1362
+ throughput/total_tokens=338,880,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1766/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=339,072,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1767/500000]
+ train/ActionL1Loss=0.1157
+ throughput/total_tokens=339,264,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1768/500000]
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=339,456,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1769/500000]
+ train/ActionL1Loss=0.1346
+ throughput/total_tokens=339,648,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=1770/500000]
+ train/ActionL1Loss=0.1043
+ throughput/total_tokens=339,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1771/500000]
+ train/ActionL1Loss=0.1285
+ throughput/total_tokens=340,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1772/500000]
+ train/ActionL1Loss=0.1312
+ throughput/total_tokens=340,224,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1773/500000]
+ train/ActionL1Loss=0.1363
+ throughput/total_tokens=340,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1774/500000]
+ train/ActionL1Loss=0.1284
+ throughput/total_tokens=340,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1775/500000]
+ train/ActionL1Loss=0.1347
+ throughput/total_tokens=340,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1776/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=340,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1777/500000]
+ train/ActionL1Loss=0.1334
+ throughput/total_tokens=341,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1778/500000]
+ train/ActionL1Loss=0.1274
+ throughput/total_tokens=341,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1779/500000]
+ train/ActionL1Loss=0.1349
+ throughput/total_tokens=341,568,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1780/500000]
+ optim/total_grad_norm=7.074
+ train/ActionL1Loss=0.1169
+ throughput/total_tokens=341,760,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1781/500000]
+ train/ActionL1Loss=0.1491
+ throughput/total_tokens=341,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1782/500000]
+ train/ActionL1Loss=0.1368
+ throughput/total_tokens=342,144,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1783/500000]
+ train/ActionL1Loss=0.1208
+ throughput/total_tokens=342,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1784/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=342,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1785/500000]
+ train/ActionL1Loss=0.1262
+ throughput/total_tokens=342,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1786/500000]
+ train/ActionL1Loss=0.1297
+ throughput/total_tokens=342,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1787/500000]
+ train/ActionL1Loss=0.1158
+ throughput/total_tokens=343,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1788/500000]
+ train/ActionL1Loss=0.1335
+ throughput/total_tokens=343,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1789/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=343,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1790/500000]
+ train/ActionL1Loss=0.1350
+ throughput/total_tokens=343,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1791/500000]
+ train/ActionL1Loss=0.1411
+ throughput/total_tokens=343,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1792/500000]
+ train/ActionL1Loss=0.1310
+ throughput/total_tokens=344,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1793/500000]
+ train/ActionL1Loss=0.1301
+ throughput/total_tokens=344,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1794/500000]
+ train/ActionL1Loss=0.1295
+ throughput/total_tokens=344,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1795/500000]
+ train/ActionL1Loss=0.1393
+ throughput/total_tokens=344,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1796/500000]
+ train/ActionL1Loss=0.1361
+ throughput/total_tokens=344,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1797/500000]
+ train/ActionL1Loss=0.1509
+ throughput/total_tokens=345,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1798/500000]
+ train/ActionL1Loss=0.1361
+ throughput/total_tokens=345,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1799/500000]
+ train/ActionL1Loss=0.1344
+ throughput/total_tokens=345,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1800/500000]
+ optim/total_grad_norm=6.884
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=345,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1801/500000]
+ train/ActionL1Loss=0.1274
+ throughput/total_tokens=345,792,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1802/500000]
+ train/ActionL1Loss=0.1352
+ throughput/total_tokens=345,984,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1803/500000]
+ train/ActionL1Loss=0.1229
+ throughput/total_tokens=346,176,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1804/500000]
+ train/ActionL1Loss=0.1254
+ throughput/total_tokens=346,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1805/500000]
+ train/ActionL1Loss=0.1270
+ throughput/total_tokens=346,560,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1806/500000]
+ train/ActionL1Loss=0.1313
+ throughput/total_tokens=346,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1807/500000]
+ train/ActionL1Loss=0.1262
+ throughput/total_tokens=346,944,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1808/500000]
+ train/ActionL1Loss=0.1249
+ throughput/total_tokens=347,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1809/500000]
+ train/ActionL1Loss=0.1243
+ throughput/total_tokens=347,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1810/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=347,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1811/500000]
+ train/ActionL1Loss=0.1143
+ throughput/total_tokens=347,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1812/500000]
+ train/ActionL1Loss=0.1243
+ throughput/total_tokens=347,904,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1813/500000]
+ train/ActionL1Loss=0.1261
+ throughput/total_tokens=348,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1814/500000]
+ train/ActionL1Loss=0.1293
+ throughput/total_tokens=348,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1815/500000]
+ train/ActionL1Loss=0.1232
+ throughput/total_tokens=348,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1816/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=348,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1817/500000]
+ train/ActionL1Loss=0.1379
+ throughput/total_tokens=348,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1818/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=349,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1819/500000]
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=349,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1820/500000]
+ optim/total_grad_norm=6.608
+ train/ActionL1Loss=0.1340
+ throughput/total_tokens=349,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1821/500000]
+ train/ActionL1Loss=0.1264
+ throughput/total_tokens=349,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1822/500000]
+ train/ActionL1Loss=0.1106
+ throughput/total_tokens=349,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1823/500000]
+ train/ActionL1Loss=0.1461
+ throughput/total_tokens=350,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1824/500000]
+ train/ActionL1Loss=0.1098
+ throughput/total_tokens=350,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1825/500000]
+ train/ActionL1Loss=0.1303
+ throughput/total_tokens=350,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1826/500000]
+ train/ActionL1Loss=0.1287
+ throughput/total_tokens=350,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1827/500000]
+ train/ActionL1Loss=0.1205
+ throughput/total_tokens=350,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1828/500000]
+ train/ActionL1Loss=0.1222
+ throughput/total_tokens=350,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1829/500000]
+ train/ActionL1Loss=0.1257
+ throughput/total_tokens=351,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1830/500000]
+ train/ActionL1Loss=0.1253
+ throughput/total_tokens=351,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1831/500000]
+ train/ActionL1Loss=0.1301
+ throughput/total_tokens=351,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1832/500000]
+ train/ActionL1Loss=0.1136
+ throughput/total_tokens=351,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1833/500000]
+ train/ActionL1Loss=0.1411
+ throughput/total_tokens=351,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1834/500000]
+ train/ActionL1Loss=0.1186
+ throughput/total_tokens=352,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1835/500000]
+ train/ActionL1Loss=0.1363
+ throughput/total_tokens=352,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1836/500000]
+ train/ActionL1Loss=0.1181
+ throughput/total_tokens=352,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1837/500000]
+ train/ActionL1Loss=0.1460
+ throughput/total_tokens=352,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1838/500000]
+ train/ActionL1Loss=0.1294
+ throughput/total_tokens=352,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1839/500000]
+ train/ActionL1Loss=0.1030
+ throughput/total_tokens=353,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1840/500000]
+ optim/total_grad_norm=7.481
+ train/ActionL1Loss=0.1261
+ throughput/total_tokens=353,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1841/500000]
+ train/ActionL1Loss=0.1356
+ throughput/total_tokens=353,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1842/500000]
+ train/ActionL1Loss=0.1367
+ throughput/total_tokens=353,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1843/500000]
+ train/ActionL1Loss=0.1547
+ throughput/total_tokens=353,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1844/500000]
+ train/ActionL1Loss=0.1454
+ throughput/total_tokens=354,048,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1845/500000]
+ train/ActionL1Loss=0.1411
+ throughput/total_tokens=354,240,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1846/500000]
+ train/ActionL1Loss=0.1234
+ throughput/total_tokens=354,432,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1847/500000]
+ train/ActionL1Loss=0.1215
+ throughput/total_tokens=354,624,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1848/500000]
+ train/ActionL1Loss=0.1259
+ throughput/total_tokens=354,816,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1849/500000]
+ train/ActionL1Loss=0.1216
+ throughput/total_tokens=355,008,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1850/500000]
+ train/ActionL1Loss=0.1380
+ throughput/total_tokens=355,200,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1851/500000]
+ train/ActionL1Loss=0.1497
+ throughput/total_tokens=355,392,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1852/500000]
+ train/ActionL1Loss=0.1398
+ throughput/total_tokens=355,584,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1853/500000]
+ train/ActionL1Loss=0.1382
+ throughput/total_tokens=355,776,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1854/500000]
+ train/ActionL1Loss=0.1267
+ throughput/total_tokens=355,968,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=1855/500000]
+ train/ActionL1Loss=0.1396
+ throughput/total_tokens=356,160,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1856/500000]
+ train/ActionL1Loss=0.1118
+ throughput/total_tokens=356,352,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1857/500000]
+ train/ActionL1Loss=0.1335
+ throughput/total_tokens=356,544,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1858/500000]
+ train/ActionL1Loss=0.1329
+ throughput/total_tokens=356,736,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1859/500000]
+ train/ActionL1Loss=0.1290
+ throughput/total_tokens=356,928,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1860/500000]
+ optim/total_grad_norm=7.482
+ train/ActionL1Loss=0.1414
+ throughput/total_tokens=357,120,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1861/500000]
+ train/ActionL1Loss=0.1384
+ throughput/total_tokens=357,312,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1862/500000]
+ train/ActionL1Loss=0.1345
+ throughput/total_tokens=357,504,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1863/500000]
+ train/ActionL1Loss=0.1194
+ throughput/total_tokens=357,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1864/500000]
+ train/ActionL1Loss=0.1080
+ throughput/total_tokens=357,888,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1865/500000]
+ train/ActionL1Loss=0.1301
+ throughput/total_tokens=358,080,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1866/500000]
+ train/ActionL1Loss=0.1045
+ throughput/total_tokens=358,272,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1867/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=358,464,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1868/500000]
+ train/ActionL1Loss=0.1257
+ throughput/total_tokens=358,656,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1869/500000]
+ train/ActionL1Loss=0.1228
+ throughput/total_tokens=358,848,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1870/500000]
+ train/ActionL1Loss=0.1147
+ throughput/total_tokens=359,040,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1871/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=359,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1872/500000]
+ train/ActionL1Loss=0.1014
+ throughput/total_tokens=359,424,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1873/500000]
+ train/ActionL1Loss=0.1130
+ throughput/total_tokens=359,616,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1874/500000]
+ train/ActionL1Loss=0.1327
+ throughput/total_tokens=359,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1875/500000]
+ train/ActionL1Loss=0.1279
+ throughput/total_tokens=360,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1876/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=360,192,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1877/500000]
+ train/ActionL1Loss=0.1457
+ throughput/total_tokens=360,384,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1878/500000]
+ train/ActionL1Loss=0.1307
+ throughput/total_tokens=360,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1879/500000]
+ train/ActionL1Loss=0.1368
+ throughput/total_tokens=360,768,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1880/500000]
+ optim/total_grad_norm=8.312
+ train/ActionL1Loss=0.1379
+ throughput/total_tokens=360,960,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1881/500000]
+ train/ActionL1Loss=0.1503
+ throughput/total_tokens=361,152,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1882/500000]
+ train/ActionL1Loss=0.1314
+ throughput/total_tokens=361,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1883/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=361,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1884/500000]
+ train/ActionL1Loss=0.1394
+ throughput/total_tokens=361,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1885/500000]
+ train/ActionL1Loss=0.1157
+ throughput/total_tokens=361,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1886/500000]
+ train/ActionL1Loss=0.1174
+ throughput/total_tokens=362,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1887/500000]
+ train/ActionL1Loss=0.1202
+ throughput/total_tokens=362,304,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1888/500000]
+ train/ActionL1Loss=0.1114
+ throughput/total_tokens=362,496,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1889/500000]
+ train/ActionL1Loss=0.1199
+ throughput/total_tokens=362,688,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1890/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=362,880,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1891/500000]
+ train/ActionL1Loss=0.1054
+ throughput/total_tokens=363,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1892/500000]
+ train/ActionL1Loss=0.1157
+ throughput/total_tokens=363,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1893/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=363,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1894/500000]
+ train/ActionL1Loss=0.1055
+ throughput/total_tokens=363,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1895/500000]
+ train/ActionL1Loss=0.1367
+ throughput/total_tokens=363,840,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1896/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=364,032,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1897/500000]
+ train/ActionL1Loss=0.1111
+ throughput/total_tokens=364,224,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1898/500000]
+ train/ActionL1Loss=0.1205
+ throughput/total_tokens=364,416,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1899/500000]
+ train/ActionL1Loss=0.1224
+ throughput/total_tokens=364,608,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1900/500000]
+ optim/total_grad_norm=7.113
+ train/ActionL1Loss=0.1343
+ throughput/total_tokens=364,800,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1901/500000]
+ train/ActionL1Loss=0.1142
+ throughput/total_tokens=364,992,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1902/500000]
+ train/ActionL1Loss=0.1121
+ throughput/total_tokens=365,184,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1903/500000]
+ train/ActionL1Loss=0.1150
+ throughput/total_tokens=365,376,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1904/500000]
+ train/ActionL1Loss=0.1210
+ throughput/total_tokens=365,568,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1905/500000]
+ train/ActionL1Loss=0.1109
+ throughput/total_tokens=365,760,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1906/500000]
+ train/ActionL1Loss=0.1260
+ throughput/total_tokens=365,952,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1907/500000]
+ train/ActionL1Loss=0.1176
+ throughput/total_tokens=366,144,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1908/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=366,336,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1909/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=366,528,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1910/500000]
+ train/ActionL1Loss=0.1260
+ throughput/total_tokens=366,720,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=1911/500000]
+ train/ActionL1Loss=0.1322
+ throughput/total_tokens=366,912,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1912/500000]
+ train/ActionL1Loss=0.1298
+ throughput/total_tokens=367,104,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1913/500000]
+ train/ActionL1Loss=0.1276
+ throughput/total_tokens=367,296,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1914/500000]
+ train/ActionL1Loss=0.1183
+ throughput/total_tokens=367,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1915/500000]
+ train/ActionL1Loss=0.1348
+ throughput/total_tokens=367,680,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1916/500000]
+ train/ActionL1Loss=0.1060
+ throughput/total_tokens=367,872,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1917/500000]
+ train/ActionL1Loss=0.1111
+ throughput/total_tokens=368,064,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1918/500000]
+ train/ActionL1Loss=0.1161
+ throughput/total_tokens=368,256,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1919/500000]
+ train/ActionL1Loss=0.1490
+ throughput/total_tokens=368,448,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1920/500000]
+ optim/total_grad_norm=8.388
+ train/ActionL1Loss=0.1171
+ throughput/total_tokens=368,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1921/500000]
+ train/ActionL1Loss=0.1289
+ throughput/total_tokens=368,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1922/500000]
+ train/ActionL1Loss=0.1358
+ throughput/total_tokens=369,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1923/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=369,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1924/500000]
+ train/ActionL1Loss=0.1202
+ throughput/total_tokens=369,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1925/500000]
+ train/ActionL1Loss=0.1167
+ throughput/total_tokens=369,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1926/500000]
+ train/ActionL1Loss=0.1156
+ throughput/total_tokens=369,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=1927/500000]
+ train/ActionL1Loss=0.1159
+ throughput/total_tokens=369,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1928/500000]
+ train/ActionL1Loss=0.1231
+ throughput/total_tokens=370,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1929/500000]
+ train/ActionL1Loss=0.1123
+ throughput/total_tokens=370,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1930/500000]
+ train/ActionL1Loss=0.1137
+ throughput/total_tokens=370,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1931/500000]
+ train/ActionL1Loss=0.1014
+ throughput/total_tokens=370,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1932/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=370,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1933/500000]
+ train/ActionL1Loss=0.1153
+ throughput/total_tokens=371,136,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1934/500000]
+ train/ActionL1Loss=0.1232
+ throughput/total_tokens=371,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1935/500000]
+ train/ActionL1Loss=0.1134
+ throughput/total_tokens=371,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1936/500000]
+ train/ActionL1Loss=0.1049
+ throughput/total_tokens=371,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1937/500000]
+ train/ActionL1Loss=0.1179
+ throughput/total_tokens=371,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1938/500000]
+ train/ActionL1Loss=0.1495
+ throughput/total_tokens=372,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1939/500000]
+ train/ActionL1Loss=0.1532
+ throughput/total_tokens=372,288,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1940/500000]
+ optim/total_grad_norm=6.180
+ train/ActionL1Loss=0.1249
+ throughput/total_tokens=372,480,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1941/500000]
+ train/ActionL1Loss=0.1362
+ throughput/total_tokens=372,672,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1942/500000]
+ train/ActionL1Loss=0.1396
+ throughput/total_tokens=372,864,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1943/500000]
+ train/ActionL1Loss=0.1124
+ throughput/total_tokens=373,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1944/500000]
+ train/ActionL1Loss=0.1116
+ throughput/total_tokens=373,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1945/500000]
+ train/ActionL1Loss=0.1276
+ throughput/total_tokens=373,440,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1946/500000]
+ train/ActionL1Loss=0.1237
+ throughput/total_tokens=373,632,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1947/500000]
+ train/ActionL1Loss=0.1396
+ throughput/total_tokens=373,824,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1948/500000]
+ train/ActionL1Loss=0.1399
+ throughput/total_tokens=374,016,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1949/500000]
+ train/ActionL1Loss=0.1296
+ throughput/total_tokens=374,208,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=1950/500000]
+ train/ActionL1Loss=0.1423
+ throughput/total_tokens=374,400,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1951/500000]
+ train/ActionL1Loss=0.1255
+ throughput/total_tokens=374,592,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1952/500000]
+ train/ActionL1Loss=0.1033
+ throughput/total_tokens=374,784,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1953/500000]
+ train/ActionL1Loss=0.1382
+ throughput/total_tokens=374,976,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=1954/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=375,168,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1955/500000]
+ train/ActionL1Loss=0.1232
+ throughput/total_tokens=375,360,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1956/500000]
+ train/ActionL1Loss=0.1245
+ throughput/total_tokens=375,552,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1957/500000]
+ train/ActionL1Loss=0.1127
+ throughput/total_tokens=375,744,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1958/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=375,936,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1959/500000]
+ train/ActionL1Loss=0.1079
+ throughput/total_tokens=376,128,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1960/500000]
+ optim/total_grad_norm=7.108
+ train/ActionL1Loss=0.1195
+ throughput/total_tokens=376,320,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=1961/500000]
+ train/ActionL1Loss=0.1003
+ throughput/total_tokens=376,512,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1962/500000]
+ train/ActionL1Loss=0.1216
+ throughput/total_tokens=376,704,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1963/500000]
+ train/ActionL1Loss=0.1138
+ throughput/total_tokens=376,896,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1964/500000]
+ train/ActionL1Loss=0.1309
+ throughput/total_tokens=377,088,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1965/500000]
+ train/ActionL1Loss=0.1245
+ throughput/total_tokens=377,280,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1966/500000]
+ train/ActionL1Loss=0.1212
+ throughput/total_tokens=377,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=1967/500000]
+ train/ActionL1Loss=0.1089
+ throughput/total_tokens=377,664,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1968/500000]
+ train/ActionL1Loss=0.1227
+ throughput/total_tokens=377,856,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=1969/500000]
+ train/ActionL1Loss=0.1405
+ throughput/total_tokens=378,048,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=1970/500000]
+ train/ActionL1Loss=0.1119
+ throughput/total_tokens=378,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1971/500000]
+ train/ActionL1Loss=0.1200
+ throughput/total_tokens=378,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1972/500000]
+ train/ActionL1Loss=0.1093
+ throughput/total_tokens=378,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1973/500000]
+ train/ActionL1Loss=0.1162
+ throughput/total_tokens=378,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1974/500000]
+ train/ActionL1Loss=0.1366
+ throughput/total_tokens=379,008,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1975/500000]
+ train/ActionL1Loss=0.1092
+ throughput/total_tokens=379,200,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1976/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=379,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1977/500000]
+ train/ActionL1Loss=0.1275
+ throughput/total_tokens=379,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1978/500000]
+ train/ActionL1Loss=0.1586
+ throughput/total_tokens=379,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1979/500000]
+ train/ActionL1Loss=0.1303
+ throughput/total_tokens=379,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1980/500000]
+ optim/total_grad_norm=6.323
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=380,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1981/500000]
+ train/ActionL1Loss=0.1182
+ throughput/total_tokens=380,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1982/500000]
+ train/ActionL1Loss=0.1274
+ throughput/total_tokens=380,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1983/500000]
+ train/ActionL1Loss=0.1006
+ throughput/total_tokens=380,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1984/500000]
+ train/ActionL1Loss=0.1361
+ throughput/total_tokens=380,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1985/500000]
+ train/ActionL1Loss=0.1267
+ throughput/total_tokens=381,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1986/500000]
+ train/ActionL1Loss=0.1009
+ throughput/total_tokens=381,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1987/500000]
+ train/ActionL1Loss=0.1113
+ throughput/total_tokens=381,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1988/500000]
+ train/ActionL1Loss=0.1209
+ throughput/total_tokens=381,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1989/500000]
+ train/ActionL1Loss=0.1052
+ throughput/total_tokens=381,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1990/500000]
+ train/ActionL1Loss=0.1047
+ throughput/total_tokens=382,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=1991/500000]
+ train/ActionL1Loss=0.1102
+ throughput/total_tokens=382,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1992/500000]
+ train/ActionL1Loss=0.1346
+ throughput/total_tokens=382,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1993/500000]
+ train/ActionL1Loss=0.1000
+ throughput/total_tokens=382,656,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1994/500000]
+ train/ActionL1Loss=0.1122
+ throughput/total_tokens=382,848,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1995/500000]
+ train/ActionL1Loss=0.1267
+ throughput/total_tokens=383,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1996/500000]
+ train/ActionL1Loss=0.1080
+ throughput/total_tokens=383,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=1997/500000]
+ train/ActionL1Loss=0.1361
+ throughput/total_tokens=383,424,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1998/500000]
+ train/ActionL1Loss=0.1232
+ throughput/total_tokens=383,616,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=1999/500000]
+ train/ActionL1Loss=0.1345
+ throughput/total_tokens=383,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2000/500000]
+ optim/total_grad_norm=5.123
+ train/ActionL1Loss=0.1279
+ throughput/total_tokens=384,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+10/01 [00:59:45] INFO | >> Saving config... checkpoint.py:608
+10/01 [01:00:16] INFO | >> Saving model state... checkpoint.py:796
+10/01 [01:01:27] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [01:02:54] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=2001/500000]
+ train/ActionL1Loss=0.1126
+ throughput/total_tokens=384,192,000
+ throughput/device/tokens_per_second=1,210
+ throughput/device/batches_per_second=0.0504
+[step=2002/500000]
+ train/ActionL1Loss=0.1416
+ throughput/total_tokens=384,384,000
+ throughput/device/tokens_per_second=1,184
+ throughput/device/batches_per_second=0.0494
+[step=2003/500000]
+ train/ActionL1Loss=0.1239
+ throughput/total_tokens=384,576,000
+ throughput/device/tokens_per_second=1,176
+ throughput/device/batches_per_second=0.0490
+[step=2004/500000]
+ train/ActionL1Loss=0.1171
+ throughput/total_tokens=384,768,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=2005/500000]
+ train/ActionL1Loss=0.1255
+ throughput/total_tokens=384,960,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=2006/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=385,152,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=2007/500000]
+ train/ActionL1Loss=0.1405
+ throughput/total_tokens=385,344,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=2008/500000]
+ train/ActionL1Loss=0.1373
+ throughput/total_tokens=385,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=2009/500000]
+ train/ActionL1Loss=0.1390
+ throughput/total_tokens=385,728,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=2010/500000]
+ train/ActionL1Loss=0.1319
+ throughput/total_tokens=385,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=2011/500000]
+ train/ActionL1Loss=0.1137
+ throughput/total_tokens=386,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=2012/500000]
+ train/ActionL1Loss=0.1170
+ throughput/total_tokens=386,304,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2013/500000]
+ train/ActionL1Loss=0.1200
+ throughput/total_tokens=386,496,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2014/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=386,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2015/500000]
+ train/ActionL1Loss=0.1278
+ throughput/total_tokens=386,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2016/500000]
+ train/ActionL1Loss=0.1324
+ throughput/total_tokens=387,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2017/500000]
+ train/ActionL1Loss=0.1225
+ throughput/total_tokens=387,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2018/500000]
+ train/ActionL1Loss=0.1297
+ throughput/total_tokens=387,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2019/500000]
+ train/ActionL1Loss=0.1146
+ throughput/total_tokens=387,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2020/500000]
+ optim/total_grad_norm=8.507
+ train/ActionL1Loss=0.1273
+ throughput/total_tokens=387,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2021/500000]
+ train/ActionL1Loss=0.1200
+ throughput/total_tokens=388,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2022/500000]
+ train/ActionL1Loss=0.1137
+ throughput/total_tokens=388,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2023/500000]
+ train/ActionL1Loss=0.1347
+ throughput/total_tokens=388,416,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2024/500000]
+ train/ActionL1Loss=0.1315
+ throughput/total_tokens=388,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2025/500000]
+ train/ActionL1Loss=0.1423
+ throughput/total_tokens=388,800,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2026/500000]
+ train/ActionL1Loss=0.1306
+ throughput/total_tokens=388,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2027/500000]
+ train/ActionL1Loss=0.1244
+ throughput/total_tokens=389,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2028/500000]
+ train/ActionL1Loss=0.1273
+ throughput/total_tokens=389,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2029/500000]
+ train/ActionL1Loss=0.1303
+ throughput/total_tokens=389,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2030/500000]
+ train/ActionL1Loss=0.1186
+ throughput/total_tokens=389,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2031/500000]
+ train/ActionL1Loss=0.1081
+ throughput/total_tokens=389,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2032/500000]
+ train/ActionL1Loss=0.1141
+ throughput/total_tokens=390,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2033/500000]
+ train/ActionL1Loss=0.1242
+ throughput/total_tokens=390,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2034/500000]
+ train/ActionL1Loss=0.1250
+ throughput/total_tokens=390,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2035/500000]
+ train/ActionL1Loss=0.1074
+ throughput/total_tokens=390,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2036/500000]
+ train/ActionL1Loss=0.1264
+ throughput/total_tokens=390,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2037/500000]
+ train/ActionL1Loss=0.1130
+ throughput/total_tokens=391,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2038/500000]
+ train/ActionL1Loss=0.0977
+ throughput/total_tokens=391,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2039/500000]
+ train/ActionL1Loss=0.1270
+ throughput/total_tokens=391,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2040/500000]
+ optim/total_grad_norm=8.681
+ train/ActionL1Loss=0.1248
+ throughput/total_tokens=391,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2041/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=391,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2042/500000]
+ train/ActionL1Loss=0.1197
+ throughput/total_tokens=392,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2043/500000]
+ train/ActionL1Loss=0.1158
+ throughput/total_tokens=392,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2044/500000]
+ train/ActionL1Loss=0.1326
+ throughput/total_tokens=392,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2045/500000]
+ train/ActionL1Loss=0.1247
+ throughput/total_tokens=392,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2046/500000]
+ train/ActionL1Loss=0.1481
+ throughput/total_tokens=392,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2047/500000]
+ train/ActionL1Loss=0.0987
+ throughput/total_tokens=393,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2048/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=393,216,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2049/500000]
+ train/ActionL1Loss=0.1278
+ throughput/total_tokens=393,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2050/500000]
+ train/ActionL1Loss=0.1388
+ throughput/total_tokens=393,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2051/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=393,792,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2052/500000]
+ train/ActionL1Loss=0.1292
+ throughput/total_tokens=393,984,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2053/500000]
+ train/ActionL1Loss=0.1029
+ throughput/total_tokens=394,176,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2054/500000]
+ train/ActionL1Loss=0.1042
+ throughput/total_tokens=394,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2055/500000]
+ train/ActionL1Loss=0.0995
+ throughput/total_tokens=394,560,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2056/500000]
+ train/ActionL1Loss=0.1139
+ throughput/total_tokens=394,752,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2057/500000]
+ train/ActionL1Loss=0.1063
+ throughput/total_tokens=394,944,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2058/500000]
+ train/ActionL1Loss=0.1166
+ throughput/total_tokens=395,136,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2059/500000]
+ train/ActionL1Loss=0.1084
+ throughput/total_tokens=395,328,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2060/500000]
+ optim/total_grad_norm=7.364
+ train/ActionL1Loss=0.1137
+ throughput/total_tokens=395,520,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2061/500000]
+ train/ActionL1Loss=0.1105
+ throughput/total_tokens=395,712,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2062/500000]
+ train/ActionL1Loss=0.1443
+ throughput/total_tokens=395,904,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2063/500000]
+ train/ActionL1Loss=0.1344
+ throughput/total_tokens=396,096,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2064/500000]
+ train/ActionL1Loss=0.1158
+ throughput/total_tokens=396,288,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2065/500000]
+ train/ActionL1Loss=0.1174
+ throughput/total_tokens=396,480,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2066/500000]
+ train/ActionL1Loss=0.0993
+ throughput/total_tokens=396,672,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2067/500000]
+ train/ActionL1Loss=0.1219
+ throughput/total_tokens=396,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2068/500000]
+ train/ActionL1Loss=0.1265
+ throughput/total_tokens=397,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2069/500000]
+ train/ActionL1Loss=0.1083
+ throughput/total_tokens=397,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2070/500000]
+ train/ActionL1Loss=0.1065
+ throughput/total_tokens=397,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2071/500000]
+ train/ActionL1Loss=0.1231
+ throughput/total_tokens=397,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2072/500000]
+ train/ActionL1Loss=0.1113
+ throughput/total_tokens=397,824,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2073/500000]
+ train/ActionL1Loss=0.1374
+ throughput/total_tokens=398,016,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2074/500000]
+ train/ActionL1Loss=0.1076
+ throughput/total_tokens=398,208,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2075/500000]
+ train/ActionL1Loss=0.1079
+ throughput/total_tokens=398,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2076/500000]
+ train/ActionL1Loss=0.1060
+ throughput/total_tokens=398,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2077/500000]
+ train/ActionL1Loss=0.1206
+ throughput/total_tokens=398,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2078/500000]
+ train/ActionL1Loss=0.1213
+ throughput/total_tokens=398,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2079/500000]
+ train/ActionL1Loss=0.1003
+ throughput/total_tokens=399,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2080/500000]
+ optim/total_grad_norm=7.127
+ train/ActionL1Loss=0.1095
+ throughput/total_tokens=399,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2081/500000]
+ train/ActionL1Loss=0.1328
+ throughput/total_tokens=399,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2082/500000]
+ train/ActionL1Loss=0.1384
+ throughput/total_tokens=399,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2083/500000]
+ train/ActionL1Loss=0.1268
+ throughput/total_tokens=399,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2084/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=400,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2085/500000]
+ train/ActionL1Loss=0.1197
+ throughput/total_tokens=400,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2086/500000]
+ train/ActionL1Loss=0.1227
+ throughput/total_tokens=400,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2087/500000]
+ train/ActionL1Loss=0.1155
+ throughput/total_tokens=400,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2088/500000]
+ train/ActionL1Loss=0.1248
+ throughput/total_tokens=400,896,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2089/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=401,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2090/500000]
+ train/ActionL1Loss=0.1067
+ throughput/total_tokens=401,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2091/500000]
+ train/ActionL1Loss=0.1307
+ throughput/total_tokens=401,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2092/500000]
+ train/ActionL1Loss=0.1300
+ throughput/total_tokens=401,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2093/500000]
+ train/ActionL1Loss=0.1398
+ throughput/total_tokens=401,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2094/500000]
+ train/ActionL1Loss=0.1088
+ throughput/total_tokens=402,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2095/500000]
+ train/ActionL1Loss=0.1033
+ throughput/total_tokens=402,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2096/500000]
+ train/ActionL1Loss=0.1029
+ throughput/total_tokens=402,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2097/500000]
+ train/ActionL1Loss=0.1169
+ throughput/total_tokens=402,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2098/500000]
+ train/ActionL1Loss=0.1268
+ throughput/total_tokens=402,816,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2099/500000]
+ train/ActionL1Loss=0.1213
+ throughput/total_tokens=403,008,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2100/500000]
+ optim/total_grad_norm=6.811
+ train/ActionL1Loss=0.1087
+ throughput/total_tokens=403,200,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2101/500000]
+ train/ActionL1Loss=0.1170
+ throughput/total_tokens=403,392,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2102/500000]
+ train/ActionL1Loss=0.1057
+ throughput/total_tokens=403,584,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2103/500000]
+ train/ActionL1Loss=0.1138
+ throughput/total_tokens=403,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2104/500000]
+ train/ActionL1Loss=0.1301
+ throughput/total_tokens=403,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2105/500000]
+ train/ActionL1Loss=0.1360
+ throughput/total_tokens=404,160,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2106/500000]
+ train/ActionL1Loss=0.1105
+ throughput/total_tokens=404,352,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2107/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=404,544,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2108/500000]
+ train/ActionL1Loss=0.1057
+ throughput/total_tokens=404,736,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2109/500000]
+ train/ActionL1Loss=0.1302
+ throughput/total_tokens=404,928,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2110/500000]
+ train/ActionL1Loss=0.1044
+ throughput/total_tokens=405,120,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2111/500000]
+ train/ActionL1Loss=0.1208
+ throughput/total_tokens=405,312,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2112/500000]
+ train/ActionL1Loss=0.1154
+ throughput/total_tokens=405,504,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2113/500000]
+ train/ActionL1Loss=0.1292
+ throughput/total_tokens=405,696,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2114/500000]
+ train/ActionL1Loss=0.1200
+ throughput/total_tokens=405,888,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2115/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=406,080,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2116/500000]
+ train/ActionL1Loss=0.1116
+ throughput/total_tokens=406,272,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2117/500000]
+ train/ActionL1Loss=0.1168
+ throughput/total_tokens=406,464,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2118/500000]
+ train/ActionL1Loss=0.1180
+ throughput/total_tokens=406,656,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2119/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=406,848,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2120/500000]
+ optim/total_grad_norm=5.559
+ train/ActionL1Loss=0.1271
+ throughput/total_tokens=407,040,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2121/500000]
+ train/ActionL1Loss=0.1273
+ throughput/total_tokens=407,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2122/500000]
+ train/ActionL1Loss=0.1039
+ throughput/total_tokens=407,424,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2123/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=407,616,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2124/500000]
+ train/ActionL1Loss=0.1266
+ throughput/total_tokens=407,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2125/500000]
+ train/ActionL1Loss=0.1074
+ throughput/total_tokens=408,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2126/500000]
+ train/ActionL1Loss=0.1281
+ throughput/total_tokens=408,192,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2127/500000]
+ train/ActionL1Loss=0.1110
+ throughput/total_tokens=408,384,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2128/500000]
+ train/ActionL1Loss=0.1035
+ throughput/total_tokens=408,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2129/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=408,768,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2130/500000]
+ train/ActionL1Loss=0.1173
+ throughput/total_tokens=408,960,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2131/500000]
+ train/ActionL1Loss=0.1103
+ throughput/total_tokens=409,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2132/500000]
+ train/ActionL1Loss=0.1069
+ throughput/total_tokens=409,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2133/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=409,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2134/500000]
+ train/ActionL1Loss=0.1200
+ throughput/total_tokens=409,728,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2135/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=409,920,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2136/500000]
+ train/ActionL1Loss=0.1152
+ throughput/total_tokens=410,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2137/500000]
+ train/ActionL1Loss=0.1095
+ throughput/total_tokens=410,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2138/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=410,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2139/500000]
+ train/ActionL1Loss=0.1037
+ throughput/total_tokens=410,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2140/500000]
+ optim/total_grad_norm=9.173
+ train/ActionL1Loss=0.1195
+ throughput/total_tokens=410,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2141/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=411,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2142/500000]
+ train/ActionL1Loss=0.1084
+ throughput/total_tokens=411,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2143/500000]
+ train/ActionL1Loss=0.1050
+ throughput/total_tokens=411,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2144/500000]
+ train/ActionL1Loss=0.1187
+ throughput/total_tokens=411,648,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2145/500000]
+ train/ActionL1Loss=0.1260
+ throughput/total_tokens=411,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2146/500000]
+ train/ActionL1Loss=0.1299
+ throughput/total_tokens=412,032,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2147/500000]
+ train/ActionL1Loss=0.1002
+ throughput/total_tokens=412,224,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2148/500000]
+ train/ActionL1Loss=0.1211
+ throughput/total_tokens=412,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2149/500000]
+ train/ActionL1Loss=0.1178
+ throughput/total_tokens=412,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2150/500000]
+ train/ActionL1Loss=0.1090
+ throughput/total_tokens=412,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2151/500000]
+ train/ActionL1Loss=0.1030
+ throughput/total_tokens=412,992,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2152/500000]
+ train/ActionL1Loss=0.1280
+ throughput/total_tokens=413,184,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2153/500000]
+ train/ActionL1Loss=0.1352
+ throughput/total_tokens=413,376,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2154/500000]
+ train/ActionL1Loss=0.1112
+ throughput/total_tokens=413,568,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2155/500000]
+ train/ActionL1Loss=0.1044
+ throughput/total_tokens=413,760,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=2156/500000]
+ train/ActionL1Loss=0.1278
+ throughput/total_tokens=413,952,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=2157/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=414,144,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2158/500000]
+ train/ActionL1Loss=0.1163
+ throughput/total_tokens=414,336,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2159/500000]
+ train/ActionL1Loss=0.1344
+ throughput/total_tokens=414,528,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2160/500000]
+ optim/total_grad_norm=5.818
+ train/ActionL1Loss=0.1238
+ throughput/total_tokens=414,720,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2161/500000]
+ train/ActionL1Loss=0.1060
+ throughput/total_tokens=414,912,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2162/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=415,104,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2163/500000]
+ train/ActionL1Loss=0.1144
+ throughput/total_tokens=415,296,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2164/500000]
+ train/ActionL1Loss=0.1136
+ throughput/total_tokens=415,488,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2165/500000]
+ train/ActionL1Loss=0.1121
+ throughput/total_tokens=415,680,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2166/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=415,872,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2167/500000]
+ train/ActionL1Loss=0.1015
+ throughput/total_tokens=416,064,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2168/500000]
+ train/ActionL1Loss=0.1146
+ throughput/total_tokens=416,256,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2169/500000]
+ train/ActionL1Loss=0.1101
+ throughput/total_tokens=416,448,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2170/500000]
+ train/ActionL1Loss=0.1332
+ throughput/total_tokens=416,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2171/500000]
+ train/ActionL1Loss=0.1035
+ throughput/total_tokens=416,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2172/500000]
+ train/ActionL1Loss=0.1071
+ throughput/total_tokens=417,024,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2173/500000]
+ train/ActionL1Loss=0.1235
+ throughput/total_tokens=417,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2174/500000]
+ train/ActionL1Loss=0.1240
+ throughput/total_tokens=417,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2175/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=417,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2176/500000]
+ train/ActionL1Loss=0.1302
+ throughput/total_tokens=417,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2177/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=417,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2178/500000]
+ train/ActionL1Loss=0.1114
+ throughput/total_tokens=418,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2179/500000]
+ train/ActionL1Loss=0.1105
+ throughput/total_tokens=418,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2180/500000]
+ optim/total_grad_norm=6.645
+ train/ActionL1Loss=0.1180
+ throughput/total_tokens=418,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2181/500000]
+ train/ActionL1Loss=0.1100
+ throughput/total_tokens=418,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2182/500000]
+ train/ActionL1Loss=0.1332
+ throughput/total_tokens=418,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2183/500000]
+ train/ActionL1Loss=0.1236
+ throughput/total_tokens=419,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2184/500000]
+ train/ActionL1Loss=0.1040
+ throughput/total_tokens=419,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2185/500000]
+ train/ActionL1Loss=0.1000
+ throughput/total_tokens=419,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2186/500000]
+ train/ActionL1Loss=0.1144
+ throughput/total_tokens=419,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2187/500000]
+ train/ActionL1Loss=0.1090
+ throughput/total_tokens=419,904,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2188/500000]
+ train/ActionL1Loss=0.1231
+ throughput/total_tokens=420,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2189/500000]
+ train/ActionL1Loss=0.1337
+ throughput/total_tokens=420,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2190/500000]
+ train/ActionL1Loss=0.1162
+ throughput/total_tokens=420,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2191/500000]
+ train/ActionL1Loss=0.0983
+ throughput/total_tokens=420,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2192/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=420,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2193/500000]
+ train/ActionL1Loss=0.1106
+ throughput/total_tokens=421,056,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2194/500000]
+ train/ActionL1Loss=0.1381
+ throughput/total_tokens=421,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2195/500000]
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=421,440,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2196/500000]
+ train/ActionL1Loss=0.1392
+ throughput/total_tokens=421,632,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2197/500000]
+ train/ActionL1Loss=0.1123
+ throughput/total_tokens=421,824,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2198/500000]
+ train/ActionL1Loss=0.1132
+ throughput/total_tokens=422,016,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2199/500000]
+ train/ActionL1Loss=0.1164
+ throughput/total_tokens=422,208,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2200/500000]
+ optim/total_grad_norm=7.288
+ train/ActionL1Loss=0.1134
+ throughput/total_tokens=422,400,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2201/500000]
+ train/ActionL1Loss=0.1197
+ throughput/total_tokens=422,592,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2202/500000]
+ train/ActionL1Loss=0.1172
+ throughput/total_tokens=422,784,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2203/500000]
+ train/ActionL1Loss=0.1292
+ throughput/total_tokens=422,976,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2204/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=423,168,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2205/500000]
+ train/ActionL1Loss=0.1012
+ throughput/total_tokens=423,360,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2206/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=423,552,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2207/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=423,744,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2208/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=423,936,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2209/500000]
+ train/ActionL1Loss=0.1026
+ throughput/total_tokens=424,128,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2210/500000]
+ train/ActionL1Loss=0.1302
+ throughput/total_tokens=424,320,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2211/500000]
+ train/ActionL1Loss=0.1167
+ throughput/total_tokens=424,512,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2212/500000]
+ train/ActionL1Loss=0.1142
+ throughput/total_tokens=424,704,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2213/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=424,896,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2214/500000]
+ train/ActionL1Loss=0.1126
+ throughput/total_tokens=425,088,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2215/500000]
+ train/ActionL1Loss=0.1036
+ throughput/total_tokens=425,280,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2216/500000]
+ train/ActionL1Loss=0.1097
+ throughput/total_tokens=425,472,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2217/500000]
+ train/ActionL1Loss=0.0972
+ throughput/total_tokens=425,664,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2218/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=425,856,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2219/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=426,048,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2220/500000]
+ optim/total_grad_norm=8.429
+ train/ActionL1Loss=0.1286
+ throughput/total_tokens=426,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2221/500000]
+ train/ActionL1Loss=0.0960
+ throughput/total_tokens=426,432,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2222/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=426,624,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2223/500000]
+ train/ActionL1Loss=0.1221
+ throughput/total_tokens=426,816,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2224/500000]
+ train/ActionL1Loss=0.1257
+ throughput/total_tokens=427,008,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2225/500000]
+ train/ActionL1Loss=0.1280
+ throughput/total_tokens=427,200,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2226/500000]
+ train/ActionL1Loss=0.1394
+ throughput/total_tokens=427,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2227/500000]
+ train/ActionL1Loss=0.1037
+ throughput/total_tokens=427,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2228/500000]
+ train/ActionL1Loss=0.0984
+ throughput/total_tokens=427,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2229/500000]
+ train/ActionL1Loss=0.1321
+ throughput/total_tokens=427,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2230/500000]
+ train/ActionL1Loss=0.1135
+ throughput/total_tokens=428,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2231/500000]
+ train/ActionL1Loss=0.1102
+ throughput/total_tokens=428,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2232/500000]
+ train/ActionL1Loss=0.0905
+ throughput/total_tokens=428,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2233/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=428,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2234/500000]
+ train/ActionL1Loss=0.1259
+ throughput/total_tokens=428,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2235/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=429,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2236/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=429,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2237/500000]
+ train/ActionL1Loss=0.1001
+ throughput/total_tokens=429,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2238/500000]
+ train/ActionL1Loss=0.0964
+ throughput/total_tokens=429,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2239/500000]
+ train/ActionL1Loss=0.1169
+ throughput/total_tokens=429,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2240/500000]
+ optim/total_grad_norm=6.407
+ train/ActionL1Loss=0.1079
+ throughput/total_tokens=430,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2241/500000]
+ train/ActionL1Loss=0.1169
+ throughput/total_tokens=430,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2242/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=430,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2243/500000]
+ train/ActionL1Loss=0.1263
+ throughput/total_tokens=430,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2244/500000]
+ train/ActionL1Loss=0.1069
+ throughput/total_tokens=430,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2245/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=431,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2246/500000]
+ train/ActionL1Loss=0.1089
+ throughput/total_tokens=431,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2247/500000]
+ train/ActionL1Loss=0.1017
+ throughput/total_tokens=431,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2248/500000]
+ train/ActionL1Loss=0.0995
+ throughput/total_tokens=431,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2249/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=431,808,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2250/500000]
+ train/ActionL1Loss=0.1099
+ throughput/total_tokens=432,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2251/500000]
+ train/ActionL1Loss=0.1151
+ throughput/total_tokens=432,192,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2252/500000]
+ train/ActionL1Loss=0.1163
+ throughput/total_tokens=432,384,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2253/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=432,576,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2254/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=432,768,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2255/500000]
+ train/ActionL1Loss=0.1011
+ throughput/total_tokens=432,960,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2256/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=433,152,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2257/500000]
+ train/ActionL1Loss=0.0836
+ throughput/total_tokens=433,344,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2258/500000]
+ train/ActionL1Loss=0.1107
+ throughput/total_tokens=433,536,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2259/500000]
+ train/ActionL1Loss=0.1008
+ throughput/total_tokens=433,728,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2260/500000]
+ optim/total_grad_norm=6.029
+ train/ActionL1Loss=0.1150
+ throughput/total_tokens=433,920,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2261/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=434,112,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2262/500000]
+ train/ActionL1Loss=0.1172
+ throughput/total_tokens=434,304,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2263/500000]
+ train/ActionL1Loss=0.0974
+ throughput/total_tokens=434,496,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2264/500000]
+ train/ActionL1Loss=0.1380
+ throughput/total_tokens=434,688,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2265/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=434,880,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2266/500000]
+ train/ActionL1Loss=0.1275
+ throughput/total_tokens=435,072,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2267/500000]
+ train/ActionL1Loss=0.1257
+ throughput/total_tokens=435,264,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2268/500000]
+ train/ActionL1Loss=0.1205
+ throughput/total_tokens=435,456,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2269/500000]
+ train/ActionL1Loss=0.1303
+ throughput/total_tokens=435,648,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2270/500000]
+ train/ActionL1Loss=0.1115
+ throughput/total_tokens=435,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2271/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=436,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2272/500000]
+ train/ActionL1Loss=0.1282
+ throughput/total_tokens=436,224,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2273/500000]
+ train/ActionL1Loss=0.1170
+ throughput/total_tokens=436,416,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2274/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=436,608,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2275/500000]
+ train/ActionL1Loss=0.1055
+ throughput/total_tokens=436,800,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2276/500000]
+ train/ActionL1Loss=0.0950
+ throughput/total_tokens=436,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2277/500000]
+ train/ActionL1Loss=0.1310
+ throughput/total_tokens=437,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2278/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=437,376,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2279/500000]
+ train/ActionL1Loss=0.1295
+ throughput/total_tokens=437,568,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2280/500000]
+ optim/total_grad_norm=5.443
+ train/ActionL1Loss=0.1022
+ throughput/total_tokens=437,760,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2281/500000]
+ train/ActionL1Loss=0.1113
+ throughput/total_tokens=437,952,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2282/500000]
+ train/ActionL1Loss=0.1032
+ throughput/total_tokens=438,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2283/500000]
+ train/ActionL1Loss=0.1211
+ throughput/total_tokens=438,336,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2284/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=438,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2285/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=438,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2286/500000]
+ train/ActionL1Loss=0.1141
+ throughput/total_tokens=438,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2287/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=439,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2288/500000]
+ train/ActionL1Loss=0.1143
+ throughput/total_tokens=439,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2289/500000]
+ train/ActionL1Loss=0.1082
+ throughput/total_tokens=439,488,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2290/500000]
+ train/ActionL1Loss=0.1057
+ throughput/total_tokens=439,680,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2291/500000]
+ train/ActionL1Loss=0.1216
+ throughput/total_tokens=439,872,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2292/500000]
+ train/ActionL1Loss=0.1178
+ throughput/total_tokens=440,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2293/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=440,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2294/500000]
+ train/ActionL1Loss=0.1069
+ throughput/total_tokens=440,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2295/500000]
+ train/ActionL1Loss=0.0960
+ throughput/total_tokens=440,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2296/500000]
+ train/ActionL1Loss=0.0985
+ throughput/total_tokens=440,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2297/500000]
+ train/ActionL1Loss=0.1211
+ throughput/total_tokens=441,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2298/500000]
+ train/ActionL1Loss=0.1207
+ throughput/total_tokens=441,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2299/500000]
+ train/ActionL1Loss=0.1099
+ throughput/total_tokens=441,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2300/500000]
+ optim/total_grad_norm=6.286
+ train/ActionL1Loss=0.1293
+ throughput/total_tokens=441,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2301/500000]
+ train/ActionL1Loss=0.1115
+ throughput/total_tokens=441,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2302/500000]
+ train/ActionL1Loss=0.1414
+ throughput/total_tokens=441,984,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2303/500000]
+ train/ActionL1Loss=0.1174
+ throughput/total_tokens=442,176,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2304/500000]
+ train/ActionL1Loss=0.1234
+ throughput/total_tokens=442,368,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2305/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=442,560,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2306/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=442,752,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2307/500000]
+ train/ActionL1Loss=0.1063
+ throughput/total_tokens=442,944,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2308/500000]
+ train/ActionL1Loss=0.1158
+ throughput/total_tokens=443,136,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2309/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=443,328,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2310/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=443,520,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2311/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=443,712,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2312/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=443,904,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2313/500000]
+ train/ActionL1Loss=0.1055
+ throughput/total_tokens=444,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2314/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=444,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2315/500000]
+ train/ActionL1Loss=0.1000
+ throughput/total_tokens=444,480,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2316/500000]
+ train/ActionL1Loss=0.1149
+ throughput/total_tokens=444,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2317/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=444,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2318/500000]
+ train/ActionL1Loss=0.1018
+ throughput/total_tokens=445,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2319/500000]
+ train/ActionL1Loss=0.1115
+ throughput/total_tokens=445,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2320/500000]
+ optim/total_grad_norm=7.541
+ train/ActionL1Loss=0.1095
+ throughput/total_tokens=445,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2321/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=445,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2322/500000]
+ train/ActionL1Loss=0.1061
+ throughput/total_tokens=445,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2323/500000]
+ train/ActionL1Loss=0.1117
+ throughput/total_tokens=446,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2324/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=446,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2325/500000]
+ train/ActionL1Loss=0.1155
+ throughput/total_tokens=446,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2326/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=446,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2327/500000]
+ train/ActionL1Loss=0.0953
+ throughput/total_tokens=446,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2328/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=446,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2329/500000]
+ train/ActionL1Loss=0.1252
+ throughput/total_tokens=447,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2330/500000]
+ train/ActionL1Loss=0.1359
+ throughput/total_tokens=447,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2331/500000]
+ train/ActionL1Loss=0.1357
+ throughput/total_tokens=447,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2332/500000]
+ train/ActionL1Loss=0.1118
+ throughput/total_tokens=447,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2333/500000]
+ train/ActionL1Loss=0.1239
+ throughput/total_tokens=447,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2334/500000]
+ train/ActionL1Loss=0.1122
+ throughput/total_tokens=448,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2335/500000]
+ train/ActionL1Loss=0.1067
+ throughput/total_tokens=448,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2336/500000]
+ train/ActionL1Loss=0.1279
+ throughput/total_tokens=448,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2337/500000]
+ train/ActionL1Loss=0.1076
+ throughput/total_tokens=448,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2338/500000]
+ train/ActionL1Loss=0.1075
+ throughput/total_tokens=448,896,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2339/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=449,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2340/500000]
+ optim/total_grad_norm=5.573
+ train/ActionL1Loss=0.0977
+ throughput/total_tokens=449,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2341/500000]
+ train/ActionL1Loss=0.1218
+ throughput/total_tokens=449,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2342/500000]
+ train/ActionL1Loss=0.1036
+ throughput/total_tokens=449,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2343/500000]
+ train/ActionL1Loss=0.1061
+ throughput/total_tokens=449,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2344/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=450,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2345/500000]
+ train/ActionL1Loss=0.1169
+ throughput/total_tokens=450,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2346/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=450,432,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2347/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=450,624,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2348/500000]
+ train/ActionL1Loss=0.1117
+ throughput/total_tokens=450,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2349/500000]
+ train/ActionL1Loss=0.1055
+ throughput/total_tokens=451,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2350/500000]
+ train/ActionL1Loss=0.1028
+ throughput/total_tokens=451,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2351/500000]
+ train/ActionL1Loss=0.1166
+ throughput/total_tokens=451,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2352/500000]
+ train/ActionL1Loss=0.1046
+ throughput/total_tokens=451,584,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2353/500000]
+ train/ActionL1Loss=0.1107
+ throughput/total_tokens=451,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2354/500000]
+ train/ActionL1Loss=0.1205
+ throughput/total_tokens=451,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2355/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=452,160,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2356/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=452,352,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2357/500000]
+ train/ActionL1Loss=0.1151
+ throughput/total_tokens=452,544,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2358/500000]
+ train/ActionL1Loss=0.1109
+ throughput/total_tokens=452,736,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2359/500000]
+ train/ActionL1Loss=0.1124
+ throughput/total_tokens=452,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2360/500000]
+ optim/total_grad_norm=6.147
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=453,120,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2361/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=453,312,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2362/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=453,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2363/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=453,696,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2364/500000]
+ train/ActionL1Loss=0.1026
+ throughput/total_tokens=453,888,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2365/500000]
+ train/ActionL1Loss=0.1258
+ throughput/total_tokens=454,080,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2366/500000]
+ train/ActionL1Loss=0.1036
+ throughput/total_tokens=454,272,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2367/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=454,464,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2368/500000]
+ train/ActionL1Loss=0.0947
+ throughput/total_tokens=454,656,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2369/500000]
+ train/ActionL1Loss=0.1022
+ throughput/total_tokens=454,848,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2370/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=455,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2371/500000]
+ train/ActionL1Loss=0.1266
+ throughput/total_tokens=455,232,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2372/500000]
+ train/ActionL1Loss=0.1106
+ throughput/total_tokens=455,424,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2373/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=455,616,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2374/500000]
+ train/ActionL1Loss=0.1034
+ throughput/total_tokens=455,808,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2375/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=456,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2376/500000]
+ train/ActionL1Loss=0.1113
+ throughput/total_tokens=456,192,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2377/500000]
+ train/ActionL1Loss=0.1308
+ throughput/total_tokens=456,384,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2378/500000]
+ train/ActionL1Loss=0.1093
+ throughput/total_tokens=456,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2379/500000]
+ train/ActionL1Loss=0.1110
+ throughput/total_tokens=456,768,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2380/500000]
+ optim/total_grad_norm=8.316
+ train/ActionL1Loss=0.0972
+ throughput/total_tokens=456,960,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2381/500000]
+ train/ActionL1Loss=0.1046
+ throughput/total_tokens=457,152,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2382/500000]
+ train/ActionL1Loss=0.1135
+ throughput/total_tokens=457,344,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2383/500000]
+ train/ActionL1Loss=0.1194
+ throughput/total_tokens=457,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2384/500000]
+ train/ActionL1Loss=0.1189
+ throughput/total_tokens=457,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2385/500000]
+ train/ActionL1Loss=0.1028
+ throughput/total_tokens=457,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2386/500000]
+ train/ActionL1Loss=0.1140
+ throughput/total_tokens=458,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2387/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=458,304,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2388/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=458,496,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2389/500000]
+ train/ActionL1Loss=0.1285
+ throughput/total_tokens=458,688,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2390/500000]
+ train/ActionL1Loss=0.1186
+ throughput/total_tokens=458,880,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2391/500000]
+ train/ActionL1Loss=0.1077
+ throughput/total_tokens=459,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2392/500000]
+ train/ActionL1Loss=0.1209
+ throughput/total_tokens=459,264,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2393/500000]
+ train/ActionL1Loss=0.0998
+ throughput/total_tokens=459,456,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2394/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=459,648,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2395/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=459,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2396/500000]
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=460,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2397/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=460,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2398/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=460,416,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2399/500000]
+ train/ActionL1Loss=0.1042
+ throughput/total_tokens=460,608,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2400/500000]
+ optim/total_grad_norm=6.326
+ train/ActionL1Loss=0.0887
+ throughput/total_tokens=460,800,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2401/500000]
+ train/ActionL1Loss=0.1030
+ throughput/total_tokens=460,992,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2402/500000]
+ train/ActionL1Loss=0.1030
+ throughput/total_tokens=461,184,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2403/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=461,376,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2404/500000]
+ train/ActionL1Loss=0.0943
+ throughput/total_tokens=461,568,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2405/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=461,760,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2406/500000]
+ train/ActionL1Loss=0.0979
+ throughput/total_tokens=461,952,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2407/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=462,144,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2408/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=462,336,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2409/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=462,528,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2410/500000]
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=462,720,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2411/500000]
+ train/ActionL1Loss=0.0982
+ throughput/total_tokens=462,912,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2412/500000]
+ train/ActionL1Loss=0.1175
+ throughput/total_tokens=463,104,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2413/500000]
+ train/ActionL1Loss=0.1100
+ throughput/total_tokens=463,296,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2414/500000]
+ train/ActionL1Loss=0.1080
+ throughput/total_tokens=463,488,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2415/500000]
+ train/ActionL1Loss=0.1055
+ throughput/total_tokens=463,680,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2416/500000]
+ train/ActionL1Loss=0.1012
+ throughput/total_tokens=463,872,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2417/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=464,064,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2418/500000]
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=464,256,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2419/500000]
+ train/ActionL1Loss=0.1004
+ throughput/total_tokens=464,448,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2420/500000]
+ optim/total_grad_norm=6.530
+ train/ActionL1Loss=0.1097
+ throughput/total_tokens=464,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2421/500000]
+ train/ActionL1Loss=0.1114
+ throughput/total_tokens=464,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2422/500000]
+ train/ActionL1Loss=0.1184
+ throughput/total_tokens=465,024,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2423/500000]
+ train/ActionL1Loss=0.1033
+ throughput/total_tokens=465,216,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2424/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=465,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2425/500000]
+ train/ActionL1Loss=0.1016
+ throughput/total_tokens=465,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2426/500000]
+ train/ActionL1Loss=0.1162
+ throughput/total_tokens=465,792,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2427/500000]
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=465,984,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2428/500000]
+ train/ActionL1Loss=0.1177
+ throughput/total_tokens=466,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2429/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=466,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2430/500000]
+ train/ActionL1Loss=0.0996
+ throughput/total_tokens=466,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2431/500000]
+ train/ActionL1Loss=0.1187
+ throughput/total_tokens=466,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2432/500000]
+ train/ActionL1Loss=0.1015
+ throughput/total_tokens=466,944,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=2433/500000]
+ train/ActionL1Loss=0.1003
+ throughput/total_tokens=467,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2434/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=467,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2435/500000]
+ train/ActionL1Loss=0.1247
+ throughput/total_tokens=467,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2436/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=467,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2437/500000]
+ train/ActionL1Loss=0.1040
+ throughput/total_tokens=467,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2438/500000]
+ train/ActionL1Loss=0.1082
+ throughput/total_tokens=468,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2439/500000]
+ train/ActionL1Loss=0.1084
+ throughput/total_tokens=468,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2440/500000]
+ optim/total_grad_norm=6.589
+ train/ActionL1Loss=0.1016
+ throughput/total_tokens=468,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2441/500000]
+ train/ActionL1Loss=0.0961
+ throughput/total_tokens=468,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2442/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=468,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2443/500000]
+ train/ActionL1Loss=0.1010
+ throughput/total_tokens=469,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2444/500000]
+ train/ActionL1Loss=0.0985
+ throughput/total_tokens=469,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2445/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=469,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2446/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=469,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2447/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=469,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2448/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=470,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2449/500000]
+ train/ActionL1Loss=0.0941
+ throughput/total_tokens=470,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2450/500000]
+ train/ActionL1Loss=0.1101
+ throughput/total_tokens=470,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2451/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=470,592,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2452/500000]
+ train/ActionL1Loss=0.1136
+ throughput/total_tokens=470,784,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2453/500000]
+ train/ActionL1Loss=0.1182
+ throughput/total_tokens=470,976,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2454/500000]
+ train/ActionL1Loss=0.1091
+ throughput/total_tokens=471,168,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2455/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=471,360,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2456/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=471,552,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2457/500000]
+ train/ActionL1Loss=0.1061
+ throughput/total_tokens=471,744,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2458/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=471,936,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2459/500000]
+ train/ActionL1Loss=0.0948
+ throughput/total_tokens=472,128,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2460/500000]
+ optim/total_grad_norm=6.485
+ train/ActionL1Loss=0.1007
+ throughput/total_tokens=472,320,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2461/500000]
+ train/ActionL1Loss=0.0877
+ throughput/total_tokens=472,512,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2462/500000]
+ train/ActionL1Loss=0.1176
+ throughput/total_tokens=472,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2463/500000]
+ train/ActionL1Loss=0.1131
+ throughput/total_tokens=472,896,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2464/500000]
+ train/ActionL1Loss=0.1025
+ throughput/total_tokens=473,088,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2465/500000]
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=473,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2466/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=473,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2467/500000]
+ train/ActionL1Loss=0.1206
+ throughput/total_tokens=473,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2468/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=473,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2469/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=474,048,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2470/500000]
+ train/ActionL1Loss=0.0984
+ throughput/total_tokens=474,240,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2471/500000]
+ train/ActionL1Loss=0.0903
+ throughput/total_tokens=474,432,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2472/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=474,624,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2473/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=474,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2474/500000]
+ train/ActionL1Loss=0.1025
+ throughput/total_tokens=475,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2475/500000]
+ train/ActionL1Loss=0.1013
+ throughput/total_tokens=475,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2476/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=475,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2477/500000]
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=475,584,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2478/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=475,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2479/500000]
+ train/ActionL1Loss=0.0887
+ throughput/total_tokens=475,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2480/500000]
+ optim/total_grad_norm=7.005
+ train/ActionL1Loss=0.0983
+ throughput/total_tokens=476,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2481/500000]
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=476,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2482/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=476,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2483/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=476,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2484/500000]
+ train/ActionL1Loss=0.0935
+ throughput/total_tokens=476,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2485/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=477,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2486/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=477,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2487/500000]
+ train/ActionL1Loss=0.0957
+ throughput/total_tokens=477,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2488/500000]
+ train/ActionL1Loss=0.0943
+ throughput/total_tokens=477,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2489/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=477,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2490/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=478,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2491/500000]
+ train/ActionL1Loss=0.1065
+ throughput/total_tokens=478,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2492/500000]
+ train/ActionL1Loss=0.1010
+ throughput/total_tokens=478,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2493/500000]
+ train/ActionL1Loss=0.1181
+ throughput/total_tokens=478,656,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2494/500000]
+ train/ActionL1Loss=0.1035
+ throughput/total_tokens=478,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2495/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=479,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2496/500000]
+ train/ActionL1Loss=0.1041
+ throughput/total_tokens=479,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2497/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=479,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2498/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=479,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2499/500000]
+ train/ActionL1Loss=0.1007
+ throughput/total_tokens=479,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2500/500000]
+ optim/total_grad_norm=8.205
+ train/ActionL1Loss=0.1049
+ throughput/total_tokens=480,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+10/01 [03:56:32] INFO | >> Saving config... checkpoint.py:608
+10/01 [03:57:01] INFO | >> Saving model state... checkpoint.py:796
+10/01 [03:58:12] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [03:59:44] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=2501/500000]
+ train/ActionL1Loss=0.1201
+ throughput/total_tokens=480,192,000
+ throughput/device/tokens_per_second=1,211
+ throughput/device/batches_per_second=0.0505
+[step=2502/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=480,384,000
+ throughput/device/tokens_per_second=1,185
+ throughput/device/batches_per_second=0.0494
+[step=2503/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=480,576,000
+ throughput/device/tokens_per_second=1,176
+ throughput/device/batches_per_second=0.0490
+[step=2504/500000]
+ train/ActionL1Loss=0.1051
+ throughput/total_tokens=480,768,000
+ throughput/device/tokens_per_second=1,174
+ throughput/device/batches_per_second=0.0489
+[step=2505/500000]
+ train/ActionL1Loss=0.1016
+ throughput/total_tokens=480,960,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0488
+[step=2506/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=481,152,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=2507/500000]
+ train/ActionL1Loss=0.1212
+ throughput/total_tokens=481,344,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=2508/500000]
+ train/ActionL1Loss=0.0981
+ throughput/total_tokens=481,536,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=2509/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=481,728,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=2510/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=481,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=2511/500000]
+ train/ActionL1Loss=0.1039
+ throughput/total_tokens=482,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=2512/500000]
+ train/ActionL1Loss=0.0933
+ throughput/total_tokens=482,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=2513/500000]
+ train/ActionL1Loss=0.1000
+ throughput/total_tokens=482,496,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=2514/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=482,688,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=2515/500000]
+ train/ActionL1Loss=0.0836
+ throughput/total_tokens=482,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=2516/500000]
+ train/ActionL1Loss=0.1051
+ throughput/total_tokens=483,072,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=2517/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=483,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=2518/500000]
+ train/ActionL1Loss=0.0957
+ throughput/total_tokens=483,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=2519/500000]
+ train/ActionL1Loss=0.0945
+ throughput/total_tokens=483,648,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2520/500000]
+ optim/total_grad_norm=7.614
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=483,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2521/500000]
+ train/ActionL1Loss=0.1188
+ throughput/total_tokens=484,032,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2522/500000]
+ train/ActionL1Loss=0.1043
+ throughput/total_tokens=484,224,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2523/500000]
+ train/ActionL1Loss=0.1001
+ throughput/total_tokens=484,416,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2524/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=484,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2525/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=484,800,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2526/500000]
+ train/ActionL1Loss=0.1058
+ throughput/total_tokens=484,992,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2527/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=485,184,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2528/500000]
+ train/ActionL1Loss=0.0900
+ throughput/total_tokens=485,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2529/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=485,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2530/500000]
+ train/ActionL1Loss=0.1178
+ throughput/total_tokens=485,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2531/500000]
+ train/ActionL1Loss=0.0901
+ throughput/total_tokens=485,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2532/500000]
+ train/ActionL1Loss=0.0947
+ throughput/total_tokens=486,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2533/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=486,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2534/500000]
+ train/ActionL1Loss=0.0982
+ throughput/total_tokens=486,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2535/500000]
+ train/ActionL1Loss=0.0991
+ throughput/total_tokens=486,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2536/500000]
+ train/ActionL1Loss=0.1069
+ throughput/total_tokens=486,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2537/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=487,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2538/500000]
+ train/ActionL1Loss=0.0987
+ throughput/total_tokens=487,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2539/500000]
+ train/ActionL1Loss=0.0983
+ throughput/total_tokens=487,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2540/500000]
+ optim/total_grad_norm=5.796
+ train/ActionL1Loss=0.1192
+ throughput/total_tokens=487,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2541/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=487,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2542/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=488,064,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2543/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=488,256,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2544/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=488,448,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2545/500000]
+ train/ActionL1Loss=0.1167
+ throughput/total_tokens=488,640,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2546/500000]
+ train/ActionL1Loss=0.1306
+ throughput/total_tokens=488,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2547/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=489,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2548/500000]
+ train/ActionL1Loss=0.0964
+ throughput/total_tokens=489,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2549/500000]
+ train/ActionL1Loss=0.1026
+ throughput/total_tokens=489,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2550/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=489,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2551/500000]
+ train/ActionL1Loss=0.1130
+ throughput/total_tokens=489,792,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2552/500000]
+ train/ActionL1Loss=0.0979
+ throughput/total_tokens=489,984,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2553/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=490,176,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2554/500000]
+ train/ActionL1Loss=0.0983
+ throughput/total_tokens=490,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2555/500000]
+ train/ActionL1Loss=0.0960
+ throughput/total_tokens=490,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2556/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=490,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2557/500000]
+ train/ActionL1Loss=0.1013
+ throughput/total_tokens=490,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2558/500000]
+ train/ActionL1Loss=0.0995
+ throughput/total_tokens=491,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2559/500000]
+ train/ActionL1Loss=0.1014
+ throughput/total_tokens=491,328,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2560/500000]
+ optim/total_grad_norm=5.827
+ train/ActionL1Loss=0.0981
+ throughput/total_tokens=491,520,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2561/500000]
+ train/ActionL1Loss=0.1049
+ throughput/total_tokens=491,712,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2562/500000]
+ train/ActionL1Loss=0.1025
+ throughput/total_tokens=491,904,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2563/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=492,096,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2564/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=492,288,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2565/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=492,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2566/500000]
+ train/ActionL1Loss=0.0952
+ throughput/total_tokens=492,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2567/500000]
+ train/ActionL1Loss=0.1109
+ throughput/total_tokens=492,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2568/500000]
+ train/ActionL1Loss=0.1063
+ throughput/total_tokens=493,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2569/500000]
+ train/ActionL1Loss=0.0921
+ throughput/total_tokens=493,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2570/500000]
+ train/ActionL1Loss=0.1112
+ throughput/total_tokens=493,440,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2571/500000]
+ train/ActionL1Loss=0.0975
+ throughput/total_tokens=493,632,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2572/500000]
+ train/ActionL1Loss=0.0917
+ throughput/total_tokens=493,824,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2573/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=494,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2574/500000]
+ train/ActionL1Loss=0.0951
+ throughput/total_tokens=494,208,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2575/500000]
+ train/ActionL1Loss=0.1098
+ throughput/total_tokens=494,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2576/500000]
+ train/ActionL1Loss=0.1221
+ throughput/total_tokens=494,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2577/500000]
+ train/ActionL1Loss=0.1220
+ throughput/total_tokens=494,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2578/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=494,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2579/500000]
+ train/ActionL1Loss=0.0991
+ throughput/total_tokens=495,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2580/500000]
+ optim/total_grad_norm=6.464
+ train/ActionL1Loss=0.1159
+ throughput/total_tokens=495,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2581/500000]
+ train/ActionL1Loss=0.1135
+ throughput/total_tokens=495,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2582/500000]
+ train/ActionL1Loss=0.1261
+ throughput/total_tokens=495,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=2583/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=495,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=2584/500000]
+ train/ActionL1Loss=0.0994
+ throughput/total_tokens=496,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=2585/500000]
+ train/ActionL1Loss=0.1112
+ throughput/total_tokens=496,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2586/500000]
+ train/ActionL1Loss=0.1048
+ throughput/total_tokens=496,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2587/500000]
+ train/ActionL1Loss=0.1071
+ throughput/total_tokens=496,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2588/500000]
+ train/ActionL1Loss=0.1180
+ throughput/total_tokens=496,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2589/500000]
+ train/ActionL1Loss=0.1240
+ throughput/total_tokens=497,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2590/500000]
+ train/ActionL1Loss=0.0947
+ throughput/total_tokens=497,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2591/500000]
+ train/ActionL1Loss=0.1091
+ throughput/total_tokens=497,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2592/500000]
+ train/ActionL1Loss=0.1080
+ throughput/total_tokens=497,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2593/500000]
+ train/ActionL1Loss=0.1034
+ throughput/total_tokens=497,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2594/500000]
+ train/ActionL1Loss=0.1084
+ throughput/total_tokens=498,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2595/500000]
+ train/ActionL1Loss=0.1156
+ throughput/total_tokens=498,240,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2596/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=498,432,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2597/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=498,624,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2598/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=498,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2599/500000]
+ train/ActionL1Loss=0.1186
+ throughput/total_tokens=499,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2600/500000]
+ optim/total_grad_norm=7.044
+ train/ActionL1Loss=0.1001
+ throughput/total_tokens=499,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2601/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=499,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2602/500000]
+ train/ActionL1Loss=0.1016
+ throughput/total_tokens=499,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2603/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=499,776,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2604/500000]
+ train/ActionL1Loss=0.1050
+ throughput/total_tokens=499,968,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2605/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=500,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2606/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=500,352,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2607/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=500,544,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2608/500000]
+ train/ActionL1Loss=0.1082
+ throughput/total_tokens=500,736,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2609/500000]
+ train/ActionL1Loss=0.1149
+ throughput/total_tokens=500,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2610/500000]
+ train/ActionL1Loss=0.0880
+ throughput/total_tokens=501,120,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2611/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=501,312,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2612/500000]
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=501,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2613/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=501,696,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2614/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=501,888,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2615/500000]
+ train/ActionL1Loss=0.0844
+ throughput/total_tokens=502,080,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2616/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=502,272,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2617/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=502,464,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2618/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=502,656,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2619/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=502,848,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2620/500000]
+ optim/total_grad_norm=6.395
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=503,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2621/500000]
+ train/ActionL1Loss=0.0912
+ throughput/total_tokens=503,232,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2622/500000]
+ train/ActionL1Loss=0.1093
+ throughput/total_tokens=503,424,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2623/500000]
+ train/ActionL1Loss=0.1223
+ throughput/total_tokens=503,616,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2624/500000]
+ train/ActionL1Loss=0.0943
+ throughput/total_tokens=503,808,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2625/500000]
+ train/ActionL1Loss=0.1006
+ throughput/total_tokens=504,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2626/500000]
+ train/ActionL1Loss=0.1088
+ throughput/total_tokens=504,192,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2627/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=504,384,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2628/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=504,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2629/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=504,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2630/500000]
+ train/ActionL1Loss=0.1036
+ throughput/total_tokens=504,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2631/500000]
+ train/ActionL1Loss=0.1194
+ throughput/total_tokens=505,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2632/500000]
+ train/ActionL1Loss=0.1146
+ throughput/total_tokens=505,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2633/500000]
+ train/ActionL1Loss=0.1045
+ throughput/total_tokens=505,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2634/500000]
+ train/ActionL1Loss=0.0901
+ throughput/total_tokens=505,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2635/500000]
+ train/ActionL1Loss=0.1002
+ throughput/total_tokens=505,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2636/500000]
+ train/ActionL1Loss=0.1082
+ throughput/total_tokens=506,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2637/500000]
+ train/ActionL1Loss=0.0852
+ throughput/total_tokens=506,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2638/500000]
+ train/ActionL1Loss=0.1094
+ throughput/total_tokens=506,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2639/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=506,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2640/500000]
+ optim/total_grad_norm=8.300
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=506,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2641/500000]
+ train/ActionL1Loss=0.0955
+ throughput/total_tokens=507,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2642/500000]
+ train/ActionL1Loss=0.1047
+ throughput/total_tokens=507,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2643/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=507,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2644/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=507,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2645/500000]
+ train/ActionL1Loss=0.0930
+ throughput/total_tokens=507,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2646/500000]
+ train/ActionL1Loss=0.1043
+ throughput/total_tokens=508,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2647/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=508,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2648/500000]
+ train/ActionL1Loss=0.1059
+ throughput/total_tokens=508,416,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2649/500000]
+ train/ActionL1Loss=0.1057
+ throughput/total_tokens=508,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2650/500000]
+ train/ActionL1Loss=0.0989
+ throughput/total_tokens=508,800,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2651/500000]
+ train/ActionL1Loss=0.1196
+ throughput/total_tokens=508,992,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2652/500000]
+ train/ActionL1Loss=0.1181
+ throughput/total_tokens=509,184,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2653/500000]
+ train/ActionL1Loss=0.1110
+ throughput/total_tokens=509,376,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2654/500000]
+ train/ActionL1Loss=0.1014
+ throughput/total_tokens=509,568,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2655/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=509,760,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2656/500000]
+ train/ActionL1Loss=0.1028
+ throughput/total_tokens=509,952,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2657/500000]
+ train/ActionL1Loss=0.0866
+ throughput/total_tokens=510,144,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2658/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=510,336,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2659/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=510,528,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2660/500000]
+ optim/total_grad_norm=7.744
+ train/ActionL1Loss=0.0933
+ throughput/total_tokens=510,720,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=2661/500000]
+ train/ActionL1Loss=0.0964
+ throughput/total_tokens=510,912,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2662/500000]
+ train/ActionL1Loss=0.1042
+ throughput/total_tokens=511,104,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2663/500000]
+ train/ActionL1Loss=0.1023
+ throughput/total_tokens=511,296,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2664/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=511,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2665/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=511,680,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2666/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=511,872,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2667/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=512,064,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2668/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=512,256,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2669/500000]
+ train/ActionL1Loss=0.1081
+ throughput/total_tokens=512,448,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2670/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=512,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2671/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=512,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2672/500000]
+ train/ActionL1Loss=0.0836
+ throughput/total_tokens=513,024,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2673/500000]
+ train/ActionL1Loss=0.0934
+ throughput/total_tokens=513,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2674/500000]
+ train/ActionL1Loss=0.1059
+ throughput/total_tokens=513,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2675/500000]
+ train/ActionL1Loss=0.1044
+ throughput/total_tokens=513,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2676/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=513,792,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2677/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=513,984,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2678/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=514,176,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2679/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=514,368,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2680/500000]
+ optim/total_grad_norm=5.150
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=514,560,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2681/500000]
+ train/ActionL1Loss=0.1164
+ throughput/total_tokens=514,752,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2682/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=514,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2683/500000]
+ train/ActionL1Loss=0.1025
+ throughput/total_tokens=515,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2684/500000]
+ train/ActionL1Loss=0.1134
+ throughput/total_tokens=515,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2685/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=515,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2686/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=515,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2687/500000]
+ train/ActionL1Loss=0.1142
+ throughput/total_tokens=515,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2688/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=516,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2689/500000]
+ train/ActionL1Loss=0.1011
+ throughput/total_tokens=516,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2690/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=516,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2691/500000]
+ train/ActionL1Loss=0.0975
+ throughput/total_tokens=516,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2692/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=516,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2693/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=517,056,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2694/500000]
+ train/ActionL1Loss=0.0916
+ throughput/total_tokens=517,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2695/500000]
+ train/ActionL1Loss=0.1090
+ throughput/total_tokens=517,440,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2696/500000]
+ train/ActionL1Loss=0.0955
+ throughput/total_tokens=517,632,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2697/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=517,824,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2698/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=518,016,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2699/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=518,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2700/500000]
+ optim/total_grad_norm=5.884
+ train/ActionL1Loss=0.0948
+ throughput/total_tokens=518,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2701/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=518,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2702/500000]
+ train/ActionL1Loss=0.1015
+ throughput/total_tokens=518,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2703/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=518,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2704/500000]
+ train/ActionL1Loss=0.0901
+ throughput/total_tokens=519,168,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2705/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=519,360,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2706/500000]
+ train/ActionL1Loss=0.1047
+ throughput/total_tokens=519,552,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2707/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=519,744,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2708/500000]
+ train/ActionL1Loss=0.1208
+ throughput/total_tokens=519,936,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2709/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=520,128,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2710/500000]
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=520,320,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2711/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=520,512,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2712/500000]
+ train/ActionL1Loss=0.1157
+ throughput/total_tokens=520,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2713/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=520,896,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2714/500000]
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=521,088,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2715/500000]
+ train/ActionL1Loss=0.1082
+ throughput/total_tokens=521,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2716/500000]
+ train/ActionL1Loss=0.1143
+ throughput/total_tokens=521,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2717/500000]
+ train/ActionL1Loss=0.0994
+ throughput/total_tokens=521,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2718/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=521,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2719/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=522,048,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2720/500000]
+ optim/total_grad_norm=6.687
+ train/ActionL1Loss=0.1164
+ throughput/total_tokens=522,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2721/500000]
+ train/ActionL1Loss=0.1004
+ throughput/total_tokens=522,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2722/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=522,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2723/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=522,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2724/500000]
+ train/ActionL1Loss=0.1049
+ throughput/total_tokens=523,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2725/500000]
+ train/ActionL1Loss=0.1102
+ throughput/total_tokens=523,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2726/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=523,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2727/500000]
+ train/ActionL1Loss=0.1034
+ throughput/total_tokens=523,584,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2728/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=523,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2729/500000]
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=523,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2730/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=524,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2731/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=524,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2732/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=524,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2733/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=524,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2734/500000]
+ train/ActionL1Loss=0.1041
+ throughput/total_tokens=524,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2735/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=525,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2736/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=525,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2737/500000]
+ train/ActionL1Loss=0.0916
+ throughput/total_tokens=525,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2738/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=525,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2739/500000]
+ train/ActionL1Loss=0.1090
+ throughput/total_tokens=525,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2740/500000]
+ optim/total_grad_norm=7.723
+ train/ActionL1Loss=0.0961
+ throughput/total_tokens=526,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2741/500000]
+ train/ActionL1Loss=0.1123
+ throughput/total_tokens=526,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2742/500000]
+ train/ActionL1Loss=0.1049
+ throughput/total_tokens=526,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2743/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=526,656,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2744/500000]
+ train/ActionL1Loss=0.1089
+ throughput/total_tokens=526,848,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2745/500000]
+ train/ActionL1Loss=0.0976
+ throughput/total_tokens=527,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2746/500000]
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=527,232,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2747/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=527,424,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2748/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=527,616,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2749/500000]
+ train/ActionL1Loss=0.1046
+ throughput/total_tokens=527,808,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2750/500000]
+ train/ActionL1Loss=0.0952
+ throughput/total_tokens=528,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2751/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=528,192,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2752/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=528,384,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2753/500000]
+ train/ActionL1Loss=0.0930
+ throughput/total_tokens=528,576,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2754/500000]
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=528,768,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2755/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=528,960,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2756/500000]
+ train/ActionL1Loss=0.1265
+ throughput/total_tokens=529,152,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2757/500000]
+ train/ActionL1Loss=0.1178
+ throughput/total_tokens=529,344,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2758/500000]
+ train/ActionL1Loss=0.0868
+ throughput/total_tokens=529,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2759/500000]
+ train/ActionL1Loss=0.1059
+ throughput/total_tokens=529,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2760/500000]
+ optim/total_grad_norm=6.502
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=529,920,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2761/500000]
+ train/ActionL1Loss=0.1011
+ throughput/total_tokens=530,112,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2762/500000]
+ train/ActionL1Loss=0.1167
+ throughput/total_tokens=530,304,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2763/500000]
+ train/ActionL1Loss=0.1030
+ throughput/total_tokens=530,496,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2764/500000]
+ train/ActionL1Loss=0.1038
+ throughput/total_tokens=530,688,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2765/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=530,880,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2766/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=531,072,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2767/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=531,264,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2768/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=531,456,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2769/500000]
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=531,648,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2770/500000]
+ train/ActionL1Loss=0.1127
+ throughput/total_tokens=531,840,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2771/500000]
+ train/ActionL1Loss=0.0943
+ throughput/total_tokens=532,032,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2772/500000]
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=532,224,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2773/500000]
+ train/ActionL1Loss=0.0985
+ throughput/total_tokens=532,416,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2774/500000]
+ train/ActionL1Loss=0.0921
+ throughput/total_tokens=532,608,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2775/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=532,800,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2776/500000]
+ train/ActionL1Loss=0.1236
+ throughput/total_tokens=532,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2777/500000]
+ train/ActionL1Loss=0.1065
+ throughput/total_tokens=533,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2778/500000]
+ train/ActionL1Loss=0.1056
+ throughput/total_tokens=533,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2779/500000]
+ train/ActionL1Loss=0.1076
+ throughput/total_tokens=533,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2780/500000]
+ optim/total_grad_norm=5.569
+ train/ActionL1Loss=0.1082
+ throughput/total_tokens=533,760,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2781/500000]
+ train/ActionL1Loss=0.1184
+ throughput/total_tokens=533,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2782/500000]
+ train/ActionL1Loss=0.0781
+ throughput/total_tokens=534,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2783/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=534,336,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2784/500000]
+ train/ActionL1Loss=0.0930
+ throughput/total_tokens=534,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2785/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=534,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2786/500000]
+ train/ActionL1Loss=0.1116
+ throughput/total_tokens=534,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2787/500000]
+ train/ActionL1Loss=0.1175
+ throughput/total_tokens=535,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2788/500000]
+ train/ActionL1Loss=0.1126
+ throughput/total_tokens=535,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2789/500000]
+ train/ActionL1Loss=0.1034
+ throughput/total_tokens=535,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2790/500000]
+ train/ActionL1Loss=0.0919
+ throughput/total_tokens=535,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2791/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=535,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2792/500000]
+ train/ActionL1Loss=0.0930
+ throughput/total_tokens=536,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2793/500000]
+ train/ActionL1Loss=0.0880
+ throughput/total_tokens=536,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2794/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=536,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2795/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=536,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2796/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=536,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2797/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=537,024,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2798/500000]
+ train/ActionL1Loss=0.0975
+ throughput/total_tokens=537,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2799/500000]
+ train/ActionL1Loss=0.1079
+ throughput/total_tokens=537,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2800/500000]
+ optim/total_grad_norm=5.507
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=537,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2801/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=537,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2802/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=537,984,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2803/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=538,176,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2804/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=538,368,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2805/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=538,560,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2806/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=538,752,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2807/500000]
+ train/ActionL1Loss=0.0880
+ throughput/total_tokens=538,944,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2808/500000]
+ train/ActionL1Loss=0.1193
+ throughput/total_tokens=539,136,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2809/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=539,328,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2810/500000]
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=539,520,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2811/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=539,712,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2812/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=539,904,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2813/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=540,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2814/500000]
+ train/ActionL1Loss=0.1079
+ throughput/total_tokens=540,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2815/500000]
+ train/ActionL1Loss=0.1048
+ throughput/total_tokens=540,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2816/500000]
+ train/ActionL1Loss=0.1028
+ throughput/total_tokens=540,672,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2817/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=540,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2818/500000]
+ train/ActionL1Loss=0.1047
+ throughput/total_tokens=541,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2819/500000]
+ train/ActionL1Loss=0.1063
+ throughput/total_tokens=541,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2820/500000]
+ optim/total_grad_norm=6.334
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=541,440,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2821/500000]
+ train/ActionL1Loss=0.1068
+ throughput/total_tokens=541,632,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2822/500000]
+ train/ActionL1Loss=0.1026
+ throughput/total_tokens=541,824,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2823/500000]
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=542,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2824/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=542,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2825/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=542,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2826/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=542,592,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2827/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=542,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2828/500000]
+ train/ActionL1Loss=0.1132
+ throughput/total_tokens=542,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2829/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=543,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2830/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=543,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2831/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=543,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2832/500000]
+ train/ActionL1Loss=0.0998
+ throughput/total_tokens=543,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2833/500000]
+ train/ActionL1Loss=0.1090
+ throughput/total_tokens=543,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2834/500000]
+ train/ActionL1Loss=0.1069
+ throughput/total_tokens=544,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2835/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=544,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2836/500000]
+ train/ActionL1Loss=0.1127
+ throughput/total_tokens=544,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2837/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=544,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2838/500000]
+ train/ActionL1Loss=0.1126
+ throughput/total_tokens=544,896,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2839/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=545,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2840/500000]
+ optim/total_grad_norm=6.413
+ train/ActionL1Loss=0.1156
+ throughput/total_tokens=545,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2841/500000]
+ train/ActionL1Loss=0.1051
+ throughput/total_tokens=545,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2842/500000]
+ train/ActionL1Loss=0.1093
+ throughput/total_tokens=545,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2843/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=545,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2844/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=546,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2845/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=546,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2846/500000]
+ train/ActionL1Loss=0.0897
+ throughput/total_tokens=546,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2847/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=546,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2848/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=546,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2849/500000]
+ train/ActionL1Loss=0.1164
+ throughput/total_tokens=547,008,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2850/500000]
+ train/ActionL1Loss=0.1141
+ throughput/total_tokens=547,200,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2851/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=547,392,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2852/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=547,584,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2853/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=547,776,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2854/500000]
+ train/ActionL1Loss=0.1119
+ throughput/total_tokens=547,968,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2855/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=548,160,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2856/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=548,352,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2857/500000]
+ train/ActionL1Loss=0.1053
+ throughput/total_tokens=548,544,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2858/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=548,736,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2859/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=548,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2860/500000]
+ optim/total_grad_norm=6.253
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=549,120,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2861/500000]
+ train/ActionL1Loss=0.1045
+ throughput/total_tokens=549,312,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2862/500000]
+ train/ActionL1Loss=0.0950
+ throughput/total_tokens=549,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2863/500000]
+ train/ActionL1Loss=0.1023
+ throughput/total_tokens=549,696,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2864/500000]
+ train/ActionL1Loss=0.0977
+ throughput/total_tokens=549,888,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2865/500000]
+ train/ActionL1Loss=0.0868
+ throughput/total_tokens=550,080,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2866/500000]
+ train/ActionL1Loss=0.1061
+ throughput/total_tokens=550,272,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2867/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=550,464,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2868/500000]
+ train/ActionL1Loss=0.1011
+ throughput/total_tokens=550,656,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2869/500000]
+ train/ActionL1Loss=0.1053
+ throughput/total_tokens=550,848,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2870/500000]
+ train/ActionL1Loss=0.1120
+ throughput/total_tokens=551,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2871/500000]
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=551,232,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2872/500000]
+ train/ActionL1Loss=0.1009
+ throughput/total_tokens=551,424,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2873/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=551,616,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2874/500000]
+ train/ActionL1Loss=0.1017
+ throughput/total_tokens=551,808,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2875/500000]
+ train/ActionL1Loss=0.1092
+ throughput/total_tokens=552,000,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2876/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=552,192,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2877/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=552,384,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2878/500000]
+ train/ActionL1Loss=0.1015
+ throughput/total_tokens=552,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2879/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=552,768,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2880/500000]
+ optim/total_grad_norm=5.214
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=552,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2881/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=553,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2882/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=553,344,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2883/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=553,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2884/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=553,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2885/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=553,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2886/500000]
+ train/ActionL1Loss=0.0982
+ throughput/total_tokens=554,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2887/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=554,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2888/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=554,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2889/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=554,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2890/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=554,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2891/500000]
+ train/ActionL1Loss=0.0998
+ throughput/total_tokens=555,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2892/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=555,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2893/500000]
+ train/ActionL1Loss=0.0890
+ throughput/total_tokens=555,456,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2894/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=555,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2895/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=555,840,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2896/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=556,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2897/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=556,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2898/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=556,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2899/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=556,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2900/500000]
+ optim/total_grad_norm=5.524
+ train/ActionL1Loss=0.1058
+ throughput/total_tokens=556,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2901/500000]
+ train/ActionL1Loss=0.1136
+ throughput/total_tokens=556,992,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2902/500000]
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=557,184,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2903/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=557,376,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2904/500000]
+ train/ActionL1Loss=0.1016
+ throughput/total_tokens=557,568,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2905/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=557,760,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2906/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=557,952,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=2907/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=558,144,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2908/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=558,336,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2909/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=558,528,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2910/500000]
+ train/ActionL1Loss=0.1168
+ throughput/total_tokens=558,720,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2911/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=558,912,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2912/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=559,104,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2913/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=559,296,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2914/500000]
+ train/ActionL1Loss=0.1011
+ throughput/total_tokens=559,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2915/500000]
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=559,680,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2916/500000]
+ train/ActionL1Loss=0.1126
+ throughput/total_tokens=559,872,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2917/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=560,064,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2918/500000]
+ train/ActionL1Loss=0.0959
+ throughput/total_tokens=560,256,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2919/500000]
+ train/ActionL1Loss=0.1017
+ throughput/total_tokens=560,448,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=2920/500000]
+ optim/total_grad_norm=6.317
+ train/ActionL1Loss=0.0952
+ throughput/total_tokens=560,640,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2921/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=560,832,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2922/500000]
+ train/ActionL1Loss=0.1018
+ throughput/total_tokens=561,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2923/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=561,216,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2924/500000]
+ train/ActionL1Loss=0.0952
+ throughput/total_tokens=561,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2925/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=561,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2926/500000]
+ train/ActionL1Loss=0.0921
+ throughput/total_tokens=561,792,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2927/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=561,984,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2928/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=562,176,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2929/500000]
+ train/ActionL1Loss=0.0948
+ throughput/total_tokens=562,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2930/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=562,560,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2931/500000]
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=562,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2932/500000]
+ train/ActionL1Loss=0.1134
+ throughput/total_tokens=562,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2933/500000]
+ train/ActionL1Loss=0.1144
+ throughput/total_tokens=563,136,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2934/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=563,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2935/500000]
+ train/ActionL1Loss=0.1066
+ throughput/total_tokens=563,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2936/500000]
+ train/ActionL1Loss=0.1043
+ throughput/total_tokens=563,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2937/500000]
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=563,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2938/500000]
+ train/ActionL1Loss=0.0993
+ throughput/total_tokens=564,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2939/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=564,288,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2940/500000]
+ optim/total_grad_norm=6.445
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=564,480,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2941/500000]
+ train/ActionL1Loss=0.1018
+ throughput/total_tokens=564,672,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2942/500000]
+ train/ActionL1Loss=0.0934
+ throughput/total_tokens=564,864,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2943/500000]
+ train/ActionL1Loss=0.0933
+ throughput/total_tokens=565,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2944/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=565,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2945/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=565,440,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2946/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=565,632,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2947/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=565,824,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2948/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=566,016,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2949/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=566,208,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2950/500000]
+ train/ActionL1Loss=0.1171
+ throughput/total_tokens=566,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2951/500000]
+ train/ActionL1Loss=0.1013
+ throughput/total_tokens=566,592,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2952/500000]
+ train/ActionL1Loss=0.1041
+ throughput/total_tokens=566,784,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2953/500000]
+ train/ActionL1Loss=0.0951
+ throughput/total_tokens=566,976,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2954/500000]
+ train/ActionL1Loss=0.1083
+ throughput/total_tokens=567,168,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2955/500000]
+ train/ActionL1Loss=0.1138
+ throughput/total_tokens=567,360,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=2956/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=567,552,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=2957/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=567,744,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2958/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=567,936,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2959/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=568,128,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2960/500000]
+ optim/total_grad_norm=7.066
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=568,320,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2961/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=568,512,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2962/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=568,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2963/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=568,896,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2964/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=569,088,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2965/500000]
+ train/ActionL1Loss=0.1012
+ throughput/total_tokens=569,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2966/500000]
+ train/ActionL1Loss=0.1047
+ throughput/total_tokens=569,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2967/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=569,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2968/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=569,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2969/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=570,048,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2970/500000]
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=570,240,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=2971/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=570,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2972/500000]
+ train/ActionL1Loss=0.1150
+ throughput/total_tokens=570,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2973/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=570,816,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2974/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=571,008,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=2975/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=571,200,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2976/500000]
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=571,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2977/500000]
+ train/ActionL1Loss=0.1044
+ throughput/total_tokens=571,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2978/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=571,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2979/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=571,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2980/500000]
+ optim/total_grad_norm=6.865
+ train/ActionL1Loss=0.0900
+ throughput/total_tokens=572,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2981/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=572,352,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2982/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=572,544,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2983/500000]
+ train/ActionL1Loss=0.0956
+ throughput/total_tokens=572,736,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2984/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=572,928,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=2985/500000]
+ train/ActionL1Loss=0.1043
+ throughput/total_tokens=573,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2986/500000]
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=573,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2987/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=573,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2988/500000]
+ train/ActionL1Loss=0.1009
+ throughput/total_tokens=573,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2989/500000]
+ train/ActionL1Loss=0.1136
+ throughput/total_tokens=573,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2990/500000]
+ train/ActionL1Loss=0.0880
+ throughput/total_tokens=574,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=2991/500000]
+ train/ActionL1Loss=0.0986
+ throughput/total_tokens=574,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2992/500000]
+ train/ActionL1Loss=0.1004
+ throughput/total_tokens=574,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=2993/500000]
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=574,656,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2994/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=574,848,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=2995/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=575,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2996/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=575,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2997/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=575,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2998/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=575,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=2999/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=575,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3000/500000]
+ optim/total_grad_norm=4.844
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=576,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+10/01 [06:53:16] INFO | >> Saving config... checkpoint.py:608
+10/01 [06:53:57] INFO | >> Saving model state... checkpoint.py:796
+10/01 [06:55:05] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [06:56:37] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=3001/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=576,192,000
+ throughput/device/tokens_per_second=1,208
+ throughput/device/batches_per_second=0.0504
+[step=3002/500000]
+ train/ActionL1Loss=0.0892
+ throughput/total_tokens=576,384,000
+ throughput/device/tokens_per_second=1,175
+ throughput/device/batches_per_second=0.0490
+[step=3003/500000]
+ train/ActionL1Loss=0.1075
+ throughput/total_tokens=576,576,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=3004/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=576,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=3005/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=576,960,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=3006/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=577,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3007/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=577,344,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3008/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=577,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3009/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=577,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3010/500000]
+ train/ActionL1Loss=0.0930
+ throughput/total_tokens=577,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3011/500000]
+ train/ActionL1Loss=0.0932
+ throughput/total_tokens=578,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3012/500000]
+ train/ActionL1Loss=0.0989
+ throughput/total_tokens=578,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3013/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=578,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3014/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=578,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3015/500000]
+ train/ActionL1Loss=0.0963
+ throughput/total_tokens=578,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3016/500000]
+ train/ActionL1Loss=0.1036
+ throughput/total_tokens=579,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3017/500000]
+ train/ActionL1Loss=0.0998
+ throughput/total_tokens=579,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3018/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=579,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3019/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=579,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3020/500000]
+ optim/total_grad_norm=8.622
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=579,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3021/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=580,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3022/500000]
+ train/ActionL1Loss=0.0996
+ throughput/total_tokens=580,224,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3023/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=580,416,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3024/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=580,608,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3025/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=580,800,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3026/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=580,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3027/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=581,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3028/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=581,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3029/500000]
+ train/ActionL1Loss=0.1088
+ throughput/total_tokens=581,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3030/500000]
+ train/ActionL1Loss=0.1012
+ throughput/total_tokens=581,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3031/500000]
+ train/ActionL1Loss=0.1038
+ throughput/total_tokens=581,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3032/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=582,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3033/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=582,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3034/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=582,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3035/500000]
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=582,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3036/500000]
+ train/ActionL1Loss=0.1041
+ throughput/total_tokens=582,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3037/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=583,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3038/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=583,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3039/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=583,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3040/500000]
+ optim/total_grad_norm=6.695
+ train/ActionL1Loss=0.0985
+ throughput/total_tokens=583,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3041/500000]
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=583,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3042/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=584,064,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3043/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=584,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3044/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=584,448,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3045/500000]
+ train/ActionL1Loss=0.1226
+ throughput/total_tokens=584,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3046/500000]
+ train/ActionL1Loss=0.1036
+ throughput/total_tokens=584,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3047/500000]
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=585,024,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3048/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=585,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3049/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=585,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3050/500000]
+ train/ActionL1Loss=0.1088
+ throughput/total_tokens=585,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3051/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=585,792,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3052/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=585,984,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3053/500000]
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=586,176,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3054/500000]
+ train/ActionL1Loss=0.1111
+ throughput/total_tokens=586,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3055/500000]
+ train/ActionL1Loss=0.1103
+ throughput/total_tokens=586,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3056/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=586,752,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3057/500000]
+ train/ActionL1Loss=0.0995
+ throughput/total_tokens=586,944,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3058/500000]
+ train/ActionL1Loss=0.0957
+ throughput/total_tokens=587,136,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3059/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=587,328,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3060/500000]
+ optim/total_grad_norm=7.434
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=587,520,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3061/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=587,712,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3062/500000]
+ train/ActionL1Loss=0.0941
+ throughput/total_tokens=587,904,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3063/500000]
+ train/ActionL1Loss=0.1033
+ throughput/total_tokens=588,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3064/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=588,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3065/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=588,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3066/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=588,672,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3067/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=588,864,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3068/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=589,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3069/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=589,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3070/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=589,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3071/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=589,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3072/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=589,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3073/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=590,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3074/500000]
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=590,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3075/500000]
+ train/ActionL1Loss=0.0948
+ throughput/total_tokens=590,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3076/500000]
+ train/ActionL1Loss=0.1000
+ throughput/total_tokens=590,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3077/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=590,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3078/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=590,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3079/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=591,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3080/500000]
+ optim/total_grad_norm=6.495
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=591,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3081/500000]
+ train/ActionL1Loss=0.0987
+ throughput/total_tokens=591,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3082/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=591,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3083/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=591,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3084/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=592,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3085/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=592,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3086/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=592,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3087/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=592,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3088/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=592,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3089/500000]
+ train/ActionL1Loss=0.0955
+ throughput/total_tokens=593,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3090/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=593,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3091/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=593,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3092/500000]
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=593,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3093/500000]
+ train/ActionL1Loss=0.1149
+ throughput/total_tokens=593,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3094/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=594,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3095/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=594,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3096/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=594,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3097/500000]
+ train/ActionL1Loss=0.0963
+ throughput/total_tokens=594,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3098/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=594,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3099/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=595,008,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3100/500000]
+ optim/total_grad_norm=5.678
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=595,200,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3101/500000]
+ train/ActionL1Loss=0.1046
+ throughput/total_tokens=595,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3102/500000]
+ train/ActionL1Loss=0.0953
+ throughput/total_tokens=595,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3103/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=595,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3104/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=595,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3105/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=596,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3106/500000]
+ train/ActionL1Loss=0.1061
+ throughput/total_tokens=596,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3107/500000]
+ train/ActionL1Loss=0.1051
+ throughput/total_tokens=596,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3108/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=596,736,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3109/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=596,928,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3110/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=597,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3111/500000]
+ train/ActionL1Loss=0.1040
+ throughput/total_tokens=597,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3112/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=597,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3113/500000]
+ train/ActionL1Loss=0.0979
+ throughput/total_tokens=597,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3114/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=597,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3115/500000]
+ train/ActionL1Loss=0.0953
+ throughput/total_tokens=598,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3116/500000]
+ train/ActionL1Loss=0.0951
+ throughput/total_tokens=598,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3117/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=598,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3118/500000]
+ train/ActionL1Loss=0.1149
+ throughput/total_tokens=598,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3119/500000]
+ train/ActionL1Loss=0.1025
+ throughput/total_tokens=598,848,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3120/500000]
+ optim/total_grad_norm=7.251
+ train/ActionL1Loss=0.1032
+ throughput/total_tokens=599,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3121/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=599,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3122/500000]
+ train/ActionL1Loss=0.0917
+ throughput/total_tokens=599,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3123/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=599,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3124/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=599,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3125/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=600,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3126/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=600,192,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3127/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=600,384,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3128/500000]
+ train/ActionL1Loss=0.1035
+ throughput/total_tokens=600,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3129/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=600,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3130/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=600,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3131/500000]
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=601,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3132/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=601,344,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3133/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=601,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3134/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=601,728,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3135/500000]
+ train/ActionL1Loss=0.0959
+ throughput/total_tokens=601,920,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3136/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=602,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3137/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=602,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3138/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=602,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3139/500000]
+ train/ActionL1Loss=0.0825
+ throughput/total_tokens=602,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3140/500000]
+ optim/total_grad_norm=6.995
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=602,880,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3141/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=603,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3142/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=603,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3143/500000]
+ train/ActionL1Loss=0.0916
+ throughput/total_tokens=603,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3144/500000]
+ train/ActionL1Loss=0.1123
+ throughput/total_tokens=603,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3145/500000]
+ train/ActionL1Loss=0.1004
+ throughput/total_tokens=603,840,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3146/500000]
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=604,032,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3147/500000]
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=604,224,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3148/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=604,416,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3149/500000]
+ train/ActionL1Loss=0.1021
+ throughput/total_tokens=604,608,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3150/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=604,800,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3151/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=604,992,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3152/500000]
+ train/ActionL1Loss=0.0922
+ throughput/total_tokens=605,184,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3153/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=605,376,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3154/500000]
+ train/ActionL1Loss=0.1064
+ throughput/total_tokens=605,568,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3155/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=605,760,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3156/500000]
+ train/ActionL1Loss=0.0866
+ throughput/total_tokens=605,952,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3157/500000]
+ train/ActionL1Loss=0.0876
+ throughput/total_tokens=606,144,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3158/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=606,336,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3159/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=606,528,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3160/500000]
+ optim/total_grad_norm=6.536
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=606,720,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3161/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=606,912,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3162/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=607,104,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3163/500000]
+ train/ActionL1Loss=0.0955
+ throughput/total_tokens=607,296,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3164/500000]
+ train/ActionL1Loss=0.0981
+ throughput/total_tokens=607,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3165/500000]
+ train/ActionL1Loss=0.0996
+ throughput/total_tokens=607,680,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3166/500000]
+ train/ActionL1Loss=0.0895
+ throughput/total_tokens=607,872,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3167/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=608,064,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3168/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=608,256,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3169/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=608,448,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3170/500000]
+ train/ActionL1Loss=0.0950
+ throughput/total_tokens=608,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3171/500000]
+ train/ActionL1Loss=0.1058
+ throughput/total_tokens=608,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3172/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=609,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3173/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=609,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3174/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=609,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3175/500000]
+ train/ActionL1Loss=0.1002
+ throughput/total_tokens=609,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3176/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=609,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3177/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=609,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3178/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=610,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3179/500000]
+ train/ActionL1Loss=0.0915
+ throughput/total_tokens=610,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3180/500000]
+ optim/total_grad_norm=6.877
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=610,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3181/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=610,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3182/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=610,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3183/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=611,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3184/500000]
+ train/ActionL1Loss=0.1037
+ throughput/total_tokens=611,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3185/500000]
+ train/ActionL1Loss=0.1057
+ throughput/total_tokens=611,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3186/500000]
+ train/ActionL1Loss=0.1165
+ throughput/total_tokens=611,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3187/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=611,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3188/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=612,096,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3189/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=612,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3190/500000]
+ train/ActionL1Loss=0.0948
+ throughput/total_tokens=612,480,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3191/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=612,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3192/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=612,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3193/500000]
+ train/ActionL1Loss=0.0956
+ throughput/total_tokens=613,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3194/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=613,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3195/500000]
+ train/ActionL1Loss=0.1132
+ throughput/total_tokens=613,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3196/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=613,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3197/500000]
+ train/ActionL1Loss=0.0917
+ throughput/total_tokens=613,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3198/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=614,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3199/500000]
+ train/ActionL1Loss=0.1062
+ throughput/total_tokens=614,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3200/500000]
+ optim/total_grad_norm=6.297
+ train/ActionL1Loss=0.1058
+ throughput/total_tokens=614,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3201/500000]
+ train/ActionL1Loss=0.1042
+ throughput/total_tokens=614,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3202/500000]
+ train/ActionL1Loss=0.0994
+ throughput/total_tokens=614,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3203/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=614,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3204/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=615,168,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3205/500000]
+ train/ActionL1Loss=0.0921
+ throughput/total_tokens=615,360,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3206/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=615,552,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3207/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=615,744,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3208/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=615,936,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3209/500000]
+ train/ActionL1Loss=0.1043
+ throughput/total_tokens=616,128,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3210/500000]
+ train/ActionL1Loss=0.1002
+ throughput/total_tokens=616,320,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3211/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=616,512,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3212/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=616,704,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3213/500000]
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=616,896,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3214/500000]
+ train/ActionL1Loss=0.0935
+ throughput/total_tokens=617,088,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3215/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=617,280,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3216/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=617,472,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3217/500000]
+ train/ActionL1Loss=0.1143
+ throughput/total_tokens=617,664,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3218/500000]
+ train/ActionL1Loss=0.1225
+ throughput/total_tokens=617,856,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3219/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=618,048,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3220/500000]
+ optim/total_grad_norm=5.921
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=618,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3221/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=618,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3222/500000]
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=618,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3223/500000]
+ train/ActionL1Loss=0.1124
+ throughput/total_tokens=618,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3224/500000]
+ train/ActionL1Loss=0.1095
+ throughput/total_tokens=619,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3225/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=619,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3226/500000]
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=619,392,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3227/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=619,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3228/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=619,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3229/500000]
+ train/ActionL1Loss=0.0926
+ throughput/total_tokens=619,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3230/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=620,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3231/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=620,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3232/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=620,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3233/500000]
+ train/ActionL1Loss=0.0897
+ throughput/total_tokens=620,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3234/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=620,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3235/500000]
+ train/ActionL1Loss=0.0854
+ throughput/total_tokens=621,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3236/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=621,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3237/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=621,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3238/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=621,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3239/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=621,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3240/500000]
+ optim/total_grad_norm=6.464
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=622,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3241/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=622,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3242/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=622,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3243/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=622,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3244/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=622,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3245/500000]
+ train/ActionL1Loss=0.0839
+ throughput/total_tokens=623,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3246/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=623,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3247/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=623,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3248/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=623,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3249/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=623,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3250/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=624,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3251/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=624,192,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3252/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=624,384,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3253/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=624,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3254/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=624,768,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3255/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=624,960,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3256/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=625,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3257/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=625,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3258/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=625,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3259/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=625,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3260/500000]
+ optim/total_grad_norm=6.841
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=625,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3261/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=626,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3262/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=626,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3263/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=626,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3264/500000]
+ train/ActionL1Loss=0.0868
+ throughput/total_tokens=626,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3265/500000]
+ train/ActionL1Loss=0.0961
+ throughput/total_tokens=626,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3266/500000]
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=627,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3267/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=627,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3268/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=627,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3269/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=627,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3270/500000]
+ train/ActionL1Loss=0.0876
+ throughput/total_tokens=627,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3271/500000]
+ train/ActionL1Loss=0.0993
+ throughput/total_tokens=628,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3272/500000]
+ train/ActionL1Loss=0.0933
+ throughput/total_tokens=628,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3273/500000]
+ train/ActionL1Loss=0.0996
+ throughput/total_tokens=628,416,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3274/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=628,608,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3275/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=628,800,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3276/500000]
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=628,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3277/500000]
+ train/ActionL1Loss=0.0890
+ throughput/total_tokens=629,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3278/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=629,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3279/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=629,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3280/500000]
+ optim/total_grad_norm=7.209
+ train/ActionL1Loss=0.0961
+ throughput/total_tokens=629,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3281/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=629,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3282/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=630,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3283/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=630,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3284/500000]
+ train/ActionL1Loss=0.0987
+ throughput/total_tokens=630,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3285/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=630,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3286/500000]
+ train/ActionL1Loss=0.1090
+ throughput/total_tokens=630,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3287/500000]
+ train/ActionL1Loss=0.0986
+ throughput/total_tokens=631,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3288/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=631,296,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3289/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=631,488,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3290/500000]
+ train/ActionL1Loss=0.0987
+ throughput/total_tokens=631,680,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3291/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=631,872,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3292/500000]
+ train/ActionL1Loss=0.1039
+ throughput/total_tokens=632,064,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3293/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=632,256,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3294/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=632,448,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3295/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=632,640,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3296/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=632,832,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3297/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=633,024,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3298/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=633,216,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3299/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=633,408,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3300/500000]
+ optim/total_grad_norm=7.805
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=633,600,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3301/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=633,792,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3302/500000]
+ train/ActionL1Loss=0.1010
+ throughput/total_tokens=633,984,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3303/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=634,176,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3304/500000]
+ train/ActionL1Loss=0.0956
+ throughput/total_tokens=634,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3305/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=634,560,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3306/500000]
+ train/ActionL1Loss=0.1117
+ throughput/total_tokens=634,752,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3307/500000]
+ train/ActionL1Loss=0.0934
+ throughput/total_tokens=634,944,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3308/500000]
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=635,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3309/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=635,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3310/500000]
+ train/ActionL1Loss=0.1004
+ throughput/total_tokens=635,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3311/500000]
+ train/ActionL1Loss=0.1128
+ throughput/total_tokens=635,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3312/500000]
+ train/ActionL1Loss=0.1058
+ throughput/total_tokens=635,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3313/500000]
+ train/ActionL1Loss=0.1069
+ throughput/total_tokens=636,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3314/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=636,288,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3315/500000]
+ train/ActionL1Loss=0.0903
+ throughput/total_tokens=636,480,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3316/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=636,672,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3317/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=636,864,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3318/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=637,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3319/500000]
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=637,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3320/500000]
+ optim/total_grad_norm=7.647
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=637,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3321/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=637,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3322/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=637,824,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3323/500000]
+ train/ActionL1Loss=0.1041
+ throughput/total_tokens=638,016,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3324/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=638,208,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3325/500000]
+ train/ActionL1Loss=0.0884
+ throughput/total_tokens=638,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3326/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=638,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3327/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=638,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3328/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=638,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3329/500000]
+ train/ActionL1Loss=0.0852
+ throughput/total_tokens=639,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3330/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=639,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3331/500000]
+ train/ActionL1Loss=0.0935
+ throughput/total_tokens=639,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3332/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=639,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3333/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=639,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3334/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=640,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3335/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=640,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3336/500000]
+ train/ActionL1Loss=0.0990
+ throughput/total_tokens=640,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3337/500000]
+ train/ActionL1Loss=0.0933
+ throughput/total_tokens=640,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3338/500000]
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=640,896,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3339/500000]
+ train/ActionL1Loss=0.0956
+ throughput/total_tokens=641,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3340/500000]
+ optim/total_grad_norm=6.765
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=641,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3341/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=641,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3342/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=641,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3343/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=641,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3344/500000]
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=642,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3345/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=642,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3346/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=642,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3347/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=642,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3348/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=642,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3349/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=643,008,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3350/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=643,200,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3351/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=643,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3352/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=643,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3353/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=643,776,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3354/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=643,968,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3355/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=644,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3356/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=644,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3357/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=644,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3358/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=644,736,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3359/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=644,928,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3360/500000]
+ optim/total_grad_norm=5.602
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=645,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3361/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=645,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3362/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=645,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3363/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=645,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3364/500000]
+ train/ActionL1Loss=0.0974
+ throughput/total_tokens=645,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3365/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=646,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3366/500000]
+ train/ActionL1Loss=0.0932
+ throughput/total_tokens=646,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3367/500000]
+ train/ActionL1Loss=0.1008
+ throughput/total_tokens=646,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3368/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=646,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3369/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=646,848,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3370/500000]
+ train/ActionL1Loss=0.1034
+ throughput/total_tokens=647,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3371/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=647,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3372/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=647,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3373/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=647,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3374/500000]
+ train/ActionL1Loss=0.1160
+ throughput/total_tokens=647,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3375/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=648,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3376/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=648,192,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3377/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=648,384,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3378/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=648,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3379/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=648,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3380/500000]
+ optim/total_grad_norm=7.642
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=648,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3381/500000]
+ train/ActionL1Loss=0.0912
+ throughput/total_tokens=649,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3382/500000]
+ train/ActionL1Loss=0.1114
+ throughput/total_tokens=649,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3383/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=649,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3384/500000]
+ train/ActionL1Loss=0.0934
+ throughput/total_tokens=649,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3385/500000]
+ train/ActionL1Loss=0.1054
+ throughput/total_tokens=649,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3386/500000]
+ train/ActionL1Loss=0.1006
+ throughput/total_tokens=650,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3387/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=650,304,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3388/500000]
+ train/ActionL1Loss=0.1041
+ throughput/total_tokens=650,496,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3389/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=650,688,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3390/500000]
+ train/ActionL1Loss=0.0905
+ throughput/total_tokens=650,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3391/500000]
+ train/ActionL1Loss=0.1013
+ throughput/total_tokens=651,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3392/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=651,264,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3393/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=651,456,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3394/500000]
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=651,648,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3395/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=651,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3396/500000]
+ train/ActionL1Loss=0.0953
+ throughput/total_tokens=652,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3397/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=652,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3398/500000]
+ train/ActionL1Loss=0.0984
+ throughput/total_tokens=652,416,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3399/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=652,608,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3400/500000]
+ optim/total_grad_norm=6.328
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=652,800,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3401/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=652,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3402/500000]
+ train/ActionL1Loss=0.0895
+ throughput/total_tokens=653,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3403/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=653,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3404/500000]
+ train/ActionL1Loss=0.1114
+ throughput/total_tokens=653,568,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3405/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=653,760,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3406/500000]
+ train/ActionL1Loss=0.0952
+ throughput/total_tokens=653,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3407/500000]
+ train/ActionL1Loss=0.1108
+ throughput/total_tokens=654,144,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3408/500000]
+ train/ActionL1Loss=0.1030
+ throughput/total_tokens=654,336,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3409/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=654,528,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3410/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=654,720,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3411/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=654,912,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3412/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=655,104,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3413/500000]
+ train/ActionL1Loss=0.1231
+ throughput/total_tokens=655,296,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3414/500000]
+ train/ActionL1Loss=0.0963
+ throughput/total_tokens=655,488,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3415/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=655,680,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3416/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=655,872,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3417/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=656,064,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3418/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=656,256,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3419/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=656,448,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3420/500000]
+ optim/total_grad_norm=6.612
+ train/ActionL1Loss=0.1071
+ throughput/total_tokens=656,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3421/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=656,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3422/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=657,024,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3423/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=657,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3424/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=657,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3425/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=657,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3426/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=657,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3427/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=657,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3428/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=658,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3429/500000]
+ train/ActionL1Loss=0.1060
+ throughput/total_tokens=658,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3430/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=658,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3431/500000]
+ train/ActionL1Loss=0.0935
+ throughput/total_tokens=658,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3432/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=658,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3433/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=659,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3434/500000]
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=659,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3435/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=659,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3436/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=659,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3437/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=659,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3438/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=660,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3439/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=660,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3440/500000]
+ optim/total_grad_norm=9.000
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=660,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3441/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=660,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3442/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=660,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3443/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=661,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3444/500000]
+ train/ActionL1Loss=0.0994
+ throughput/total_tokens=661,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3445/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=661,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3446/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=661,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3447/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=661,824,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3448/500000]
+ train/ActionL1Loss=0.0890
+ throughput/total_tokens=662,016,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3449/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=662,208,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3450/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=662,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3451/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=662,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3452/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=662,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3453/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=662,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3454/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=663,168,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3455/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=663,360,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3456/500000]
+ train/ActionL1Loss=0.0945
+ throughput/total_tokens=663,552,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3457/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=663,744,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3458/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=663,936,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3459/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=664,128,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3460/500000]
+ optim/total_grad_norm=6.926
+ train/ActionL1Loss=0.0979
+ throughput/total_tokens=664,320,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3461/500000]
+ train/ActionL1Loss=0.0994
+ throughput/total_tokens=664,512,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3462/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=664,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3463/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=664,896,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3464/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=665,088,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3465/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=665,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3466/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=665,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3467/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=665,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3468/500000]
+ train/ActionL1Loss=0.0884
+ throughput/total_tokens=665,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3469/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=666,048,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3470/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=666,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3471/500000]
+ train/ActionL1Loss=0.0928
+ throughput/total_tokens=666,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3472/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=666,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3473/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=666,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3474/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=667,008,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3475/500000]
+ train/ActionL1Loss=0.0887
+ throughput/total_tokens=667,200,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3476/500000]
+ train/ActionL1Loss=0.0919
+ throughput/total_tokens=667,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3477/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=667,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3478/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=667,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3479/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=667,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3480/500000]
+ optim/total_grad_norm=6.323
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=668,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3481/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=668,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3482/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=668,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3483/500000]
+ train/ActionL1Loss=0.0916
+ throughput/total_tokens=668,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3484/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=668,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3485/500000]
+ train/ActionL1Loss=0.1107
+ throughput/total_tokens=669,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3486/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=669,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3487/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=669,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3488/500000]
+ train/ActionL1Loss=0.0903
+ throughput/total_tokens=669,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3489/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=669,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3490/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=670,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3491/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=670,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3492/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=670,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3493/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=670,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3494/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=670,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3495/500000]
+ train/ActionL1Loss=0.0903
+ throughput/total_tokens=671,040,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3496/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=671,232,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3497/500000]
+ train/ActionL1Loss=0.0926
+ throughput/total_tokens=671,424,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3498/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=671,616,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3499/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=671,808,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3500/500000]
+ optim/total_grad_norm=5.764
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=672,000,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+10/01 [09:49:59] INFO | >> Saving config... checkpoint.py:608
+10/01 [09:50:43] INFO | >> Saving model state... checkpoint.py:796
+10/01 [09:51:56] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [09:53:29] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=3501/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=672,192,000
+ throughput/device/tokens_per_second=1,209
+ throughput/device/batches_per_second=0.0504
+[step=3502/500000]
+ train/ActionL1Loss=0.1006
+ throughput/total_tokens=672,384,000
+ throughput/device/tokens_per_second=1,173
+ throughput/device/batches_per_second=0.0489
+[step=3503/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=672,576,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=3504/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=672,768,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=3505/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=672,960,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=3506/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=673,152,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=3507/500000]
+ train/ActionL1Loss=0.0890
+ throughput/total_tokens=673,344,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3508/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=673,536,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3509/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=673,728,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3510/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=673,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3511/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=674,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3512/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=674,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3513/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=674,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3514/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=674,688,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3515/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=674,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3516/500000]
+ train/ActionL1Loss=0.1107
+ throughput/total_tokens=675,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3517/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=675,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3518/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=675,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3519/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=675,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3520/500000]
+ optim/total_grad_norm=5.974
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=675,840,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3521/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=676,032,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3522/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=676,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3523/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=676,416,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3524/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=676,608,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3525/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=676,800,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3526/500000]
+ train/ActionL1Loss=0.0887
+ throughput/total_tokens=676,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3527/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=677,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3528/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=677,376,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3529/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=677,568,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3530/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=677,760,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3531/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=677,952,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3532/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=678,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3533/500000]
+ train/ActionL1Loss=0.0844
+ throughput/total_tokens=678,336,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3534/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=678,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3535/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=678,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3536/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=678,912,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3537/500000]
+ train/ActionL1Loss=0.0974
+ throughput/total_tokens=679,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3538/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=679,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3539/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=679,488,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3540/500000]
+ optim/total_grad_norm=6.701
+ train/ActionL1Loss=0.1028
+ throughput/total_tokens=679,680,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3541/500000]
+ train/ActionL1Loss=0.1062
+ throughput/total_tokens=679,872,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3542/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=680,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3543/500000]
+ train/ActionL1Loss=0.0930
+ throughput/total_tokens=680,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3544/500000]
+ train/ActionL1Loss=0.0900
+ throughput/total_tokens=680,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3545/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=680,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3546/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=680,832,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3547/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=681,024,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3548/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=681,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3549/500000]
+ train/ActionL1Loss=0.0950
+ throughput/total_tokens=681,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3550/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=681,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3551/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=681,792,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3552/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=681,984,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3553/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=682,176,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3554/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=682,368,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3555/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=682,560,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3556/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=682,752,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3557/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=682,944,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3558/500000]
+ train/ActionL1Loss=0.0892
+ throughput/total_tokens=683,136,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3559/500000]
+ train/ActionL1Loss=0.0992
+ throughput/total_tokens=683,328,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3560/500000]
+ optim/total_grad_norm=7.310
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=683,520,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3561/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=683,712,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3562/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=683,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3563/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=684,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3564/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=684,288,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3565/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=684,480,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3566/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=684,672,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3567/500000]
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=684,864,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3568/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=685,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3569/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=685,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3570/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=685,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3571/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=685,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3572/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=685,824,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3573/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=686,016,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3574/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=686,208,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3575/500000]
+ train/ActionL1Loss=0.0946
+ throughput/total_tokens=686,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3576/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=686,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3577/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=686,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3578/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=686,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3579/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=687,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3580/500000]
+ optim/total_grad_norm=5.210
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=687,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3581/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=687,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3582/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=687,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3583/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=687,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3584/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=688,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3585/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=688,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3586/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=688,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3587/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=688,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3588/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=688,896,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3589/500000]
+ train/ActionL1Loss=0.1024
+ throughput/total_tokens=689,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3590/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=689,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3591/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=689,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3592/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=689,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3593/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=689,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3594/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=690,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3595/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=690,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3596/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=690,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3597/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=690,624,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3598/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=690,816,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3599/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=691,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3600/500000]
+ optim/total_grad_norm=5.721
+ train/ActionL1Loss=0.1162
+ throughput/total_tokens=691,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3601/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=691,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3602/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=691,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3603/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=691,776,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3604/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=691,968,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3605/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=692,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3606/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=692,352,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3607/500000]
+ train/ActionL1Loss=0.0948
+ throughput/total_tokens=692,544,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3608/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=692,736,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3609/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=692,928,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3610/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=693,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3611/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=693,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3612/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=693,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3613/500000]
+ train/ActionL1Loss=0.0928
+ throughput/total_tokens=693,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3614/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=693,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3615/500000]
+ train/ActionL1Loss=0.0954
+ throughput/total_tokens=694,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3616/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=694,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3617/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=694,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3618/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=694,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3619/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=694,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3620/500000]
+ optim/total_grad_norm=6.442
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=695,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3621/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=695,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3622/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=695,424,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3623/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=695,616,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3624/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=695,808,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3625/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=696,000,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3626/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=696,192,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3627/500000]
+ train/ActionL1Loss=0.1019
+ throughput/total_tokens=696,384,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3628/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=696,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3629/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=696,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3630/500000]
+ train/ActionL1Loss=0.0884
+ throughput/total_tokens=696,960,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3631/500000]
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=697,152,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3632/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=697,344,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3633/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=697,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3634/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=697,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3635/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=697,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3636/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=698,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3637/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=698,304,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3638/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=698,496,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3639/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=698,688,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3640/500000]
+ optim/total_grad_norm=7.595
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=698,880,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3641/500000]
+ train/ActionL1Loss=0.0986
+ throughput/total_tokens=699,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3642/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=699,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3643/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=699,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3644/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=699,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3645/500000]
+ train/ActionL1Loss=0.0960
+ throughput/total_tokens=699,840,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3646/500000]
+ train/ActionL1Loss=0.1049
+ throughput/total_tokens=700,032,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3647/500000]
+ train/ActionL1Loss=0.0989
+ throughput/total_tokens=700,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3648/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=700,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3649/500000]
+ train/ActionL1Loss=0.0836
+ throughput/total_tokens=700,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3650/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=700,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3651/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=700,992,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3652/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=701,184,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3653/500000]
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=701,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3654/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=701,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3655/500000]
+ train/ActionL1Loss=0.1097
+ throughput/total_tokens=701,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3656/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=701,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3657/500000]
+ train/ActionL1Loss=0.0905
+ throughput/total_tokens=702,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3658/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=702,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3659/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=702,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3660/500000]
+ optim/total_grad_norm=6.200
+ train/ActionL1Loss=0.0884
+ throughput/total_tokens=702,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3661/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=702,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3662/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=703,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3663/500000]
+ train/ActionL1Loss=0.0938
+ throughput/total_tokens=703,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3664/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=703,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3665/500000]
+ train/ActionL1Loss=0.0977
+ throughput/total_tokens=703,680,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3666/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=703,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3667/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=704,064,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3668/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=704,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3669/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=704,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3670/500000]
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=704,640,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3671/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=704,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3672/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=705,024,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3673/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=705,216,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3674/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=705,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3675/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=705,600,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3676/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=705,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3677/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=705,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3678/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=706,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3679/500000]
+ train/ActionL1Loss=0.0993
+ throughput/total_tokens=706,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3680/500000]
+ optim/total_grad_norm=5.651
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=706,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3681/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=706,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3682/500000]
+ train/ActionL1Loss=0.1086
+ throughput/total_tokens=706,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3683/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=707,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3684/500000]
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=707,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3685/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=707,520,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3686/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=707,712,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3687/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=707,904,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3688/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=708,096,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3689/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=708,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3690/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=708,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3691/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=708,672,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3692/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=708,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3693/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=709,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3694/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=709,248,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3695/500000]
+ train/ActionL1Loss=0.1031
+ throughput/total_tokens=709,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3696/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=709,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3697/500000]
+ train/ActionL1Loss=0.0963
+ throughput/total_tokens=709,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3698/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=710,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3699/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=710,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3700/500000]
+ optim/total_grad_norm=6.060
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=710,400,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3701/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=710,592,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3702/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=710,784,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3703/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=710,976,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3704/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=711,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3705/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=711,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3706/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=711,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3707/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=711,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3708/500000]
+ train/ActionL1Loss=0.0974
+ throughput/total_tokens=711,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3709/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=712,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3710/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=712,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3711/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=712,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3712/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=712,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3713/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=712,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3714/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=713,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3715/500000]
+ train/ActionL1Loss=0.1078
+ throughput/total_tokens=713,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3716/500000]
+ train/ActionL1Loss=0.1098
+ throughput/total_tokens=713,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3717/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=713,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3718/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=713,856,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3719/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=714,048,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3720/500000]
+ optim/total_grad_norm=7.046
+ train/ActionL1Loss=0.1008
+ throughput/total_tokens=714,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3721/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=714,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3722/500000]
+ train/ActionL1Loss=0.0963
+ throughput/total_tokens=714,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3723/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=714,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3724/500000]
+ train/ActionL1Loss=0.0976
+ throughput/total_tokens=715,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3725/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=715,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3726/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=715,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3727/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=715,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3728/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=715,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3729/500000]
+ train/ActionL1Loss=0.0938
+ throughput/total_tokens=715,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3730/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=716,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3731/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=716,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3732/500000]
+ train/ActionL1Loss=0.0923
+ throughput/total_tokens=716,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3733/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=716,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3734/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=716,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3735/500000]
+ train/ActionL1Loss=0.1134
+ throughput/total_tokens=717,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3736/500000]
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=717,312,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3737/500000]
+ train/ActionL1Loss=0.1065
+ throughput/total_tokens=717,504,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3738/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=717,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3739/500000]
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=717,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3740/500000]
+ optim/total_grad_norm=5.588
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=718,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3741/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=718,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3742/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=718,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3743/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=718,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3744/500000]
+ train/ActionL1Loss=0.1197
+ throughput/total_tokens=718,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3745/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=719,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3746/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=719,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3747/500000]
+ train/ActionL1Loss=0.1249
+ throughput/total_tokens=719,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3748/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=719,616,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3749/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=719,808,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3750/500000]
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=720,000,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3751/500000]
+ train/ActionL1Loss=0.0912
+ throughput/total_tokens=720,192,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3752/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=720,384,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3753/500000]
+ train/ActionL1Loss=0.0884
+ throughput/total_tokens=720,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3754/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=720,768,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3755/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=720,960,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3756/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=721,152,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3757/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=721,344,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3758/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=721,536,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3759/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=721,728,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3760/500000]
+ optim/total_grad_norm=6.713
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=721,920,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3761/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=722,112,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3762/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=722,304,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3763/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=722,496,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3764/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=722,688,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3765/500000]
+ train/ActionL1Loss=0.0563
+ throughput/total_tokens=722,880,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3766/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=723,072,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3767/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=723,264,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3768/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=723,456,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3769/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=723,648,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3770/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=723,840,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3771/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=724,032,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3772/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=724,224,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3773/500000]
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=724,416,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3774/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=724,608,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3775/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=724,800,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3776/500000]
+ train/ActionL1Loss=0.0971
+ throughput/total_tokens=724,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3777/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=725,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3778/500000]
+ train/ActionL1Loss=0.0901
+ throughput/total_tokens=725,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3779/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=725,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3780/500000]
+ optim/total_grad_norm=7.539
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=725,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3781/500000]
+ train/ActionL1Loss=0.1076
+ throughput/total_tokens=725,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3782/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=726,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3783/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=726,336,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3784/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=726,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3785/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=726,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3786/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=726,912,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3787/500000]
+ train/ActionL1Loss=0.0989
+ throughput/total_tokens=727,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3788/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=727,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3789/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=727,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3790/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=727,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3791/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=727,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3792/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=728,064,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3793/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=728,256,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3794/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=728,448,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3795/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=728,640,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3796/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=728,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3797/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=729,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3798/500000]
+ train/ActionL1Loss=0.0922
+ throughput/total_tokens=729,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3799/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=729,408,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3800/500000]
+ optim/total_grad_norm=5.345
+ train/ActionL1Loss=0.0934
+ throughput/total_tokens=729,600,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3801/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=729,792,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=3802/500000]
+ train/ActionL1Loss=0.0950
+ throughput/total_tokens=729,984,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3803/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=730,176,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3804/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=730,368,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3805/500000]
+ train/ActionL1Loss=0.0931
+ throughput/total_tokens=730,560,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3806/500000]
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=730,752,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3807/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=730,944,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3808/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=731,136,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3809/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=731,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3810/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=731,520,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3811/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=731,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3812/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=731,904,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3813/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=732,096,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3814/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=732,288,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3815/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=732,480,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3816/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=732,672,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=3817/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=732,864,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3818/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=733,056,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=3819/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=733,248,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3820/500000]
+ optim/total_grad_norm=5.372
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=733,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3821/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=733,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3822/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=733,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3823/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=734,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3824/500000]
+ train/ActionL1Loss=0.0978
+ throughput/total_tokens=734,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3825/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=734,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3826/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=734,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3827/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=734,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3828/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=734,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3829/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=735,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3830/500000]
+ train/ActionL1Loss=0.1005
+ throughput/total_tokens=735,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3831/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=735,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3832/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=735,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3833/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=735,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3834/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=736,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3835/500000]
+ train/ActionL1Loss=0.0890
+ throughput/total_tokens=736,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3836/500000]
+ train/ActionL1Loss=0.0923
+ throughput/total_tokens=736,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3837/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=736,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3838/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=736,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3839/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=737,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3840/500000]
+ optim/total_grad_norm=5.516
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=737,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3841/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=737,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3842/500000]
+ train/ActionL1Loss=0.0895
+ throughput/total_tokens=737,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3843/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=737,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3844/500000]
+ train/ActionL1Loss=0.1159
+ throughput/total_tokens=738,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3845/500000]
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=738,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3846/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=738,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3847/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=738,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3848/500000]
+ train/ActionL1Loss=0.1014
+ throughput/total_tokens=738,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3849/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=739,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3850/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=739,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3851/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=739,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3852/500000]
+ train/ActionL1Loss=0.0876
+ throughput/total_tokens=739,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3853/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=739,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3854/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=739,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3855/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=740,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3856/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=740,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3857/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=740,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3858/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=740,736,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3859/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=740,928,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3860/500000]
+ optim/total_grad_norm=6.403
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=741,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3861/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=741,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3862/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=741,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3863/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=741,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3864/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=741,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3865/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=742,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3866/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=742,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3867/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=742,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3868/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=742,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3869/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=742,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3870/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=743,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3871/500000]
+ train/ActionL1Loss=0.0963
+ throughput/total_tokens=743,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3872/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=743,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3873/500000]
+ train/ActionL1Loss=0.0955
+ throughput/total_tokens=743,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3874/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=743,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3875/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=744,000,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3876/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=744,192,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3877/500000]
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=744,384,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3878/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=744,576,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3879/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=744,768,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3880/500000]
+ optim/total_grad_norm=8.045
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=744,960,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3881/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=745,152,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3882/500000]
+ train/ActionL1Loss=0.0935
+ throughput/total_tokens=745,344,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3883/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=745,536,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3884/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=745,728,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3885/500000]
+ train/ActionL1Loss=0.1072
+ throughput/total_tokens=745,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3886/500000]
+ train/ActionL1Loss=0.1003
+ throughput/total_tokens=746,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3887/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=746,304,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3888/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=746,496,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3889/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=746,688,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3890/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=746,880,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3891/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=747,072,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3892/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=747,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3893/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=747,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3894/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=747,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3895/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=747,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3896/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=748,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3897/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=748,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3898/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=748,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3899/500000]
+ train/ActionL1Loss=0.1014
+ throughput/total_tokens=748,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3900/500000]
+ optim/total_grad_norm=6.681
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=748,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3901/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=748,992,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3902/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=749,184,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3903/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=749,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3904/500000]
+ train/ActionL1Loss=0.1133
+ throughput/total_tokens=749,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3905/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=749,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3906/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=749,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3907/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=750,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3908/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=750,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3909/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=750,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=3910/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=750,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=3911/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=750,912,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3912/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=751,104,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=3913/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=751,296,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3914/500000]
+ train/ActionL1Loss=0.0854
+ throughput/total_tokens=751,488,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3915/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=751,680,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3916/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=751,872,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3917/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=752,064,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3918/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=752,256,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3919/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=752,448,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=3920/500000]
+ optim/total_grad_norm=6.155
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=752,640,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3921/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=752,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3922/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=753,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3923/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=753,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3924/500000]
+ train/ActionL1Loss=0.0919
+ throughput/total_tokens=753,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3925/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=753,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3926/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=753,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3927/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=753,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3928/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=754,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3929/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=754,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3930/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=754,560,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3931/500000]
+ train/ActionL1Loss=0.0890
+ throughput/total_tokens=754,752,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3932/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=754,944,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3933/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=755,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3934/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=755,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3935/500000]
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=755,520,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3936/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=755,712,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3937/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=755,904,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3938/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=756,096,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3939/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=756,288,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3940/500000]
+ optim/total_grad_norm=6.576
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=756,480,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3941/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=756,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3942/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=756,864,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3943/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=757,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3944/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=757,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3945/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=757,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3946/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=757,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3947/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=757,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3948/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=758,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3949/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=758,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3950/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=758,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3951/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=758,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3952/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=758,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3953/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=758,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3954/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=759,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=3955/500000]
+ train/ActionL1Loss=0.0938
+ throughput/total_tokens=759,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3956/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=759,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3957/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=759,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3958/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=759,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3959/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=760,128,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3960/500000]
+ optim/total_grad_norm=5.031
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=760,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3961/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=760,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3962/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=760,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3963/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=760,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3964/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=761,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3965/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=761,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3966/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=761,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3967/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=761,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3968/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=761,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3969/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=762,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=3970/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=762,240,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3971/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=762,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=3972/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=762,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=3973/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=762,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=3974/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=763,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3975/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=763,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3976/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=763,392,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3977/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=763,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3978/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=763,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3979/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=763,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3980/500000]
+ optim/total_grad_norm=6.595
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=764,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=3981/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=764,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3982/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=764,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3983/500000]
+ train/ActionL1Loss=0.0929
+ throughput/total_tokens=764,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3984/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=764,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3985/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=765,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3986/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=765,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3987/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=765,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3988/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=765,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3989/500000]
+ train/ActionL1Loss=0.0903
+ throughput/total_tokens=765,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3990/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=766,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=3991/500000]
+ train/ActionL1Loss=0.0950
+ throughput/total_tokens=766,272,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=3992/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=766,464,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3993/500000]
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=766,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3994/500000]
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=766,848,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=3995/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=767,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3996/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=767,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3997/500000]
+ train/ActionL1Loss=0.0935
+ throughput/total_tokens=767,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3998/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=767,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=3999/500000]
+ train/ActionL1Loss=0.0989
+ throughput/total_tokens=767,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4000/500000]
+ optim/total_grad_norm=5.346
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=768,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+10/01 [12:47:00] INFO | >> Saving config... checkpoint.py:608
+10/01 [12:47:33] INFO | >> Saving model state... checkpoint.py:796
+10/01 [12:48:45] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [12:50:13] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=4001/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=768,192,000
+ throughput/device/tokens_per_second=1,208
+ throughput/device/batches_per_second=0.0504
+[step=4002/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=768,384,000
+ throughput/device/tokens_per_second=1,186
+ throughput/device/batches_per_second=0.0494
+[step=4003/500000]
+ train/ActionL1Loss=0.1040
+ throughput/total_tokens=768,576,000
+ throughput/device/tokens_per_second=1,178
+ throughput/device/batches_per_second=0.0491
+[step=4004/500000]
+ train/ActionL1Loss=0.0966
+ throughput/total_tokens=768,768,000
+ throughput/device/tokens_per_second=1,174
+ throughput/device/batches_per_second=0.0489
+[step=4005/500000]
+ train/ActionL1Loss=0.0900
+ throughput/total_tokens=768,960,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0488
+[step=4006/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=769,152,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=4007/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=769,344,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=4008/500000]
+ train/ActionL1Loss=0.1063
+ throughput/total_tokens=769,536,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=4009/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=769,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=4010/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=769,920,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=4011/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=770,112,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=4012/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=770,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=4013/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=770,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4014/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=770,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4015/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=770,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4016/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=771,072,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4017/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=771,264,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4018/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=771,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4019/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=771,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4020/500000]
+ optim/total_grad_norm=7.769
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=771,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4021/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=772,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4022/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=772,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4023/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=772,416,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4024/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=772,608,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4025/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=772,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4026/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=772,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4027/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=773,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4028/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=773,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4029/500000]
+ train/ActionL1Loss=0.0877
+ throughput/total_tokens=773,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4030/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=773,760,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4031/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=773,952,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4032/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=774,144,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4033/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=774,336,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4034/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=774,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4035/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=774,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4036/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=774,912,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4037/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=775,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4038/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=775,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4039/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=775,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4040/500000]
+ optim/total_grad_norm=7.766
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=775,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4041/500000]
+ train/ActionL1Loss=0.0983
+ throughput/total_tokens=775,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4042/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=776,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4043/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=776,256,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4044/500000]
+ train/ActionL1Loss=0.0866
+ throughput/total_tokens=776,448,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4045/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=776,640,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4046/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=776,832,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4047/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=777,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4048/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=777,216,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4049/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=777,408,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4050/500000]
+ train/ActionL1Loss=0.1044
+ throughput/total_tokens=777,600,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4051/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=777,792,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4052/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=777,984,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4053/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=778,176,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4054/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=778,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4055/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=778,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4056/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=778,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4057/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=778,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4058/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=779,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4059/500000]
+ train/ActionL1Loss=0.1051
+ throughput/total_tokens=779,328,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4060/500000]
+ optim/total_grad_norm=7.752
+ train/ActionL1Loss=0.0921
+ throughput/total_tokens=779,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4061/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=779,712,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4062/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=779,904,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4063/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=780,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4064/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=780,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4065/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=780,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4066/500000]
+ train/ActionL1Loss=0.0982
+ throughput/total_tokens=780,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4067/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=780,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4068/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=781,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4069/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=781,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4070/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=781,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4071/500000]
+ train/ActionL1Loss=0.1063
+ throughput/total_tokens=781,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4072/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=781,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4073/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=782,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4074/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=782,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4075/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=782,400,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4076/500000]
+ train/ActionL1Loss=0.0917
+ throughput/total_tokens=782,592,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4077/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=782,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4078/500000]
+ train/ActionL1Loss=0.1176
+ throughput/total_tokens=782,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4079/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=783,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4080/500000]
+ optim/total_grad_norm=5.745
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=783,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4081/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=783,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4082/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=783,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4083/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=783,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4084/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=784,128,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4085/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=784,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4086/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=784,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4087/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=784,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4088/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=784,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4089/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=785,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4090/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=785,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4091/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=785,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4092/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=785,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4093/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=785,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4094/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=786,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4095/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=786,240,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4096/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=786,432,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4097/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=786,624,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4098/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=786,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4099/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=787,008,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4100/500000]
+ optim/total_grad_norm=4.718
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=787,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4101/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=787,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4102/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=787,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4103/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=787,776,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4104/500000]
+ train/ActionL1Loss=0.1070
+ throughput/total_tokens=787,968,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4105/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=788,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4106/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=788,352,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4107/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=788,544,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4108/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=788,736,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4109/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=788,928,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4110/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=789,120,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4111/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=789,312,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4112/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=789,504,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4113/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=789,696,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4114/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=789,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4115/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=790,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4116/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=790,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4117/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=790,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4118/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=790,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4119/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=790,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4120/500000]
+ optim/total_grad_norm=5.062
+ train/ActionL1Loss=0.1062
+ throughput/total_tokens=791,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4121/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=791,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4122/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=791,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4123/500000]
+ train/ActionL1Loss=0.0896
+ throughput/total_tokens=791,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4124/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=791,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4125/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=792,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4126/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=792,192,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4127/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=792,384,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4128/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=792,576,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4129/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=792,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4130/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=792,960,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4131/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=793,152,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4132/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=793,344,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4133/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=793,536,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4134/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=793,728,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4135/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=793,920,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4136/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=794,112,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4137/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=794,304,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4138/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=794,496,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4139/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=794,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4140/500000]
+ optim/total_grad_norm=5.645
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=794,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4141/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=795,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4142/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=795,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4143/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=795,456,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4144/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=795,648,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4145/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=795,840,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4146/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=796,032,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4147/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=796,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4148/500000]
+ train/ActionL1Loss=0.0956
+ throughput/total_tokens=796,416,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4149/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=796,608,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4150/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=796,800,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4151/500000]
+ train/ActionL1Loss=0.0884
+ throughput/total_tokens=796,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4152/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=797,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4153/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=797,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4154/500000]
+ train/ActionL1Loss=0.0958
+ throughput/total_tokens=797,568,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4155/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=797,760,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4156/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=797,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4157/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=798,144,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4158/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=798,336,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4159/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=798,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4160/500000]
+ optim/total_grad_norm=7.920
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=798,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4161/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=798,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4162/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=799,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4163/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=799,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4164/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=799,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4165/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=799,680,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4166/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=799,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4167/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=800,064,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4168/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=800,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4169/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=800,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4170/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=800,640,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4171/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=800,832,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4172/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=801,024,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4173/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=801,216,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4174/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=801,408,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4175/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=801,600,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4176/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=801,792,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4177/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=801,984,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4178/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=802,176,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4179/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=802,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4180/500000]
+ optim/total_grad_norm=5.140
+ train/ActionL1Loss=0.0912
+ throughput/total_tokens=802,560,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4181/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=802,752,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4182/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=802,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4183/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=803,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4184/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=803,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4185/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=803,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4186/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=803,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4187/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=803,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4188/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=804,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4189/500000]
+ train/ActionL1Loss=0.1050
+ throughput/total_tokens=804,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4190/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=804,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4191/500000]
+ train/ActionL1Loss=0.0972
+ throughput/total_tokens=804,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4192/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=804,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4193/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=805,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4194/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=805,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4195/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=805,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4196/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=805,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4197/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=805,824,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4198/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=806,016,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4199/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=806,208,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4200/500000]
+ optim/total_grad_norm=6.487
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=806,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4201/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=806,592,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4202/500000]
+ train/ActionL1Loss=0.0934
+ throughput/total_tokens=806,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4203/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=806,976,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4204/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=807,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4205/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=807,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4206/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=807,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4207/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=807,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4208/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=807,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4209/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=808,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4210/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=808,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4211/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=808,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4212/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=808,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4213/500000]
+ train/ActionL1Loss=0.0925
+ throughput/total_tokens=808,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4214/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=809,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4215/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=809,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4216/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=809,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4217/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=809,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4218/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=809,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4219/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=810,048,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4220/500000]
+ optim/total_grad_norm=5.329
+ train/ActionL1Loss=0.0866
+ throughput/total_tokens=810,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4221/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=810,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4222/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=810,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4223/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=810,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4224/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=811,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4225/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=811,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4226/500000]
+ train/ActionL1Loss=0.0897
+ throughput/total_tokens=811,392,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4227/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=811,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4228/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=811,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4229/500000]
+ train/ActionL1Loss=0.1045
+ throughput/total_tokens=811,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4230/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=812,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4231/500000]
+ train/ActionL1Loss=0.0877
+ throughput/total_tokens=812,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4232/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=812,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4233/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=812,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4234/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=812,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4235/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=813,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4236/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=813,312,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4237/500000]
+ train/ActionL1Loss=0.0852
+ throughput/total_tokens=813,504,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4238/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=813,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4239/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=813,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4240/500000]
+ optim/total_grad_norm=5.508
+ train/ActionL1Loss=0.1117
+ throughput/total_tokens=814,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4241/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=814,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4242/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=814,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4243/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=814,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4244/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=814,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4245/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=815,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4246/500000]
+ train/ActionL1Loss=0.0901
+ throughput/total_tokens=815,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4247/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=815,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4248/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=815,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4249/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=815,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4250/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=816,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4251/500000]
+ train/ActionL1Loss=0.0781
+ throughput/total_tokens=816,192,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4252/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=816,384,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4253/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=816,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4254/500000]
+ train/ActionL1Loss=0.0953
+ throughput/total_tokens=816,768,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4255/500000]
+ train/ActionL1Loss=0.0916
+ throughput/total_tokens=816,960,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4256/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=817,152,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4257/500000]
+ train/ActionL1Loss=0.0868
+ throughput/total_tokens=817,344,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4258/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=817,536,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4259/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=817,728,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4260/500000]
+ optim/total_grad_norm=7.235
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=817,920,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4261/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=818,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4262/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=818,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4263/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=818,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4264/500000]
+ train/ActionL1Loss=0.0978
+ throughput/total_tokens=818,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4265/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=818,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4266/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=819,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4267/500000]
+ train/ActionL1Loss=0.0877
+ throughput/total_tokens=819,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4268/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=819,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4269/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=819,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4270/500000]
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=819,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4271/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=820,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4272/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=820,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4273/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=820,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4274/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=820,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4275/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=820,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4276/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=820,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4277/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=821,184,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4278/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=821,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4279/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=821,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4280/500000]
+ optim/total_grad_norm=7.295
+ train/ActionL1Loss=0.1101
+ throughput/total_tokens=821,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4281/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=821,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4282/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=822,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4283/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=822,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4284/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=822,528,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4285/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=822,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4286/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=822,912,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4287/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=823,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4288/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=823,296,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4289/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=823,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4290/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=823,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4291/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=823,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4292/500000]
+ train/ActionL1Loss=0.0914
+ throughput/total_tokens=824,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4293/500000]
+ train/ActionL1Loss=0.0866
+ throughput/total_tokens=824,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4294/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=824,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4295/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=824,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4296/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=824,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4297/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=825,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4298/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=825,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4299/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=825,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4300/500000]
+ optim/total_grad_norm=6.946
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=825,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4301/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=825,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4302/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=825,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4303/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=826,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4304/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=826,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4305/500000]
+ train/ActionL1Loss=0.0849
+ throughput/total_tokens=826,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4306/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=826,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4307/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=826,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4308/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=827,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4309/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=827,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4310/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=827,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4311/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=827,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4312/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=827,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4313/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=828,096,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4314/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=828,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4315/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=828,480,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4316/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=828,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4317/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=828,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4318/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=829,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4319/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=829,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4320/500000]
+ optim/total_grad_norm=6.188
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=829,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4321/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=829,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4322/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=829,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4323/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=830,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4324/500000]
+ train/ActionL1Loss=0.1093
+ throughput/total_tokens=830,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4325/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=830,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4326/500000]
+ train/ActionL1Loss=0.0905
+ throughput/total_tokens=830,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4327/500000]
+ train/ActionL1Loss=0.0781
+ throughput/total_tokens=830,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4328/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=830,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4329/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=831,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4330/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=831,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4331/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=831,552,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4332/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=831,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4333/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=831,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4334/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=832,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4335/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=832,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4336/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=832,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4337/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=832,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4338/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=832,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4339/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=833,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4340/500000]
+ optim/total_grad_norm=7.227
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=833,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4341/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=833,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4342/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=833,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4343/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=833,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4344/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=834,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4345/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=834,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4346/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=834,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4347/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=834,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4348/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=834,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4349/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=835,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4350/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=835,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4351/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=835,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4352/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=835,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4353/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=835,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4354/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=835,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4355/500000]
+ train/ActionL1Loss=0.0976
+ throughput/total_tokens=836,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4356/500000]
+ train/ActionL1Loss=0.0926
+ throughput/total_tokens=836,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4357/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=836,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4358/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=836,736,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4359/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=836,928,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4360/500000]
+ optim/total_grad_norm=5.045
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=837,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4361/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=837,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4362/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=837,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4363/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=837,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4364/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=837,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4365/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=838,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4366/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=838,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4367/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=838,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4368/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=838,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4369/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=838,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4370/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=839,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4371/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=839,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4372/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=839,424,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4373/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=839,616,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4374/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=839,808,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4375/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=840,000,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4376/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=840,192,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4377/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=840,384,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4378/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=840,576,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4379/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=840,768,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4380/500000]
+ optim/total_grad_norm=6.402
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=840,960,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4381/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=841,152,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4382/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=841,344,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4383/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=841,536,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4384/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=841,728,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4385/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=841,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4386/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=842,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4387/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=842,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4388/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=842,496,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4389/500000]
+ train/ActionL1Loss=0.0514
+ throughput/total_tokens=842,688,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4390/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=842,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4391/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=843,072,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4392/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=843,264,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4393/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=843,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4394/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=843,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4395/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=843,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4396/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=844,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4397/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=844,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4398/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=844,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4399/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=844,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4400/500000]
+ optim/total_grad_norm=5.066
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=844,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4401/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=844,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4402/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=845,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4403/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=845,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4404/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=845,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4405/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=845,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4406/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=845,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4407/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=846,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4408/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=846,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4409/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=846,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4410/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=846,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4411/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=846,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4412/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=847,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4413/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=847,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4414/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=847,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4415/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=847,680,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4416/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=847,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4417/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=848,064,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4418/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=848,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4419/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=848,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4420/500000]
+ optim/total_grad_norm=7.887
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=848,640,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4421/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=848,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4422/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=849,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4423/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=849,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4424/500000]
+ train/ActionL1Loss=0.0916
+ throughput/total_tokens=849,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4425/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=849,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4426/500000]
+ train/ActionL1Loss=0.1027
+ throughput/total_tokens=849,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4427/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=849,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4428/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=850,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4429/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=850,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4430/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=850,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4431/500000]
+ train/ActionL1Loss=0.0932
+ throughput/total_tokens=850,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4432/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=850,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4433/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=851,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4434/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=851,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4435/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=851,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4436/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=851,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4437/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=851,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4438/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=852,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4439/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=852,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4440/500000]
+ optim/total_grad_norm=6.849
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=852,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4441/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=852,672,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4442/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=852,864,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4443/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=853,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4444/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=853,248,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4445/500000]
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=853,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4446/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=853,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4447/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=853,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4448/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=854,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4449/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=854,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4450/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=854,400,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4451/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=854,592,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4452/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=854,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4453/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=854,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4454/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=855,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4455/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=855,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4456/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=855,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4457/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=855,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4458/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=855,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4459/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=856,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4460/500000]
+ optim/total_grad_norm=8.478
+ train/ActionL1Loss=0.0892
+ throughput/total_tokens=856,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4461/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=856,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4462/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=856,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4463/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=856,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4464/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=857,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4465/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=857,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4466/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=857,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4467/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=857,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4468/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=857,856,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4469/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=858,048,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4470/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=858,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4471/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=858,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4472/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=858,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4473/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=858,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4474/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=859,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4475/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=859,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4476/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=859,392,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4477/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=859,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4478/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=859,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4479/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=859,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4480/500000]
+ optim/total_grad_norm=7.960
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=860,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4481/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=860,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4482/500000]
+ train/ActionL1Loss=0.0900
+ throughput/total_tokens=860,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4483/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=860,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4484/500000]
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=860,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4485/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=861,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4486/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=861,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4487/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=861,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4488/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=861,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4489/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=861,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4490/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=862,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4491/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=862,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4492/500000]
+ train/ActionL1Loss=0.1046
+ throughput/total_tokens=862,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4493/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=862,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4494/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=862,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4495/500000]
+ train/ActionL1Loss=0.0836
+ throughput/total_tokens=863,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4496/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=863,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4497/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=863,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4498/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=863,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4499/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=863,808,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4500/500000]
+ optim/total_grad_norm=6.681
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=864,000,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+10/01 [15:43:27] INFO | >> Saving config... checkpoint.py:608
+10/01 [15:44:12] INFO | >> Saving model state... checkpoint.py:796
+10/01 [15:45:23] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [15:46:49] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=4501/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=864,192,000
+ throughput/device/tokens_per_second=1,211
+ throughput/device/batches_per_second=0.0505
+[step=4502/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=864,384,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=4503/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=864,576,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4504/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=864,768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4505/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=864,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4506/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=865,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4507/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=865,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4508/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=865,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4509/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=865,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4510/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=865,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4511/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=866,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4512/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=866,304,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4513/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=866,496,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4514/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=866,688,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4515/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=866,880,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4516/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=867,072,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4517/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=867,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4518/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=867,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4519/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=867,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4520/500000]
+ optim/total_grad_norm=5.821
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=867,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4521/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=868,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4522/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=868,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4523/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=868,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4524/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=868,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4525/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=868,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4526/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=868,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4527/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=869,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4528/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=869,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4529/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=869,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4530/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=869,760,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4531/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=869,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4532/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=870,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4533/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=870,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4534/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=870,528,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4535/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=870,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4536/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=870,912,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4537/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=871,104,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4538/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=871,296,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4539/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=871,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4540/500000]
+ optim/total_grad_norm=5.997
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=871,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4541/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=871,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4542/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=872,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4543/500000]
+ train/ActionL1Loss=0.0996
+ throughput/total_tokens=872,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4544/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=872,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4545/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=872,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4546/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=872,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4547/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=873,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4548/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=873,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4549/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=873,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4550/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=873,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4551/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=873,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4552/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=873,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4553/500000]
+ train/ActionL1Loss=0.0900
+ throughput/total_tokens=874,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4554/500000]
+ train/ActionL1Loss=0.0993
+ throughput/total_tokens=874,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4555/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=874,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4556/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=874,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4557/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=874,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4558/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=875,136,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4559/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=875,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4560/500000]
+ optim/total_grad_norm=7.121
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=875,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4561/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=875,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4562/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=875,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4563/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=876,096,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4564/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=876,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4565/500000]
+ train/ActionL1Loss=0.0919
+ throughput/total_tokens=876,480,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4566/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=876,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4567/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=876,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4568/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=877,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4569/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=877,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4570/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=877,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4571/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=877,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4572/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=877,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4573/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=878,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4574/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=878,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4575/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=878,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4576/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=878,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4577/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=878,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4578/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=878,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4579/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=879,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4580/500000]
+ optim/total_grad_norm=4.994
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=879,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4581/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=879,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4582/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=879,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4583/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=879,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4584/500000]
+ train/ActionL1Loss=0.0933
+ throughput/total_tokens=880,128,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4585/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=880,320,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4586/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=880,512,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4587/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=880,704,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4588/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=880,896,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4589/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=881,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4590/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=881,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4591/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=881,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4592/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=881,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4593/500000]
+ train/ActionL1Loss=0.0988
+ throughput/total_tokens=881,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4594/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=882,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4595/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=882,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4596/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=882,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4597/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=882,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4598/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=882,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4599/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=883,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4600/500000]
+ optim/total_grad_norm=7.739
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=883,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4601/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=883,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4602/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=883,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4603/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=883,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4604/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=883,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4605/500000]
+ train/ActionL1Loss=0.1050
+ throughput/total_tokens=884,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4606/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=884,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4607/500000]
+ train/ActionL1Loss=0.1061
+ throughput/total_tokens=884,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4608/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=884,736,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4609/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=884,928,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4610/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=885,120,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4611/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=885,312,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4612/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=885,504,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4613/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=885,696,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4614/500000]
+ train/ActionL1Loss=0.0885
+ throughput/total_tokens=885,888,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4615/500000]
+ train/ActionL1Loss=0.1097
+ throughput/total_tokens=886,080,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4616/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=886,272,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4617/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=886,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4618/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=886,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4619/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=886,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4620/500000]
+ optim/total_grad_norm=6.824
+ train/ActionL1Loss=0.0932
+ throughput/total_tokens=887,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4621/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=887,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4622/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=887,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4623/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=887,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4624/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=887,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4625/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=888,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4626/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=888,192,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4627/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=888,384,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4628/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=888,576,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4629/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=888,768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4630/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=888,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4631/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=889,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4632/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=889,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4633/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=889,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4634/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=889,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4635/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=889,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4636/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=890,112,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4637/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=890,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4638/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=890,496,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4639/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=890,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4640/500000]
+ optim/total_grad_norm=7.559
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=890,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4641/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=891,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4642/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=891,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4643/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=891,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4644/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=891,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4645/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=891,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4646/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=892,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4647/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=892,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4648/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=892,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4649/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=892,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4650/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=892,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4651/500000]
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=892,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4652/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=893,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4653/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=893,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4654/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=893,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4655/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=893,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4656/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=893,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4657/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=894,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4658/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=894,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4659/500000]
+ train/ActionL1Loss=0.0939
+ throughput/total_tokens=894,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4660/500000]
+ optim/total_grad_norm=6.307
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=894,720,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4661/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=894,912,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4662/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=895,104,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4663/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=895,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4664/500000]
+ train/ActionL1Loss=0.0851
+ throughput/total_tokens=895,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4665/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=895,680,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4666/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=895,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4667/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=896,064,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4668/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=896,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4669/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=896,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4670/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=896,640,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4671/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=896,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4672/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=897,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4673/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=897,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4674/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=897,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4675/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=897,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4676/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=897,792,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4677/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=897,984,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4678/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=898,176,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4679/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=898,368,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4680/500000]
+ optim/total_grad_norm=5.929
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=898,560,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4681/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=898,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4682/500000]
+ train/ActionL1Loss=0.0781
+ throughput/total_tokens=898,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4683/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=899,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4684/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=899,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4685/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=899,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4686/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=899,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4687/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=899,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4688/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=900,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4689/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=900,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4690/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=900,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4691/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=900,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4692/500000]
+ train/ActionL1Loss=0.0962
+ throughput/total_tokens=900,864,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4693/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=901,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4694/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=901,248,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4695/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=901,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4696/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=901,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4697/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=901,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4698/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=902,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4699/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=902,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4700/500000]
+ optim/total_grad_norm=6.346
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=902,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4701/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=902,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4702/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=902,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4703/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=902,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4704/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=903,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4705/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=903,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4706/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=903,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4707/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=903,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4708/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=903,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4709/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=904,128,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4710/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=904,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4711/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=904,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4712/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=904,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4713/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=904,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4714/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=905,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4715/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=905,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4716/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=905,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4717/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=905,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4718/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=905,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4719/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=906,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4720/500000]
+ optim/total_grad_norm=5.812
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=906,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=4721/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=906,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=4722/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=906,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=4723/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=906,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=4724/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=907,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=4725/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=907,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=4726/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=907,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4727/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=907,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4728/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=907,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4729/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=907,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4730/500000]
+ train/ActionL1Loss=0.0961
+ throughput/total_tokens=908,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4731/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=908,352,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4732/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=908,544,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4733/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=908,736,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4734/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=908,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4735/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=909,120,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4736/500000]
+ train/ActionL1Loss=0.0968
+ throughput/total_tokens=909,312,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4737/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=909,504,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4738/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=909,696,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4739/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=909,888,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4740/500000]
+ optim/total_grad_norm=7.200
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=910,080,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4741/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=910,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4742/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=910,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4743/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=910,656,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4744/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=910,848,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4745/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=911,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4746/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=911,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4747/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=911,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4748/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=911,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4749/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=911,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4750/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=912,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4751/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=912,192,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=4752/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=912,384,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=4753/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=912,576,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=4754/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=912,768,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0481
+[step=4755/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=912,960,000
+ throughput/device/tokens_per_second=1,153
+ throughput/device/batches_per_second=0.0480
+[step=4756/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=913,152,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=4757/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=913,344,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=4758/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=913,536,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=4759/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=913,728,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4760/500000]
+ optim/total_grad_norm=6.466
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=913,920,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+ System/Peak GPU Memory (MB)=46,917
+[step=4761/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=914,112,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4762/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=914,304,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4763/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=914,496,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4764/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=914,688,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4765/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=914,880,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4766/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=915,072,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4767/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=915,264,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4768/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=915,456,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4769/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=915,648,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=4770/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=915,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4771/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=916,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4772/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=916,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4773/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=916,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4774/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=916,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4775/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=916,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4776/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=916,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4777/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=917,184,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4778/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=917,376,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4779/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=917,568,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4780/500000]
+ optim/total_grad_norm=4.762
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=917,760,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4781/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=917,952,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4782/500000]
+ train/ActionL1Loss=0.0944
+ throughput/total_tokens=918,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4783/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=918,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4784/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=918,528,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4785/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=918,720,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4786/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=918,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4787/500000]
+ train/ActionL1Loss=0.1073
+ throughput/total_tokens=919,104,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4788/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=919,296,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4789/500000]
+ train/ActionL1Loss=0.0898
+ throughput/total_tokens=919,488,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4790/500000]
+ train/ActionL1Loss=0.0774
+ throughput/total_tokens=919,680,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4791/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=919,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4792/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=920,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4793/500000]
+ train/ActionL1Loss=0.1059
+ throughput/total_tokens=920,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4794/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=920,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4795/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=920,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4796/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=920,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4797/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=921,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4798/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=921,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4799/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=921,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4800/500000]
+ optim/total_grad_norm=6.525
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=921,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4801/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=921,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4802/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=921,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4803/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=922,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4804/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=922,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4805/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=922,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4806/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=922,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4807/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=922,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4808/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=923,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4809/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=923,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4810/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=923,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4811/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=923,712,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4812/500000]
+ train/ActionL1Loss=0.0906
+ throughput/total_tokens=923,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4813/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=924,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4814/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=924,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4815/500000]
+ train/ActionL1Loss=0.0876
+ throughput/total_tokens=924,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4816/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=924,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4817/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=924,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4818/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=925,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4819/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=925,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4820/500000]
+ optim/total_grad_norm=6.409
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=925,440,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4821/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=925,632,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4822/500000]
+ train/ActionL1Loss=0.0970
+ throughput/total_tokens=925,824,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4823/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=926,016,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4824/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=926,208,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4825/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=926,400,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4826/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=926,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4827/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=926,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4828/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=926,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4829/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=927,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4830/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=927,360,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4831/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=927,552,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4832/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=927,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4833/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=927,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4834/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=928,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4835/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=928,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4836/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=928,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4837/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=928,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4838/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=928,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4839/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=929,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4840/500000]
+ optim/total_grad_norm=7.313
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=929,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4841/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=929,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4842/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=929,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4843/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=929,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4844/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=930,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4845/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=930,240,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4846/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=930,432,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4847/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=930,624,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4848/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=930,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4849/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=931,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4850/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=931,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4851/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=931,392,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4852/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=931,584,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4853/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=931,776,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4854/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=931,968,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4855/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=932,160,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4856/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=932,352,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=4857/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=932,544,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=4858/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=932,736,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=4859/500000]
+ train/ActionL1Loss=0.0781
+ throughput/total_tokens=932,928,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=4860/500000]
+ optim/total_grad_norm=7.210
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=933,120,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4861/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=933,312,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=4862/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=933,504,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=4863/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=933,696,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=4864/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=933,888,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4865/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=934,080,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4866/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=934,272,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4867/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=934,464,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4868/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=934,656,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4869/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=934,848,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4870/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=935,040,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4871/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=935,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4872/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=935,424,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4873/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=935,616,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4874/500000]
+ train/ActionL1Loss=0.0854
+ throughput/total_tokens=935,808,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4875/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=936,000,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4876/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=936,192,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4877/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=936,384,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4878/500000]
+ train/ActionL1Loss=0.0854
+ throughput/total_tokens=936,576,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4879/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=936,768,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4880/500000]
+ optim/total_grad_norm=5.196
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=936,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4881/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=937,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4882/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=937,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4883/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=937,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4884/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=937,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4885/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=937,920,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4886/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=938,112,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4887/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=938,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4888/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=938,496,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4889/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=938,688,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4890/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=938,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4891/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=939,072,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4892/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=939,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4893/500000]
+ train/ActionL1Loss=0.1056
+ throughput/total_tokens=939,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4894/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=939,648,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4895/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=939,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4896/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=940,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4897/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=940,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4898/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=940,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4899/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=940,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4900/500000]
+ optim/total_grad_norm=4.499
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=940,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4901/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=940,992,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4902/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=941,184,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4903/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=941,376,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4904/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=941,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4905/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=941,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=4906/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=941,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4907/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=942,144,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4908/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=942,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4909/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=942,528,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4910/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=942,720,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=4911/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=942,912,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4912/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=943,104,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4913/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=943,296,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4914/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=943,488,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=4915/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=943,680,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4916/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=943,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4917/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=944,064,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4918/500000]
+ train/ActionL1Loss=0.0872
+ throughput/total_tokens=944,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4919/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=944,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=4920/500000]
+ optim/total_grad_norm=6.650
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=944,640,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4921/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=944,832,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4922/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=945,024,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4923/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=945,216,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4924/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=945,408,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4925/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=945,600,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4926/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=945,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4927/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=945,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4928/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=946,176,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4929/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=946,368,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4930/500000]
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=946,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4931/500000]
+ train/ActionL1Loss=0.0879
+ throughput/total_tokens=946,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4932/500000]
+ train/ActionL1Loss=0.0874
+ throughput/total_tokens=946,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4933/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=947,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4934/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=947,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4935/500000]
+ train/ActionL1Loss=0.0508
+ throughput/total_tokens=947,520,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4936/500000]
+ train/ActionL1Loss=0.0887
+ throughput/total_tokens=947,712,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4937/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=947,904,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4938/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=948,096,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=4939/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=948,288,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4940/500000]
+ optim/total_grad_norm=7.143
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=948,480,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4941/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=948,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4942/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=948,864,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4943/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=949,056,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4944/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=949,248,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4945/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=949,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4946/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=949,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4947/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=949,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4948/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=950,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4949/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=950,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=4950/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=950,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4951/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=950,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4952/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=950,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4953/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=950,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4954/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=951,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4955/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=951,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4956/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=951,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4957/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=951,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4958/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=951,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4959/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=952,128,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4960/500000]
+ optim/total_grad_norm=6.449
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=952,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4961/500000]
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=952,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4962/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=952,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4963/500000]
+ train/ActionL1Loss=0.0980
+ throughput/total_tokens=952,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4964/500000]
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=953,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4965/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=953,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4966/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=953,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4967/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=953,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4968/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=953,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4969/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=954,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=4970/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=954,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4971/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=954,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4972/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=954,624,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4973/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=954,816,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4974/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=955,008,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4975/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=955,200,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4976/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=955,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4977/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=955,584,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4978/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=955,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4979/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=955,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4980/500000]
+ optim/total_grad_norm=7.788
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=956,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=4981/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=956,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4982/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=956,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4983/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=956,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4984/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=956,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4985/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=957,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4986/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=957,312,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=4987/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=957,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4988/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=957,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4989/500000]
+ train/ActionL1Loss=0.0836
+ throughput/total_tokens=957,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=4990/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=958,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=4991/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=958,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4992/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=958,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4993/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=958,656,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4994/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=958,848,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4995/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=959,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4996/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=959,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4997/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=959,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4998/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=959,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=4999/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=959,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5000/500000]
+ optim/total_grad_norm=9.134
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=960,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+10/01 [18:40:18] INFO | >> Saving config... checkpoint.py:608
+10/01 [18:40:48] INFO | >> Saving model state... checkpoint.py:796
+10/01 [18:42:00] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [18:43:32] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=5001/500000]
+ train/ActionL1Loss=0.1007
+ throughput/total_tokens=960,192,000
+ throughput/device/tokens_per_second=1,212
+ throughput/device/batches_per_second=0.0505
+[step=5002/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=960,384,000
+ throughput/device/tokens_per_second=1,173
+ throughput/device/batches_per_second=0.0489
+[step=5003/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=960,576,000
+ throughput/device/tokens_per_second=1,144
+ throughput/device/batches_per_second=0.0477
+[step=5004/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=960,768,000
+ throughput/device/tokens_per_second=1,142
+ throughput/device/batches_per_second=0.0476
+[step=5005/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=960,960,000
+ throughput/device/tokens_per_second=1,146
+ throughput/device/batches_per_second=0.0478
+[step=5006/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=961,152,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0479
+[step=5007/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=961,344,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=5008/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=961,536,000
+ throughput/device/tokens_per_second=1,155
+ throughput/device/batches_per_second=0.0481
+[step=5009/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=961,728,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=5010/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=961,920,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+ System/Peak GPU Memory (MB)=46,917
+[step=5011/500000]
+ train/ActionL1Loss=0.0824
+ throughput/total_tokens=962,112,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=5012/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=962,304,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=5013/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=962,496,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=5014/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=962,688,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=5015/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=962,880,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=5016/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=963,072,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=5017/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=963,264,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=5018/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=963,456,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=5019/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=963,648,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=5020/500000]
+ optim/total_grad_norm=6.731
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=963,840,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=5021/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=964,032,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=5022/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=964,224,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5023/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=964,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5024/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=964,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5025/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=964,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5026/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=964,992,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5027/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=965,184,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5028/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=965,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5029/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=965,568,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5030/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=965,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5031/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=965,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5032/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=966,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5033/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=966,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5034/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=966,528,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5035/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=966,720,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5036/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=966,912,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5037/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=967,104,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5038/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=967,296,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5039/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=967,488,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5040/500000]
+ optim/total_grad_norm=6.433
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=967,680,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5041/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=967,872,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5042/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=968,064,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5043/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=968,256,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5044/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=968,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5045/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=968,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5046/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=968,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5047/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=969,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5048/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=969,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5049/500000]
+ train/ActionL1Loss=0.0839
+ throughput/total_tokens=969,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5050/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=969,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5051/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=969,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5052/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=969,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5053/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=970,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5054/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=970,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5055/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=970,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5056/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=970,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5057/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=970,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5058/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=971,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5059/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=971,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5060/500000]
+ optim/total_grad_norm=5.551
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=971,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5061/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=971,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5062/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=971,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5063/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=972,096,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5064/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=972,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5065/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=972,480,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5066/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=972,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5067/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=972,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5068/500000]
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=973,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5069/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=973,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5070/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=973,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5071/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=973,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5072/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=973,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5073/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=974,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5074/500000]
+ train/ActionL1Loss=0.0927
+ throughput/total_tokens=974,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5075/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=974,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5076/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=974,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5077/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=974,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5078/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=974,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5079/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=975,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5080/500000]
+ optim/total_grad_norm=6.563
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=975,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5081/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=975,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5082/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=975,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5083/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=975,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5084/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=976,128,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5085/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=976,320,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5086/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=976,512,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5087/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=976,704,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5088/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=976,896,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5089/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=977,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5090/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=977,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5091/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=977,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5092/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=977,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5093/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=977,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5094/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=978,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5095/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=978,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5096/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=978,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5097/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=978,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5098/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=978,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5099/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=979,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5100/500000]
+ optim/total_grad_norm=7.381
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=979,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5101/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=979,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5102/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=979,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5103/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=979,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5104/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=979,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5105/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=980,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5106/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=980,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5107/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=980,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5108/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=980,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5109/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=980,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5110/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=981,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5111/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=981,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5112/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=981,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5113/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=981,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5114/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=981,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5115/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=982,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5116/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=982,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5117/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=982,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5118/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=982,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5119/500000]
+ train/ActionL1Loss=0.0977
+ throughput/total_tokens=982,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5120/500000]
+ optim/total_grad_norm=6.422
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=983,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5121/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=983,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5122/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=983,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5123/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=983,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5124/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=983,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5125/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=984,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5126/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=984,192,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5127/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=984,384,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5128/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=984,576,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5129/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=984,768,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5130/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=984,960,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5131/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=985,152,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5132/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=985,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5133/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=985,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5134/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=985,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5135/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=985,920,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5136/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=986,112,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5137/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=986,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5138/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=986,496,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5139/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=986,688,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5140/500000]
+ optim/total_grad_norm=6.562
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=986,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5141/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=987,072,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5142/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=987,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5143/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=987,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5144/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=987,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5145/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=987,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5146/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=988,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5147/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=988,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5148/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=988,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5149/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=988,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5150/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=988,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5151/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=988,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5152/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=989,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5153/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=989,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5154/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=989,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5155/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=989,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5156/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=989,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5157/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=990,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5158/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=990,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5159/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=990,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5160/500000]
+ optim/total_grad_norm=5.951
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=990,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5161/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=990,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5162/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=991,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5163/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=991,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5164/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=991,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5165/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=991,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5166/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=991,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5167/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=992,064,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5168/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=992,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5169/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=992,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5170/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=992,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5171/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=992,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5172/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=993,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5173/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=993,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5174/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=993,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5175/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=993,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5176/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=993,792,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5177/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=993,984,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5178/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=994,176,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5179/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=994,368,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5180/500000]
+ optim/total_grad_norm=5.814
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=994,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5181/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=994,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5182/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=994,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5183/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=995,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5184/500000]
+ train/ActionL1Loss=0.0867
+ throughput/total_tokens=995,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5185/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=995,520,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5186/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=995,712,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5187/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=995,904,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5188/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=996,096,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5189/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=996,288,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5190/500000]
+ train/ActionL1Loss=0.0854
+ throughput/total_tokens=996,480,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5191/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=996,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5192/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=996,864,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5193/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=997,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5194/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=997,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5195/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=997,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5196/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=997,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5197/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=997,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5198/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=998,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5199/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=998,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5200/500000]
+ optim/total_grad_norm=6.282
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=998,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5201/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=998,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5202/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=998,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5203/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=998,976,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5204/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=999,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5205/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=999,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5206/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=999,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5207/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=999,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5208/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=999,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5209/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=1,000,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5210/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,000,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=5211/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=1,000,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5212/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=1,000,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5213/500000]
+ train/ActionL1Loss=0.0928
+ throughput/total_tokens=1,000,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5214/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=1,001,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5215/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,001,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5216/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=1,001,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5217/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,001,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5218/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,001,856,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5219/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,002,048,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5220/500000]
+ optim/total_grad_norm=7.165
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,002,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5221/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,002,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5222/500000]
+ train/ActionL1Loss=0.0936
+ throughput/total_tokens=1,002,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5223/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,002,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5224/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,003,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5225/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=1,003,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5226/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,003,392,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5227/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=1,003,584,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5228/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,003,776,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5229/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,003,968,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5230/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=1,004,160,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5231/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,004,352,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5232/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,004,544,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5233/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,004,736,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5234/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,004,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5235/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,005,120,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5236/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=1,005,312,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5237/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=1,005,504,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5238/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,005,696,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5239/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,005,888,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5240/500000]
+ optim/total_grad_norm=6.904
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,006,080,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5241/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,006,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5242/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,006,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5243/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,006,656,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5244/500000]
+ train/ActionL1Loss=0.0845
+ throughput/total_tokens=1,006,848,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5245/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,007,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5246/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,007,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5247/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=1,007,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5248/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=1,007,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5249/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,007,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5250/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,008,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5251/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=1,008,192,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5252/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,008,384,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5253/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,008,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5254/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=1,008,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5255/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=1,008,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5256/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,009,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5257/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=1,009,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5258/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=1,009,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5259/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=1,009,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5260/500000]
+ optim/total_grad_norm=7.898
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,009,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5261/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=1,010,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5262/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=1,010,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5263/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=1,010,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5264/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=1,010,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5265/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=1,010,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5266/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=1,011,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5267/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=1,011,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5268/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=1,011,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5269/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,011,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5270/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=1,011,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5271/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=1,012,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5272/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,012,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5273/500000]
+ train/ActionL1Loss=0.0957
+ throughput/total_tokens=1,012,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5274/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,012,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5275/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,012,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5276/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=1,012,992,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5277/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=1,013,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5278/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=1,013,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5279/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=1,013,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5280/500000]
+ optim/total_grad_norm=7.729
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,013,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5281/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,013,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5282/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,014,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5283/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=1,014,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5284/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,014,528,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5285/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,014,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5286/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,014,912,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5287/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,015,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5288/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,015,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5289/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,015,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5290/500000]
+ train/ActionL1Loss=0.0913
+ throughput/total_tokens=1,015,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5291/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,015,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5292/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,016,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5293/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,016,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5294/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,016,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5295/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,016,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5296/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,016,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5297/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,017,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5298/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,017,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5299/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,017,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5300/500000]
+ optim/total_grad_norm=5.918
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,017,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5301/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=1,017,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5302/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,017,984,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5303/500000]
+ train/ActionL1Loss=0.0882
+ throughput/total_tokens=1,018,176,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5304/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=1,018,368,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5305/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,018,560,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5306/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,018,752,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5307/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,018,944,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5308/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,019,136,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5309/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,019,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5310/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,019,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=5311/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,019,712,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5312/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,019,904,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5313/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,020,096,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5314/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,020,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5315/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,020,480,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5316/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=1,020,672,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5317/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,020,864,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5318/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,021,056,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5319/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,021,248,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5320/500000]
+ optim/total_grad_norm=5.386
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,021,440,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5321/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,021,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5322/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,021,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5323/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,022,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5324/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,022,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5325/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,022,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5326/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,022,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5327/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,022,784,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5328/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=1,022,976,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5329/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,023,168,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5330/500000]
+ train/ActionL1Loss=0.0997
+ throughput/total_tokens=1,023,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5331/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,023,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5332/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,023,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5333/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,023,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5334/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,024,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5335/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,024,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5336/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,024,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5337/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=1,024,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5338/500000]
+ train/ActionL1Loss=0.0943
+ throughput/total_tokens=1,024,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5339/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,025,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5340/500000]
+ optim/total_grad_norm=6.566
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,025,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5341/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,025,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5342/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,025,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5343/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,025,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5344/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,026,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5345/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,026,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5346/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=1,026,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5347/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,026,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5348/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,026,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5349/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,027,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5350/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,027,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5351/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,027,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5352/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=1,027,584,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5353/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,027,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5354/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,027,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5355/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,028,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5356/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,028,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5357/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,028,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5358/500000]
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=1,028,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5359/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,028,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5360/500000]
+ optim/total_grad_norm=5.633
+ train/ActionL1Loss=0.0781
+ throughput/total_tokens=1,029,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5361/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,029,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5362/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,029,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5363/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,029,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5364/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,029,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5365/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,030,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5366/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=1,030,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5367/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,030,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5368/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,030,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5369/500000]
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,030,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5370/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,031,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5371/500000]
+ train/ActionL1Loss=0.0790
+ throughput/total_tokens=1,031,232,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5372/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=1,031,424,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5373/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,031,616,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5374/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,031,808,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5375/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,032,000,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5376/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,032,192,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5377/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,032,384,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5378/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,032,576,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5379/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,032,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5380/500000]
+ optim/total_grad_norm=5.925
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,032,960,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5381/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,033,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5382/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,033,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5383/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=1,033,536,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5384/500000]
+ train/ActionL1Loss=0.0902
+ throughput/total_tokens=1,033,728,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5385/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,033,920,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5386/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,034,112,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5387/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,034,304,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5388/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=1,034,496,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5389/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=1,034,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5390/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,034,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5391/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,035,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5392/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,035,264,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5393/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,035,456,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5394/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,035,648,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5395/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,035,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5396/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,036,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5397/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,036,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5398/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,036,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5399/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,036,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5400/500000]
+ optim/total_grad_norm=7.516
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,036,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5401/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,036,992,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5402/500000]
+ train/ActionL1Loss=0.0778
+ throughput/total_tokens=1,037,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5403/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,037,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5404/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=1,037,568,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=5405/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,037,760,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=5406/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,037,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5407/500000]
+ train/ActionL1Loss=0.0892
+ throughput/total_tokens=1,038,144,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5408/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=1,038,336,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5409/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=1,038,528,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5410/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,038,720,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=5411/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,038,912,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5412/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=1,039,104,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5413/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=1,039,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5414/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=1,039,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5415/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,039,680,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5416/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=1,039,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5417/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,040,064,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5418/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,040,256,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5419/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,040,448,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5420/500000]
+ optim/total_grad_norm=5.191
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,040,640,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5421/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,040,832,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5422/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,041,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5423/500000]
+ train/ActionL1Loss=0.0965
+ throughput/total_tokens=1,041,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5424/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,041,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5425/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,041,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5426/500000]
+ train/ActionL1Loss=0.0918
+ throughput/total_tokens=1,041,792,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5427/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,041,984,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5428/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=1,042,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5429/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,042,368,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5430/500000]
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=1,042,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5431/500000]
+ train/ActionL1Loss=0.0526
+ throughput/total_tokens=1,042,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5432/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,042,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5433/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,043,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5434/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=1,043,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5435/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,043,520,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5436/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=1,043,712,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5437/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,043,904,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5438/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,044,096,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5439/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,044,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5440/500000]
+ optim/total_grad_norm=5.696
+ train/ActionL1Loss=0.0877
+ throughput/total_tokens=1,044,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5441/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,044,672,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5442/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=1,044,864,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5443/500000]
+ train/ActionL1Loss=0.0833
+ throughput/total_tokens=1,045,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5444/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=1,045,248,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5445/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,045,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5446/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,045,632,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5447/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=1,045,824,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5448/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,046,016,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5449/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=1,046,208,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5450/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,046,400,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5451/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,046,592,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5452/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,046,784,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5453/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,046,976,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5454/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=1,047,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5455/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,047,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5456/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,047,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5457/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=1,047,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5458/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,047,936,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5459/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,048,128,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5460/500000]
+ optim/total_grad_norm=5.067
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,048,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=5461/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,048,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5462/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,048,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5463/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,048,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5464/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,049,088,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5465/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,049,280,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5466/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,049,472,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5467/500000]
+ train/ActionL1Loss=0.0774
+ throughput/total_tokens=1,049,664,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5468/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,049,856,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5469/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,050,048,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5470/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,050,240,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5471/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=1,050,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5472/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=1,050,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5473/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,050,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5474/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,051,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5475/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,051,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5476/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,051,392,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5477/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,051,584,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5478/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,051,776,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5479/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,051,968,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5480/500000]
+ optim/total_grad_norm=5.095
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=1,052,160,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5481/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,052,352,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5482/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,052,544,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5483/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,052,736,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5484/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,052,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5485/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,053,120,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5486/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,053,312,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5487/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,053,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5488/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,053,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5489/500000]
+ train/ActionL1Loss=0.0891
+ throughput/total_tokens=1,053,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5490/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,054,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5491/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,054,272,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5492/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,054,464,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5493/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,054,656,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5494/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,054,848,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5495/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,055,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5496/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=1,055,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5497/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,055,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5498/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,055,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5499/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,055,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5500/500000]
+ optim/total_grad_norm=5.667
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,056,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+10/01 [21:36:35] INFO | >> Saving config... checkpoint.py:608
+10/01 [21:37:26] INFO | >> Saving model state... checkpoint.py:796
+10/01 [21:38:38] INFO | >> Saving optim state... checkpoint.py:811
+10/01 [21:40:09] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=5501/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=1,056,192,000
+ throughput/device/tokens_per_second=1,208
+ throughput/device/batches_per_second=0.0504
+[step=5502/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=1,056,384,000
+ throughput/device/tokens_per_second=1,181
+ throughput/device/batches_per_second=0.0492
+[step=5503/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,056,576,000
+ throughput/device/tokens_per_second=1,176
+ throughput/device/batches_per_second=0.0490
+[step=5504/500000]
+ train/ActionL1Loss=0.1006
+ throughput/total_tokens=1,056,768,000
+ throughput/device/tokens_per_second=1,173
+ throughput/device/batches_per_second=0.0489
+[step=5505/500000]
+ train/ActionL1Loss=0.0908
+ throughput/total_tokens=1,056,960,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=5506/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,057,152,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=5507/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,057,344,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=5508/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,057,536,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=5509/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,057,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5510/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,057,920,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5511/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=1,058,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5512/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,058,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5513/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,058,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5514/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,058,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5515/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,058,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5516/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,059,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5517/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=1,059,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5518/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=1,059,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5519/500000]
+ train/ActionL1Loss=0.0895
+ throughput/total_tokens=1,059,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5520/500000]
+ optim/total_grad_norm=5.789
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,059,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5521/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,060,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5522/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,060,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5523/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,060,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5524/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=1,060,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5525/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,060,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5526/500000]
+ train/ActionL1Loss=0.0967
+ throughput/total_tokens=1,060,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5527/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,061,184,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5528/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,061,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5529/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,061,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5530/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,061,760,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5531/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=1,061,952,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5532/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=1,062,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5533/500000]
+ train/ActionL1Loss=0.1037
+ throughput/total_tokens=1,062,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5534/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,062,528,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5535/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,062,720,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5536/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,062,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5537/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,063,104,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5538/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=1,063,296,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5539/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=1,063,488,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5540/500000]
+ optim/total_grad_norm=6.441
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=1,063,680,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5541/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,063,872,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5542/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,064,064,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5543/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=1,064,256,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5544/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,064,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5545/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,064,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5546/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,064,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5547/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,065,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5548/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,065,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5549/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,065,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5550/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,065,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5551/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,065,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5552/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,065,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5553/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,066,176,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5554/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=1,066,368,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5555/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,066,560,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5556/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,066,752,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5557/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,066,944,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5558/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,067,136,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5559/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,067,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5560/500000]
+ optim/total_grad_norm=5.226
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=1,067,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5561/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,067,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5562/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,067,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5563/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,068,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5564/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,068,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5565/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=1,068,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5566/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,068,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5567/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,068,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5568/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,069,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5569/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=1,069,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5570/500000]
+ train/ActionL1Loss=0.0473
+ throughput/total_tokens=1,069,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5571/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,069,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5572/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,069,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5573/500000]
+ train/ActionL1Loss=0.0907
+ throughput/total_tokens=1,070,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5574/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,070,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5575/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,070,400,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5576/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,070,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5577/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=1,070,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5578/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,070,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5579/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,071,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5580/500000]
+ optim/total_grad_norm=4.591
+ train/ActionL1Loss=0.0458
+ throughput/total_tokens=1,071,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5581/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,071,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5582/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=1,071,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5583/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,071,936,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5584/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=1,072,128,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5585/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,072,320,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5586/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,072,512,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5587/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,072,704,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5588/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,072,896,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5589/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,073,088,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5590/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,073,280,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5591/500000]
+ train/ActionL1Loss=0.0926
+ throughput/total_tokens=1,073,472,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5592/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,073,664,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5593/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,073,856,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5594/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,074,048,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5595/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,074,240,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5596/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,074,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5597/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,074,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5598/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=1,074,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5599/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,075,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5600/500000]
+ optim/total_grad_norm=4.339
+ train/ActionL1Loss=0.0899
+ throughput/total_tokens=1,075,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5601/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,075,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5602/500000]
+ train/ActionL1Loss=0.0961
+ throughput/total_tokens=1,075,584,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5603/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=1,075,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5604/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=1,075,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5605/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,076,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5606/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,076,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5607/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,076,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5608/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,076,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5609/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,076,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5610/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,077,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5611/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,077,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5612/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,077,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5613/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,077,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5614/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,077,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5615/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=1,078,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5616/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,078,272,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5617/500000]
+ train/ActionL1Loss=0.0878
+ throughput/total_tokens=1,078,464,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5618/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=1,078,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5619/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,078,848,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5620/500000]
+ optim/total_grad_norm=7.582
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,079,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5621/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,079,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5622/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,079,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5623/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,079,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5624/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,079,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5625/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,080,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5626/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,080,192,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5627/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,080,384,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5628/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,080,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5629/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,080,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5630/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,080,960,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5631/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,081,152,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=5632/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,081,344,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5633/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,081,536,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5634/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,081,728,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5635/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,081,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5636/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,082,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5637/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,082,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5638/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,082,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5639/500000]
+ train/ActionL1Loss=0.0468
+ throughput/total_tokens=1,082,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5640/500000]
+ optim/total_grad_norm=4.931
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,082,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5641/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,083,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5642/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,083,264,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5643/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,083,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5644/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=1,083,648,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5645/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,083,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5646/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,084,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5647/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,084,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5648/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,084,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5649/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,084,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5650/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,084,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5651/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,084,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5652/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,085,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5653/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,085,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5654/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,085,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5655/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,085,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5656/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,085,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5657/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,086,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5658/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,086,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5659/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,086,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5660/500000]
+ optim/total_grad_norm=7.552
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,086,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5661/500000]
+ train/ActionL1Loss=0.0526
+ throughput/total_tokens=1,086,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5662/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=1,087,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5663/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,087,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5664/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,087,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5665/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,087,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5666/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,087,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5667/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,088,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5668/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,088,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5669/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,088,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5670/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=1,088,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5671/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,088,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5672/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,089,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5673/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,089,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5674/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,089,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5675/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=1,089,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5676/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,089,792,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5677/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,089,984,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5678/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,090,176,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5679/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,090,368,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5680/500000]
+ optim/total_grad_norm=7.175
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=1,090,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5681/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,090,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5682/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,090,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5683/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,091,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5684/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,091,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5685/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,091,520,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5686/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,091,712,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5687/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,091,904,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5688/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,092,096,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5689/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,092,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5690/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,092,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5691/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,092,672,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5692/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,092,864,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5693/500000]
+ train/ActionL1Loss=0.0919
+ throughput/total_tokens=1,093,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5694/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=1,093,248,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5695/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,093,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5696/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,093,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5697/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,093,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5698/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,094,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5699/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,094,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5700/500000]
+ optim/total_grad_norm=5.973
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,094,400,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5701/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,094,592,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5702/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,094,784,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5703/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=1,094,976,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5704/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,095,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5705/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,095,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5706/500000]
+ train/ActionL1Loss=0.0889
+ throughput/total_tokens=1,095,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5707/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=1,095,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5708/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,095,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5709/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,096,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5710/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=1,096,320,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5711/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,096,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5712/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,096,704,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5713/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,096,896,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5714/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,097,088,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5715/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=1,097,280,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5716/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,097,472,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5717/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,097,664,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5718/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,097,856,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5719/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,098,048,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5720/500000]
+ optim/total_grad_norm=5.560
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,098,240,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5721/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=1,098,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5722/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,098,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5723/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,098,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5724/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,099,008,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5725/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=1,099,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5726/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,099,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5727/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,099,584,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5728/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,099,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5729/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=1,099,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5730/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,100,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5731/500000]
+ train/ActionL1Loss=0.0765
+ throughput/total_tokens=1,100,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5732/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,100,544,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5733/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,100,736,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5734/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,100,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5735/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,101,120,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5736/500000]
+ train/ActionL1Loss=0.0856
+ throughput/total_tokens=1,101,312,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5737/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,101,504,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5738/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,101,696,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5739/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,101,888,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5740/500000]
+ optim/total_grad_norm=7.220
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,102,080,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5741/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,102,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5742/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,102,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5743/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,102,656,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5744/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=1,102,848,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5745/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=1,103,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5746/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=1,103,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5747/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,103,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5748/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,103,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5749/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,103,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5750/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=1,104,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5751/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,104,192,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5752/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,104,384,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5753/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,104,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5754/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=1,104,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5755/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,104,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5756/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,105,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5757/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,105,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5758/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,105,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5759/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,105,728,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5760/500000]
+ optim/total_grad_norm=4.860
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,105,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5761/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,106,112,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5762/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,106,304,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5763/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,106,496,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5764/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,106,688,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5765/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=1,106,880,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5766/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,107,072,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5767/500000]
+ train/ActionL1Loss=0.1093
+ throughput/total_tokens=1,107,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5768/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=1,107,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5769/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=1,107,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5770/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,107,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5771/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,108,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5772/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,108,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5773/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=1,108,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5774/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,108,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5775/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,108,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5776/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,108,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5777/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=1,109,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5778/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,109,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5779/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,109,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5780/500000]
+ optim/total_grad_norm=5.805
+ train/ActionL1Loss=0.0841
+ throughput/total_tokens=1,109,760,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5781/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=1,109,952,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5782/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,110,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5783/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,110,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5784/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,110,528,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5785/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,110,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5786/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,110,912,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5787/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,111,104,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5788/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,111,296,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5789/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=1,111,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5790/500000]
+ train/ActionL1Loss=0.0864
+ throughput/total_tokens=1,111,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5791/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,111,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5792/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,112,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5793/500000]
+ train/ActionL1Loss=0.0920
+ throughput/total_tokens=1,112,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5794/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,112,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5795/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,112,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5796/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,112,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5797/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,113,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5798/500000]
+ train/ActionL1Loss=0.0839
+ throughput/total_tokens=1,113,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5799/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=1,113,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5800/500000]
+ optim/total_grad_norm=6.939
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,113,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5801/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,113,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5802/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,113,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5803/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,114,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5804/500000]
+ train/ActionL1Loss=0.0969
+ throughput/total_tokens=1,114,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5805/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=1,114,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5806/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,114,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5807/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,114,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5808/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,115,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5809/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=1,115,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5810/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=1,115,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=5811/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,115,712,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5812/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,115,904,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5813/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,116,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5814/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,116,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5815/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,116,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5816/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=1,116,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5817/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,116,864,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5818/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,117,056,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5819/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,117,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5820/500000]
+ optim/total_grad_norm=6.581
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=1,117,440,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5821/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,117,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5822/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,117,824,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5823/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,118,016,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5824/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,118,208,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5825/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,118,400,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5826/500000]
+ train/ActionL1Loss=0.0488
+ throughput/total_tokens=1,118,592,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5827/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,118,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5828/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,118,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5829/500000]
+ train/ActionL1Loss=0.0848
+ throughput/total_tokens=1,119,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5830/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,119,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5831/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,119,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5832/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,119,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5833/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,119,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5834/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,120,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5835/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,120,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5836/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,120,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5837/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,120,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5838/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,120,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5839/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,121,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5840/500000]
+ optim/total_grad_norm=4.631
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,121,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5841/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,121,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5842/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=1,121,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5843/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=1,121,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5844/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=1,122,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5845/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,122,240,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5846/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=1,122,432,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5847/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,122,624,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5848/500000]
+ train/ActionL1Loss=0.1008
+ throughput/total_tokens=1,122,816,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5849/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,123,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5850/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,123,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5851/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,123,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5852/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,123,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5853/500000]
+ train/ActionL1Loss=0.0779
+ throughput/total_tokens=1,123,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5854/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,123,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5855/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,124,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5856/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,124,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5857/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,124,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5858/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=1,124,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5859/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=1,124,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5860/500000]
+ optim/total_grad_norm=6.363
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,125,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5861/500000]
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=1,125,312,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5862/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,125,504,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5863/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=1,125,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5864/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=1,125,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5865/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=1,126,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5866/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,126,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5867/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,126,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5868/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,126,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5869/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,126,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5870/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=1,127,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5871/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,127,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5872/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,127,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5873/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=1,127,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5874/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,127,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5875/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,128,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5876/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,128,192,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5877/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,128,384,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5878/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,128,576,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5879/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,128,768,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5880/500000]
+ optim/total_grad_norm=4.804
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,128,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5881/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,129,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5882/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,129,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5883/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,129,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5884/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,129,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5885/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,129,920,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5886/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,130,112,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5887/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,130,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5888/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,130,496,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5889/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,130,688,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5890/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,130,880,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5891/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=1,131,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5892/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,131,264,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5893/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=1,131,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=5894/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,131,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5895/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=1,131,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5896/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,132,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5897/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,132,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5898/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,132,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5899/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,132,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5900/500000]
+ optim/total_grad_norm=6.576
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,132,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5901/500000]
+ train/ActionL1Loss=0.0437
+ throughput/total_tokens=1,132,992,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5902/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=1,133,184,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5903/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,133,376,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5904/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,133,568,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5905/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,133,760,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5906/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,133,952,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5907/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,134,144,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=5908/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,134,336,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5909/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,134,528,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=5910/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=1,134,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5911/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,134,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5912/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=1,135,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5913/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,135,296,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5914/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,135,488,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5915/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,135,680,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5916/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,135,872,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=5917/500000]
+ train/ActionL1Loss=0.0774
+ throughput/total_tokens=1,136,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5918/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,136,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5919/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=1,136,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5920/500000]
+ optim/total_grad_norm=5.800
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,136,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5921/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,136,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5922/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,137,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5923/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,137,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5924/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=1,137,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5925/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,137,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5926/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=1,137,792,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5927/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,137,984,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5928/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,138,176,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5929/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,138,368,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5930/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,138,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5931/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,138,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5932/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=1,138,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5933/500000]
+ train/ActionL1Loss=0.0526
+ throughput/total_tokens=1,139,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5934/500000]
+ train/ActionL1Loss=0.0505
+ throughput/total_tokens=1,139,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5935/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,139,520,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5936/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,139,712,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5937/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,139,904,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5938/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,140,096,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5939/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,140,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5940/500000]
+ optim/total_grad_norm=6.121
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,140,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5941/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=1,140,672,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=5942/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,140,864,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5943/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,141,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5944/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,141,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5945/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,141,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5946/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,141,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5947/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,141,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5948/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,142,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5949/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=1,142,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5950/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,142,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5951/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,142,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5952/500000]
+ train/ActionL1Loss=0.0863
+ throughput/total_tokens=1,142,784,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5953/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=1,142,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5954/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,143,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5955/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=1,143,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5956/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=1,143,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5957/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,143,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5958/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=1,143,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5959/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,144,128,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5960/500000]
+ optim/total_grad_norm=7.231
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,144,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=5961/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,144,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5962/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=1,144,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5963/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,144,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5964/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=1,145,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5965/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,145,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5966/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,145,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5967/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,145,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5968/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=1,145,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=5969/500000]
+ train/ActionL1Loss=0.0492
+ throughput/total_tokens=1,146,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=5970/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=1,146,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5971/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,146,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5972/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,146,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5973/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,146,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5974/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,147,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5975/500000]
+ train/ActionL1Loss=0.0488
+ throughput/total_tokens=1,147,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5976/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,147,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5977/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,147,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5978/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,147,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5979/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=1,147,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5980/500000]
+ optim/total_grad_norm=5.722
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,148,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=5981/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,148,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5982/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,148,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5983/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,148,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=5984/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=1,148,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5985/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,149,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5986/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,149,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5987/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,149,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5988/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=1,149,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5989/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=1,149,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=5990/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,150,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=5991/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,150,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5992/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,150,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5993/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,150,656,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5994/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,150,848,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5995/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,151,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5996/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,151,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5997/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,151,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5998/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,151,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=5999/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,151,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6000/500000]
+ optim/total_grad_norm=6.525
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,152,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+10/02 [00:33:13] INFO | >> Saving config... checkpoint.py:608
+10/02 [00:34:03] INFO | >> Saving model state... checkpoint.py:796
+10/02 [00:35:10] INFO | >> Saving optim state... checkpoint.py:811
+10/02 [00:36:42] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=6001/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,152,192,000
+ throughput/device/tokens_per_second=1,208
+ throughput/device/batches_per_second=0.0503
+[step=6002/500000]
+ train/ActionL1Loss=0.0774
+ throughput/total_tokens=1,152,384,000
+ throughput/device/tokens_per_second=1,174
+ throughput/device/batches_per_second=0.0489
+[step=6003/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,152,576,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=6004/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,152,768,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=6005/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=1,152,960,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=6006/500000]
+ train/ActionL1Loss=0.0828
+ throughput/total_tokens=1,153,152,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=6007/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,153,344,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6008/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,153,536,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6009/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,153,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6010/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=1,153,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6011/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,154,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6012/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=1,154,304,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6013/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,154,496,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6014/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,154,688,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6015/500000]
+ train/ActionL1Loss=0.0807
+ throughput/total_tokens=1,154,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6016/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,155,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6017/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,155,264,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6018/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,155,456,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6019/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,155,648,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6020/500000]
+ optim/total_grad_norm=8.293
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,155,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6021/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,156,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6022/500000]
+ train/ActionL1Loss=0.0860
+ throughput/total_tokens=1,156,224,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6023/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,156,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6024/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,156,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6025/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,156,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6026/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=1,156,992,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6027/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,157,184,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6028/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,157,376,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6029/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,157,568,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6030/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=1,157,760,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6031/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,157,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6032/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,158,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6033/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,158,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6034/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,158,528,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6035/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,158,720,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6036/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,158,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6037/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=1,159,104,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6038/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,159,296,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6039/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,159,488,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6040/500000]
+ optim/total_grad_norm=4.972
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,159,680,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6041/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,159,872,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6042/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,160,064,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6043/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,160,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6044/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,160,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6045/500000]
+ train/ActionL1Loss=0.0482
+ throughput/total_tokens=1,160,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6046/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,160,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6047/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=1,161,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6048/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,161,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6049/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=1,161,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6050/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,161,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6051/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,161,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6052/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,161,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6053/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=1,162,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6054/500000]
+ train/ActionL1Loss=0.0492
+ throughput/total_tokens=1,162,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6055/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,162,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6056/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=1,162,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6057/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,162,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6058/500000]
+ train/ActionL1Loss=0.0567
+ throughput/total_tokens=1,163,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6059/500000]
+ train/ActionL1Loss=0.0855
+ throughput/total_tokens=1,163,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6060/500000]
+ optim/total_grad_norm=5.275
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,163,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6061/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=1,163,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6062/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,163,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6063/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,164,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6064/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,164,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6065/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,164,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6066/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,164,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6067/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,164,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6068/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,165,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6069/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,165,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6070/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,165,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6071/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,165,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6072/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=1,165,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6073/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,166,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6074/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,166,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6075/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,166,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6076/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,166,592,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6077/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=1,166,784,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6078/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=1,166,976,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6079/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,167,168,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6080/500000]
+ optim/total_grad_norm=6.036
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,167,360,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6081/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,167,552,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6082/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,167,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6083/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,167,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6084/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,168,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6085/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,168,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6086/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=1,168,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6087/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,168,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6088/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,168,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6089/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,169,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6090/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,169,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6091/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,169,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6092/500000]
+ train/ActionL1Loss=0.0480
+ throughput/total_tokens=1,169,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6093/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,169,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6094/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,170,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6095/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,170,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6096/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,170,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6097/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=1,170,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6098/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,170,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6099/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=1,171,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6100/500000]
+ optim/total_grad_norm=6.548
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,171,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6101/500000]
+ train/ActionL1Loss=0.0508
+ throughput/total_tokens=1,171,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6102/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,171,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6103/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=1,171,776,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6104/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,171,968,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6105/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,172,160,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6106/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,172,352,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6107/500000]
+ train/ActionL1Loss=0.0792
+ throughput/total_tokens=1,172,544,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6108/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,172,736,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6109/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,172,928,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6110/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,173,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6111/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,173,312,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6112/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,173,504,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6113/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,173,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6114/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,173,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6115/500000]
+ train/ActionL1Loss=0.0921
+ throughput/total_tokens=1,174,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6116/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,174,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6117/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,174,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6118/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,174,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6119/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,174,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6120/500000]
+ optim/total_grad_norm=4.858
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,175,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6121/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,175,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6122/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,175,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6123/500000]
+ train/ActionL1Loss=0.0846
+ throughput/total_tokens=1,175,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6124/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,175,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6125/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,176,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6126/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,176,192,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6127/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,176,384,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6128/500000]
+ train/ActionL1Loss=0.0942
+ throughput/total_tokens=1,176,576,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6129/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,176,768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6130/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,176,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6131/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,177,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6132/500000]
+ train/ActionL1Loss=0.0527
+ throughput/total_tokens=1,177,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6133/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,177,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6134/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,177,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6135/500000]
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,177,920,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6136/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,178,112,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6137/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,178,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6138/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,178,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6139/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,178,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6140/500000]
+ optim/total_grad_norm=7.645
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,178,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6141/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,179,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6142/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,179,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6143/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,179,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6144/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,179,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6145/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,179,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6146/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,180,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6147/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,180,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6148/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,180,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6149/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,180,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6150/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,180,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6151/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,180,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6152/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,181,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6153/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,181,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6154/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,181,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6155/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,181,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6156/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,181,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6157/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=1,182,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6158/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=1,182,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6159/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,182,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6160/500000]
+ optim/total_grad_norm=6.274
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,182,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6161/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,182,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6162/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,183,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6163/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,183,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6164/500000]
+ train/ActionL1Loss=0.0919
+ throughput/total_tokens=1,183,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6165/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,183,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6166/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,183,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6167/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=1,184,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6168/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,184,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6169/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,184,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6170/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,184,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6171/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,184,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6172/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,185,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6173/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,185,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6174/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,185,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6175/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,185,600,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6176/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,185,792,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6177/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=1,185,984,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6178/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,186,176,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6179/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,186,368,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6180/500000]
+ optim/total_grad_norm=6.037
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,186,560,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6181/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,186,752,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6182/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,186,944,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6183/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,187,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6184/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,187,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6185/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,187,520,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6186/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,187,712,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6187/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,187,904,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6188/500000]
+ train/ActionL1Loss=0.0472
+ throughput/total_tokens=1,188,096,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6189/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,188,288,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6190/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,188,480,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6191/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,188,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6192/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,188,864,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6193/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,189,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6194/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=1,189,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6195/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=1,189,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6196/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,189,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6197/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,189,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6198/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,190,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6199/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,190,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6200/500000]
+ optim/total_grad_norm=5.801
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,190,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6201/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,190,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6202/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,190,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6203/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,190,976,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6204/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,191,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6205/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,191,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6206/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,191,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6207/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,191,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6208/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,191,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6209/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,192,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6210/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=1,192,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6211/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,192,512,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6212/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,192,704,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6213/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,192,896,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6214/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,193,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6215/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,193,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6216/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,193,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6217/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,193,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6218/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,193,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6219/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,194,048,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6220/500000]
+ optim/total_grad_norm=8.126
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,194,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6221/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=1,194,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6222/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,194,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6223/500000]
+ train/ActionL1Loss=0.0775
+ throughput/total_tokens=1,194,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6224/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,195,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6225/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,195,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6226/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,195,392,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6227/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,195,584,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6228/500000]
+ train/ActionL1Loss=0.0892
+ throughput/total_tokens=1,195,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6229/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=1,195,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6230/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,196,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6231/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,196,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6232/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,196,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6233/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,196,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6234/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,196,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6235/500000]
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=1,197,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6236/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,197,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6237/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,197,504,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6238/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,197,696,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6239/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,197,888,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6240/500000]
+ optim/total_grad_norm=5.164
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,198,080,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6241/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,198,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6242/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,198,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6243/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,198,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6244/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,198,848,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6245/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,199,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6246/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,199,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6247/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,199,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6248/500000]
+ train/ActionL1Loss=0.0508
+ throughput/total_tokens=1,199,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6249/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,199,808,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6250/500000]
+ train/ActionL1Loss=0.0527
+ throughput/total_tokens=1,200,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6251/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,200,192,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6252/500000]
+ train/ActionL1Loss=0.0443
+ throughput/total_tokens=1,200,384,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6253/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,200,576,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6254/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,200,768,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6255/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,200,960,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6256/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=1,201,152,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6257/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=1,201,344,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6258/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,201,536,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6259/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,201,728,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6260/500000]
+ optim/total_grad_norm=6.735
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,201,920,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6261/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=1,202,112,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6262/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,202,304,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6263/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,202,496,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6264/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,202,688,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6265/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,202,880,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6266/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,203,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6267/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,203,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6268/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,203,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6269/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=1,203,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6270/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,203,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6271/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,204,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6272/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,204,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6273/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,204,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6274/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,204,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6275/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,204,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6276/500000]
+ train/ActionL1Loss=0.0416
+ throughput/total_tokens=1,204,992,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6277/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,205,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6278/500000]
+ train/ActionL1Loss=0.0508
+ throughput/total_tokens=1,205,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6279/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,205,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6280/500000]
+ optim/total_grad_norm=6.073
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,205,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6281/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,205,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6282/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,206,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6283/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,206,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6284/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,206,528,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6285/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,206,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6286/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=1,206,912,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6287/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,207,104,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6288/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,207,296,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6289/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=1,207,488,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6290/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,207,680,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6291/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,207,872,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6292/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=1,208,064,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6293/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,208,256,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6294/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,208,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6295/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,208,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6296/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,208,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6297/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,209,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6298/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,209,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6299/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=1,209,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6300/500000]
+ optim/total_grad_norm=6.784
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,209,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6301/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=1,209,792,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6302/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,209,984,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6303/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,210,176,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6304/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,210,368,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6305/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,210,560,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6306/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,210,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6307/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,210,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6308/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,211,136,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6309/500000]
+ train/ActionL1Loss=0.1023
+ throughput/total_tokens=1,211,328,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6310/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=1,211,520,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=6311/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,211,712,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6312/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,211,904,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6313/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,212,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6314/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,212,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6315/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,212,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6316/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,212,672,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=6317/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,212,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6318/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,213,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6319/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,213,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6320/500000]
+ optim/total_grad_norm=7.189
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,213,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6321/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,213,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6322/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,213,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6323/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,214,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6324/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,214,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6325/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,214,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6326/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,214,592,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6327/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,214,784,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6328/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,214,976,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6329/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,215,168,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6330/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,215,360,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6331/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,215,552,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6332/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,215,744,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6333/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,215,936,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6334/500000]
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,216,128,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6335/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,216,320,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6336/500000]
+ train/ActionL1Loss=0.0813
+ throughput/total_tokens=1,216,512,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6337/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,216,704,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6338/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,216,896,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6339/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=1,217,088,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6340/500000]
+ optim/total_grad_norm=6.386
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,217,280,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6341/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,217,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6342/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,217,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6343/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,217,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6344/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,218,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6345/500000]
+ train/ActionL1Loss=0.0949
+ throughput/total_tokens=1,218,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6346/500000]
+ train/ActionL1Loss=0.0883
+ throughput/total_tokens=1,218,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6347/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,218,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6348/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,218,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6349/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,219,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6350/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,219,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6351/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,219,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6352/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,219,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6353/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,219,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6354/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,219,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6355/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,220,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6356/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,220,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6357/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,220,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6358/500000]
+ train/ActionL1Loss=0.0450
+ throughput/total_tokens=1,220,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6359/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,220,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6360/500000]
+ optim/total_grad_norm=5.073
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=1,221,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6361/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,221,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6362/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=1,221,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6363/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,221,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6364/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,221,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6365/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,222,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6366/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,222,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6367/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,222,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6368/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,222,656,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6369/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,222,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6370/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,223,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6371/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,223,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6372/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,223,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6373/500000]
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,223,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6374/500000]
+ train/ActionL1Loss=0.0858
+ throughput/total_tokens=1,223,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6375/500000]
+ train/ActionL1Loss=0.0454
+ throughput/total_tokens=1,224,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6376/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,224,192,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6377/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,224,384,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6378/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,224,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6379/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,224,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6380/500000]
+ optim/total_grad_norm=6.973
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,224,960,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6381/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,225,152,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6382/500000]
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=1,225,344,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6383/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,225,536,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6384/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,225,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6385/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=1,225,920,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6386/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,226,112,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6387/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,226,304,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6388/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,226,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6389/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,226,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6390/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,226,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6391/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,227,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6392/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,227,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6393/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,227,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6394/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,227,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6395/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,227,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6396/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,228,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6397/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,228,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6398/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,228,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6399/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,228,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6400/500000]
+ optim/total_grad_norm=6.286
+ train/ActionL1Loss=0.1102
+ throughput/total_tokens=1,228,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6401/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,228,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6402/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,229,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6403/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,229,376,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6404/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,229,568,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6405/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,229,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6406/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,229,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6407/500000]
+ train/ActionL1Loss=0.0505
+ throughput/total_tokens=1,230,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6408/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,230,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6409/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,230,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6410/500000]
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,230,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6411/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,230,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6412/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,231,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6413/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,231,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6414/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,231,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6415/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,231,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6416/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,231,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6417/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,232,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6418/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,232,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6419/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,232,448,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6420/500000]
+ optim/total_grad_norm=5.885
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,232,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6421/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,232,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6422/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=1,233,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6423/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,233,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6424/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,233,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6425/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,233,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6426/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,233,792,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6427/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,233,984,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6428/500000]
+ train/ActionL1Loss=0.0861
+ throughput/total_tokens=1,234,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6429/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,234,368,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6430/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,234,560,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6431/500000]
+ train/ActionL1Loss=0.0515
+ throughput/total_tokens=1,234,752,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6432/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,234,944,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6433/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,235,136,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6434/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,235,328,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6435/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,235,520,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6436/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,235,712,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6437/500000]
+ train/ActionL1Loss=0.0467
+ throughput/total_tokens=1,235,904,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6438/500000]
+ train/ActionL1Loss=0.0839
+ throughput/total_tokens=1,236,096,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6439/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=1,236,288,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6440/500000]
+ optim/total_grad_norm=5.990
+ train/ActionL1Loss=0.0786
+ throughput/total_tokens=1,236,480,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6441/500000]
+ train/ActionL1Loss=0.0487
+ throughput/total_tokens=1,236,672,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6442/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,236,864,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6443/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,237,056,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6444/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,237,248,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6445/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,237,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6446/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,237,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6447/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,237,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6448/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,238,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6449/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,238,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6450/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,238,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6451/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,238,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6452/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,238,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6453/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,238,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6454/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,239,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6455/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,239,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6456/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,239,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6457/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,239,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6458/500000]
+ train/ActionL1Loss=0.0439
+ throughput/total_tokens=1,239,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6459/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,240,128,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6460/500000]
+ optim/total_grad_norm=7.166
+ train/ActionL1Loss=0.0901
+ throughput/total_tokens=1,240,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6461/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,240,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6462/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,240,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6463/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,240,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6464/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,241,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6465/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,241,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6466/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,241,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6467/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,241,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6468/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=1,241,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6469/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,242,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6470/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,242,240,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6471/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,242,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6472/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,242,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6473/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,242,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6474/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,243,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6475/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,243,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6476/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,243,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6477/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,243,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6478/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,243,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6479/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=1,243,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6480/500000]
+ optim/total_grad_norm=6.535
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,244,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6481/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,244,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6482/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,244,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6483/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,244,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6484/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,244,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6485/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,245,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6486/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,245,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6487/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=1,245,504,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6488/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,245,696,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6489/500000]
+ train/ActionL1Loss=0.0438
+ throughput/total_tokens=1,245,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6490/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,246,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6491/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,246,272,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6492/500000]
+ train/ActionL1Loss=0.0910
+ throughput/total_tokens=1,246,464,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6493/500000]
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=1,246,656,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6494/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,246,848,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6495/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,247,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6496/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,247,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6497/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=1,247,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6498/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,247,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6499/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=1,247,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6500/500000]
+ optim/total_grad_norm=9.072
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,248,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+10/02 [03:29:54] INFO | >> Saving config... checkpoint.py:608
+10/02 [03:30:22] INFO | >> Saving model state... checkpoint.py:796
+10/02 [03:31:34] INFO | >> Saving optim state... checkpoint.py:811
+10/02 [03:33:08] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=6501/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,248,192,000
+ throughput/device/tokens_per_second=1,209
+ throughput/device/batches_per_second=0.0504
+[step=6502/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,248,384,000
+ throughput/device/tokens_per_second=1,185
+ throughput/device/batches_per_second=0.0494
+[step=6503/500000]
+ train/ActionL1Loss=0.0470
+ throughput/total_tokens=1,248,576,000
+ throughput/device/tokens_per_second=1,179
+ throughput/device/batches_per_second=0.0492
+[step=6504/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,248,768,000
+ throughput/device/tokens_per_second=1,176
+ throughput/device/batches_per_second=0.0490
+[step=6505/500000]
+ train/ActionL1Loss=0.0480
+ throughput/total_tokens=1,248,960,000
+ throughput/device/tokens_per_second=1,174
+ throughput/device/batches_per_second=0.0489
+[step=6506/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,249,152,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=6507/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,249,344,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=6508/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,249,536,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=6509/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=1,249,728,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=6510/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,249,920,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6511/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,250,112,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=6512/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,250,304,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6513/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,250,496,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6514/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,250,688,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6515/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,250,880,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6516/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,251,072,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6517/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,251,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6518/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,251,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6519/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,251,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6520/500000]
+ optim/total_grad_norm=7.341
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,251,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6521/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,252,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6522/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,252,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6523/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,252,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6524/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,252,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6525/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,252,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6526/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,252,992,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6527/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,253,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6528/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,253,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6529/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,253,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6530/500000]
+ train/ActionL1Loss=0.0500
+ throughput/total_tokens=1,253,760,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6531/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,253,952,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6532/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,254,144,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6533/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,254,336,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6534/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,254,528,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6535/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,254,720,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6536/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=1,254,912,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6537/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,255,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6538/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,255,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6539/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,255,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6540/500000]
+ optim/total_grad_norm=5.022
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,255,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6541/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,255,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6542/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,256,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6543/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,256,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6544/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,256,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6545/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,256,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6546/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,256,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6547/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,257,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6548/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,257,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6549/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=1,257,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6550/500000]
+ train/ActionL1Loss=0.0796
+ throughput/total_tokens=1,257,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6551/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,257,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6552/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,257,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6553/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=1,258,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6554/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,258,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6555/500000]
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,258,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6556/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,258,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6557/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,258,944,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6558/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,259,136,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6559/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=1,259,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6560/500000]
+ optim/total_grad_norm=6.346
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,259,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6561/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,259,712,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6562/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,259,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6563/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,260,096,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6564/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=1,260,288,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6565/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,260,480,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6566/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=1,260,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6567/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,260,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6568/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,261,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6569/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,261,248,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6570/500000]
+ train/ActionL1Loss=0.0440
+ throughput/total_tokens=1,261,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6571/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,261,632,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6572/500000]
+ train/ActionL1Loss=0.0842
+ throughput/total_tokens=1,261,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6573/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,262,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6574/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=1,262,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6575/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,262,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6576/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,262,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6577/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,262,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6578/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,262,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6579/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,263,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6580/500000]
+ optim/total_grad_norm=6.072
+ train/ActionL1Loss=0.0692
+ throughput/total_tokens=1,263,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6581/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,263,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6582/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,263,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6583/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,263,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6584/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,264,128,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6585/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,264,320,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6586/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,264,512,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6587/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,264,704,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6588/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,264,896,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6589/500000]
+ train/ActionL1Loss=0.0489
+ throughput/total_tokens=1,265,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6590/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,265,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6591/500000]
+ train/ActionL1Loss=0.0450
+ throughput/total_tokens=1,265,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6592/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,265,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6593/500000]
+ train/ActionL1Loss=0.0533
+ throughput/total_tokens=1,265,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6594/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,266,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6595/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,266,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6596/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,266,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6597/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,266,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6598/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,266,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6599/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=1,267,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6600/500000]
+ optim/total_grad_norm=5.483
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,267,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6601/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=1,267,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6602/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,267,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6603/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,267,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6604/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,267,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6605/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,268,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6606/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,268,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6607/500000]
+ train/ActionL1Loss=0.0488
+ throughput/total_tokens=1,268,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6608/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=1,268,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6609/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,268,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6610/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,269,120,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6611/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,269,312,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6612/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,269,504,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6613/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,269,696,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6614/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,269,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6615/500000]
+ train/ActionL1Loss=0.0937
+ throughput/total_tokens=1,270,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6616/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,270,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6617/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,270,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6618/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,270,656,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6619/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,270,848,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6620/500000]
+ optim/total_grad_norm=4.335
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,271,040,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6621/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,271,232,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6622/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,271,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6623/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,271,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6624/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,271,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6625/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=1,272,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6626/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,272,192,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6627/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,272,384,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6628/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,272,576,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6629/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,272,768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6630/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=1,272,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6631/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,273,152,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6632/500000]
+ train/ActionL1Loss=0.0750
+ throughput/total_tokens=1,273,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6633/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,273,536,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6634/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,273,728,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6635/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,273,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6636/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,274,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6637/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,274,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6638/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,274,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6639/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,274,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6640/500000]
+ optim/total_grad_norm=6.942
+ train/ActionL1Loss=0.0466
+ throughput/total_tokens=1,274,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6641/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,275,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6642/500000]
+ train/ActionL1Loss=0.0837
+ throughput/total_tokens=1,275,264,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6643/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,275,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6644/500000]
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=1,275,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6645/500000]
+ train/ActionL1Loss=0.0486
+ throughput/total_tokens=1,275,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6646/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,276,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6647/500000]
+ train/ActionL1Loss=0.0984
+ throughput/total_tokens=1,276,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6648/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,276,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6649/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,276,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6650/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,276,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6651/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,276,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6652/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,277,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6653/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,277,376,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6654/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,277,568,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6655/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,277,760,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6656/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,277,952,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6657/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,278,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6658/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,278,336,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6659/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,278,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6660/500000]
+ optim/total_grad_norm=8.431
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,278,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6661/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,278,912,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6662/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,279,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6663/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,279,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6664/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,279,488,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6665/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,279,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6666/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,279,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6667/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,280,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6668/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,280,256,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6669/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,280,448,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6670/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,280,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6671/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,280,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6672/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,281,024,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6673/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,281,216,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6674/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,281,408,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6675/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,281,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6676/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,281,792,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6677/500000]
+ train/ActionL1Loss=0.0451
+ throughput/total_tokens=1,281,984,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6678/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,282,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6679/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,282,368,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6680/500000]
+ optim/total_grad_norm=5.009
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,282,560,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6681/500000]
+ train/ActionL1Loss=0.0523
+ throughput/total_tokens=1,282,752,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6682/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,282,944,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6683/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,283,136,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6684/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,283,328,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6685/500000]
+ train/ActionL1Loss=0.0486
+ throughput/total_tokens=1,283,520,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6686/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,283,712,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6687/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,283,904,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6688/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,284,096,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6689/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,284,288,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6690/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,284,480,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6691/500000]
+ train/ActionL1Loss=0.0488
+ throughput/total_tokens=1,284,672,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6692/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,284,864,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6693/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,285,056,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6694/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,285,248,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6695/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,285,440,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6696/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,285,632,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6697/500000]
+ train/ActionL1Loss=0.0810
+ throughput/total_tokens=1,285,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6698/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,286,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6699/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,286,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6700/500000]
+ optim/total_grad_norm=5.388
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,286,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6701/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,286,592,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6702/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,286,784,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6703/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,286,976,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6704/500000]
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,287,168,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6705/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,287,360,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6706/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,287,552,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6707/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=1,287,744,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0480
+[step=6708/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,287,936,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0479
+[step=6709/500000]
+ train/ActionL1Loss=0.0498
+ throughput/total_tokens=1,288,128,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0479
+[step=6710/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,288,320,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0479
+ System/Peak GPU Memory (MB)=46,917
+[step=6711/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,288,512,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0479
+[step=6712/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,288,704,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0479
+[step=6713/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,288,896,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0480
+[step=6714/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,289,088,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0480
+[step=6715/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,289,280,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0480
+[step=6716/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,289,472,000
+ throughput/device/tokens_per_second=1,150
+ throughput/device/batches_per_second=0.0480
+[step=6717/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,289,664,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6718/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,289,856,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6719/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,290,048,000
+ throughput/device/tokens_per_second=1,151
+ throughput/device/batches_per_second=0.0480
+[step=6720/500000]
+ optim/total_grad_norm=5.734
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,290,240,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6721/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,290,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6722/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,290,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6723/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,290,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6724/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,291,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6725/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,291,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6726/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,291,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6727/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,291,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6728/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,291,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6729/500000]
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,291,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6730/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=1,292,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6731/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,292,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6732/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,292,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6733/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=1,292,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6734/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,292,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6735/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,293,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6736/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,293,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6737/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,293,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6738/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,293,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6739/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=1,293,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6740/500000]
+ optim/total_grad_norm=7.610
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,294,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6741/500000]
+ train/ActionL1Loss=0.0909
+ throughput/total_tokens=1,294,272,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6742/500000]
+ train/ActionL1Loss=0.0870
+ throughput/total_tokens=1,294,464,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6743/500000]
+ train/ActionL1Loss=0.0505
+ throughput/total_tokens=1,294,656,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6744/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,294,848,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6745/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,295,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6746/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=1,295,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6747/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,295,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6748/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,295,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6749/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,295,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6750/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,296,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6751/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,296,192,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6752/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,296,384,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6753/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,296,576,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6754/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,296,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6755/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,296,960,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6756/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,297,152,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6757/500000]
+ train/ActionL1Loss=0.0823
+ throughput/total_tokens=1,297,344,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6758/500000]
+ train/ActionL1Loss=0.0523
+ throughput/total_tokens=1,297,536,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6759/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,297,728,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6760/500000]
+ optim/total_grad_norm=7.223
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,297,920,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6761/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,298,112,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6762/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,298,304,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6763/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,298,496,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6764/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,298,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6765/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,298,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6766/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,299,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6767/500000]
+ train/ActionL1Loss=0.0843
+ throughput/total_tokens=1,299,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6768/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,299,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6769/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,299,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6770/500000]
+ train/ActionL1Loss=0.0893
+ throughput/total_tokens=1,299,840,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6771/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=1,300,032,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6772/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,300,224,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6773/500000]
+ train/ActionL1Loss=0.0712
+ throughput/total_tokens=1,300,416,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6774/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,300,608,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6775/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,300,800,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6776/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,300,992,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6777/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,301,184,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6778/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,301,376,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6779/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,301,568,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6780/500000]
+ optim/total_grad_norm=6.110
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,301,760,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6781/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,301,952,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6782/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,302,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6783/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,302,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6784/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,302,528,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6785/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,302,720,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6786/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,302,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6787/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,303,104,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6788/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,303,296,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6789/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,303,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6790/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=1,303,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6791/500000]
+ train/ActionL1Loss=0.0486
+ throughput/total_tokens=1,303,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6792/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,304,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6793/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,304,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6794/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,304,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6795/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,304,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6796/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,304,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6797/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,305,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6798/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,305,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6799/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,305,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6800/500000]
+ optim/total_grad_norm=6.770
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,305,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6801/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,305,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6802/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,305,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6803/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,306,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6804/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,306,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6805/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,306,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6806/500000]
+ train/ActionL1Loss=0.0762
+ throughput/total_tokens=1,306,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6807/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,306,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6808/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,307,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6809/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,307,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6810/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,307,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6811/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=1,307,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6812/500000]
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=1,307,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6813/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,308,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6814/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,308,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6815/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,308,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6816/500000]
+ train/ActionL1Loss=0.0467
+ throughput/total_tokens=1,308,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6817/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,308,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6818/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,309,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6819/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,309,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6820/500000]
+ optim/total_grad_norm=6.410
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,309,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6821/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,309,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6822/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,309,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6823/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,310,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6824/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,310,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6825/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,310,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6826/500000]
+ train/ActionL1Loss=0.0422
+ throughput/total_tokens=1,310,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6827/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,310,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6828/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,310,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6829/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,311,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6830/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,311,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6831/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,311,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6832/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,311,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6833/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=1,311,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6834/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,312,128,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6835/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,312,320,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6836/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=1,312,512,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6837/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,312,704,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6838/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=1,312,896,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6839/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,313,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6840/500000]
+ optim/total_grad_norm=6.454
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,313,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6841/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,313,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6842/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,313,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6843/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,313,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6844/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,314,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6845/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,314,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6846/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,314,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6847/500000]
+ train/ActionL1Loss=0.0439
+ throughput/total_tokens=1,314,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6848/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,314,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6849/500000]
+ train/ActionL1Loss=0.0523
+ throughput/total_tokens=1,315,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6850/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,315,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6851/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=1,315,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6852/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,315,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6853/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=1,315,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6854/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,315,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6855/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,316,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6856/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,316,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6857/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,316,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6858/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=1,316,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6859/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,316,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6860/500000]
+ optim/total_grad_norm=5.439
+ train/ActionL1Loss=0.0938
+ throughput/total_tokens=1,317,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6861/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,317,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6862/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,317,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6863/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,317,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6864/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,317,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6865/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,318,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6866/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,318,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6867/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,318,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6868/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,318,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6869/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,318,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=6870/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,319,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6871/500000]
+ train/ActionL1Loss=0.0825
+ throughput/total_tokens=1,319,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6872/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,319,424,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6873/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,319,616,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6874/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,319,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6875/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,320,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6876/500000]
+ train/ActionL1Loss=0.0506
+ throughput/total_tokens=1,320,192,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6877/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,320,384,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6878/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,320,576,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6879/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,320,768,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6880/500000]
+ optim/total_grad_norm=6.151
+ train/ActionL1Loss=0.0857
+ throughput/total_tokens=1,320,960,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6881/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,321,152,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6882/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,321,344,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6883/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,321,536,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6884/500000]
+ train/ActionL1Loss=0.0469
+ throughput/total_tokens=1,321,728,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6885/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,321,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6886/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=1,322,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6887/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,322,304,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6888/500000]
+ train/ActionL1Loss=0.0838
+ throughput/total_tokens=1,322,496,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6889/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,322,688,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6890/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=1,322,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6891/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,323,072,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6892/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,323,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6893/500000]
+ train/ActionL1Loss=0.0386
+ throughput/total_tokens=1,323,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6894/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,323,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6895/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,323,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6896/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,324,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6897/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,324,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6898/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,324,416,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6899/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,324,608,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=6900/500000]
+ optim/total_grad_norm=6.920
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,324,800,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6901/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,324,992,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6902/500000]
+ train/ActionL1Loss=0.0453
+ throughput/total_tokens=1,325,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6903/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,325,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6904/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,325,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6905/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,325,760,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=6906/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,325,952,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6907/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,326,144,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6908/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,326,336,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6909/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,326,528,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6910/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,326,720,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6911/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,326,912,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6912/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=1,327,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6913/500000]
+ train/ActionL1Loss=0.0831
+ throughput/total_tokens=1,327,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6914/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,327,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6915/500000]
+ train/ActionL1Loss=0.0523
+ throughput/total_tokens=1,327,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6916/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,327,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=6917/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,328,064,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6918/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,328,256,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6919/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,328,448,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6920/500000]
+ optim/total_grad_norm=6.979
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,328,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=6921/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,328,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6922/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,329,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6923/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,329,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6924/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,329,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=6925/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,329,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6926/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,329,792,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6927/500000]
+ train/ActionL1Loss=0.0840
+ throughput/total_tokens=1,329,984,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6928/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,330,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6929/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=1,330,368,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6930/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,330,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6931/500000]
+ train/ActionL1Loss=0.0850
+ throughput/total_tokens=1,330,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6932/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,330,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6933/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,331,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6934/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,331,328,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6935/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=1,331,520,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6936/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,331,712,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6937/500000]
+ train/ActionL1Loss=0.0782
+ throughput/total_tokens=1,331,904,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6938/500000]
+ train/ActionL1Loss=0.0827
+ throughput/total_tokens=1,332,096,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6939/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,332,288,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6940/500000]
+ optim/total_grad_norm=7.339
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,332,480,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6941/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,332,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6942/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,332,864,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6943/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,333,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6944/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,333,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6945/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,333,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6946/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,333,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6947/500000]
+ train/ActionL1Loss=0.0514
+ throughput/total_tokens=1,333,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6948/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,334,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6949/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=1,334,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=6950/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,334,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6951/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,334,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6952/500000]
+ train/ActionL1Loss=0.0419
+ throughput/total_tokens=1,334,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6953/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,334,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6954/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,335,168,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6955/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,335,360,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6956/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,335,552,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6957/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=1,335,744,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6958/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,335,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6959/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,336,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=6960/500000]
+ optim/total_grad_norm=4.608
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,336,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=6961/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,336,512,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6962/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,336,704,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6963/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,336,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6964/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,337,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6965/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,337,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6966/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,337,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6967/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,337,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6968/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,337,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6969/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,338,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=6970/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,338,240,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6971/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,338,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6972/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,338,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6973/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=1,338,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6974/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,339,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6975/500000]
+ train/ActionL1Loss=0.0957
+ throughput/total_tokens=1,339,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6976/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,339,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6977/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,339,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6978/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,339,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6979/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,339,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6980/500000]
+ optim/total_grad_norm=4.978
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=1,340,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6981/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,340,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6982/500000]
+ train/ActionL1Loss=0.0488
+ throughput/total_tokens=1,340,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6983/500000]
+ train/ActionL1Loss=0.0472
+ throughput/total_tokens=1,340,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6984/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=1,340,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6985/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,341,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6986/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,341,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6987/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=1,341,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6988/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,341,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6989/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,341,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6990/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,342,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=6991/500000]
+ train/ActionL1Loss=0.0489
+ throughput/total_tokens=1,342,272,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6992/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,342,464,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6993/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,342,656,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6994/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,342,848,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6995/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,343,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6996/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,343,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6997/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,343,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6998/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,343,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=6999/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,343,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7000/500000]
+ optim/total_grad_norm=4.266
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,344,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+10/02 [06:26:18] INFO | >> Saving config... checkpoint.py:608
+10/02 [06:26:49] INFO | >> Saving model state... checkpoint.py:796
+10/02 [06:28:02] INFO | >> Saving optim state... checkpoint.py:811
+10/02 [06:29:37] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=7001/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,344,192,000
+ throughput/device/tokens_per_second=1,211
+ throughput/device/batches_per_second=0.0505
+[step=7002/500000]
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,344,384,000
+ throughput/device/tokens_per_second=1,174
+ throughput/device/batches_per_second=0.0489
+[step=7003/500000]
+ train/ActionL1Loss=0.0747
+ throughput/total_tokens=1,344,576,000
+ throughput/device/tokens_per_second=1,146
+ throughput/device/batches_per_second=0.0478
+[step=7004/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,344,768,000
+ throughput/device/tokens_per_second=1,145
+ throughput/device/batches_per_second=0.0477
+[step=7005/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,344,960,000
+ throughput/device/tokens_per_second=1,149
+ throughput/device/batches_per_second=0.0479
+[step=7006/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,345,152,000
+ throughput/device/tokens_per_second=1,152
+ throughput/device/batches_per_second=0.0480
+[step=7007/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=1,345,344,000
+ throughput/device/tokens_per_second=1,154
+ throughput/device/batches_per_second=0.0481
+[step=7008/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,345,536,000
+ throughput/device/tokens_per_second=1,156
+ throughput/device/batches_per_second=0.0482
+[step=7009/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,345,728,000
+ throughput/device/tokens_per_second=1,157
+ throughput/device/batches_per_second=0.0482
+[step=7010/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,345,920,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+ System/Peak GPU Memory (MB)=46,917
+[step=7011/500000]
+ train/ActionL1Loss=0.0803
+ throughput/total_tokens=1,346,112,000
+ throughput/device/tokens_per_second=1,158
+ throughput/device/batches_per_second=0.0483
+[step=7012/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,346,304,000
+ throughput/device/tokens_per_second=1,159
+ throughput/device/batches_per_second=0.0483
+[step=7013/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,346,496,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0483
+[step=7014/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,346,688,000
+ throughput/device/tokens_per_second=1,160
+ throughput/device/batches_per_second=0.0484
+[step=7015/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=1,346,880,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=7016/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,347,072,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=7017/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,347,264,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=7018/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,347,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7019/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,347,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7020/500000]
+ optim/total_grad_norm=6.866
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,347,840,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=7021/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,348,032,000
+ throughput/device/tokens_per_second=1,161
+ throughput/device/batches_per_second=0.0484
+[step=7022/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,348,224,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7023/500000]
+ train/ActionL1Loss=0.0801
+ throughput/total_tokens=1,348,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7024/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=1,348,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7025/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,348,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7026/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,348,992,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7027/500000]
+ train/ActionL1Loss=0.0432
+ throughput/total_tokens=1,349,184,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7028/500000]
+ train/ActionL1Loss=0.0976
+ throughput/total_tokens=1,349,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7029/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,349,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7030/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,349,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7031/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,349,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7032/500000]
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,350,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7033/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,350,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7034/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,350,528,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7035/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,350,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7036/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,350,912,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7037/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,351,104,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7038/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,351,296,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7039/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,351,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7040/500000]
+ optim/total_grad_norm=5.971
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,351,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7041/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=1,351,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7042/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,352,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7043/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,352,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7044/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,352,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7045/500000]
+ train/ActionL1Loss=0.0435
+ throughput/total_tokens=1,352,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7046/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,352,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7047/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,353,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7048/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,353,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7049/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,353,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7050/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,353,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7051/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,353,792,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7052/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,353,984,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7053/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,354,176,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7054/500000]
+ train/ActionL1Loss=0.0422
+ throughput/total_tokens=1,354,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7055/500000]
+ train/ActionL1Loss=0.0486
+ throughput/total_tokens=1,354,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7056/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,354,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7057/500000]
+ train/ActionL1Loss=0.0473
+ throughput/total_tokens=1,354,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7058/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,355,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7059/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,355,328,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7060/500000]
+ optim/total_grad_norm=5.626
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,355,520,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7061/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,355,712,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7062/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,355,904,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7063/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,356,096,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7064/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,356,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7065/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,356,480,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7066/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,356,672,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7067/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,356,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7068/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,357,056,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7069/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,357,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7070/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,357,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7071/500000]
+ train/ActionL1Loss=0.0437
+ throughput/total_tokens=1,357,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7072/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,357,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7073/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,358,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7074/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,358,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7075/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,358,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7076/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,358,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7077/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,358,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7078/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,358,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7079/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,359,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7080/500000]
+ optim/total_grad_norm=7.207
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,359,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7081/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,359,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7082/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,359,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7083/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,359,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7084/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,360,128,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7085/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,360,320,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7086/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,360,512,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7087/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,360,704,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7088/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,360,896,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7089/500000]
+ train/ActionL1Loss=0.0480
+ throughput/total_tokens=1,361,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7090/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=1,361,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7091/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,361,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7092/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,361,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7093/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,361,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7094/500000]
+ train/ActionL1Loss=0.0527
+ throughput/total_tokens=1,362,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7095/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,362,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7096/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,362,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7097/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=1,362,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7098/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,362,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7099/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,363,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7100/500000]
+ optim/total_grad_norm=5.959
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,363,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7101/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,363,392,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7102/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=1,363,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7103/500000]
+ train/ActionL1Loss=0.0527
+ throughput/total_tokens=1,363,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7104/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,363,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7105/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,364,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7106/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,364,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7107/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,364,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7108/500000]
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,364,736,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7109/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,364,928,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7110/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,365,120,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+ System/Peak GPU Memory (MB)=46,917
+[step=7111/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,365,312,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7112/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=1,365,504,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7113/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,365,696,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7114/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,365,888,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7115/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,366,080,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7116/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,366,272,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7117/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,366,464,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7118/500000]
+ train/ActionL1Loss=0.0758
+ throughput/total_tokens=1,366,656,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7119/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,366,848,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7120/500000]
+ optim/total_grad_norm=5.819
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,367,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7121/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,367,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7122/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,367,424,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7123/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,367,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7124/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,367,808,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7125/500000]
+ train/ActionL1Loss=0.0457
+ throughput/total_tokens=1,368,000,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7126/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,368,192,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7127/500000]
+ train/ActionL1Loss=0.0739
+ throughput/total_tokens=1,368,384,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7128/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=1,368,576,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7129/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,368,768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7130/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,368,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7131/500000]
+ train/ActionL1Loss=0.0533
+ throughput/total_tokens=1,369,152,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7132/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,369,344,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7133/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,369,536,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7134/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,369,728,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7135/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,369,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7136/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,370,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7137/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,370,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7138/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,370,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7139/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=1,370,688,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7140/500000]
+ optim/total_grad_norm=5.508
+ train/ActionL1Loss=0.0797
+ throughput/total_tokens=1,370,880,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7141/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=1,371,072,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7142/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,371,264,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7143/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,371,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7144/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,371,648,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7145/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,371,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7146/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,372,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7147/500000]
+ train/ActionL1Loss=0.0743
+ throughput/total_tokens=1,372,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7148/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,372,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7149/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,372,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7150/500000]
+ train/ActionL1Loss=0.0438
+ throughput/total_tokens=1,372,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7151/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,372,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7152/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,373,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7153/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,373,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7154/500000]
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,373,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7155/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,373,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7156/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,373,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7157/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,374,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7158/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,374,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7159/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,374,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7160/500000]
+ optim/total_grad_norm=6.516
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,374,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7161/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,374,912,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7162/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,375,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7163/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,375,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7164/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,375,488,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7165/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,375,680,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7166/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,375,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7167/500000]
+ train/ActionL1Loss=0.0452
+ throughput/total_tokens=1,376,064,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7168/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,376,256,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7169/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,376,448,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7170/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,376,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7171/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,376,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7172/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,377,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7173/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,377,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7174/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,377,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7175/500000]
+ train/ActionL1Loss=0.0500
+ throughput/total_tokens=1,377,600,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7176/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,377,792,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7177/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,377,984,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7178/500000]
+ train/ActionL1Loss=0.0451
+ throughput/total_tokens=1,378,176,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7179/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,378,368,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7180/500000]
+ optim/total_grad_norm=7.958
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,378,560,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7181/500000]
+ train/ActionL1Loss=0.0494
+ throughput/total_tokens=1,378,752,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7182/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,378,944,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7183/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,379,136,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7184/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,379,328,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7185/500000]
+ train/ActionL1Loss=0.0455
+ throughput/total_tokens=1,379,520,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7186/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,379,712,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7187/500000]
+ train/ActionL1Loss=0.0533
+ throughput/total_tokens=1,379,904,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7188/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,380,096,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7189/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,380,288,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7190/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,380,480,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7191/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,380,672,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7192/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,380,864,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7193/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,381,056,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7194/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,381,248,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7195/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,381,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7196/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,381,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7197/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,381,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7198/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,382,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7199/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,382,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7200/500000]
+ optim/total_grad_norm=5.271
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,382,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7201/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,382,592,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7202/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,382,784,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7203/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,382,976,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7204/500000]
+ train/ActionL1Loss=0.0832
+ throughput/total_tokens=1,383,168,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7205/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,383,360,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7206/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,383,552,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7207/500000]
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,383,744,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7208/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,383,936,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7209/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,384,128,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7210/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,384,320,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7211/500000]
+ train/ActionL1Loss=0.0453
+ throughput/total_tokens=1,384,512,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7212/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,384,704,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7213/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,384,896,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7214/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,385,088,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7215/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,385,280,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7216/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,385,472,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7217/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,385,664,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7218/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,385,856,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7219/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,386,048,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7220/500000]
+ optim/total_grad_norm=6.081
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,386,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7221/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,386,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7222/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,386,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7223/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,386,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7224/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,387,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7225/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,387,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7226/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,387,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7227/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,387,584,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7228/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,387,776,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7229/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,387,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7230/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,388,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7231/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,388,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7232/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,388,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7233/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,388,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7234/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,388,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7235/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,389,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7236/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,389,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7237/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,389,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7238/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,389,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7239/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,389,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7240/500000]
+ optim/total_grad_norm=5.700
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,390,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7241/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,390,272,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7242/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,390,464,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7243/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=1,390,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7244/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,390,848,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7245/500000]
+ train/ActionL1Loss=0.0776
+ throughput/total_tokens=1,391,040,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7246/500000]
+ train/ActionL1Loss=0.0445
+ throughput/total_tokens=1,391,232,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7247/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,391,424,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7248/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,391,616,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7249/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,391,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7250/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,392,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7251/500000]
+ train/ActionL1Loss=0.0472
+ throughput/total_tokens=1,392,192,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7252/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,392,384,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7253/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,392,576,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7254/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,392,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7255/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,392,960,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7256/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,393,152,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7257/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,393,344,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7258/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,393,536,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7259/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,393,728,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7260/500000]
+ optim/total_grad_norm=6.312
+ train/ActionL1Loss=0.0416
+ throughput/total_tokens=1,393,920,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7261/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,394,112,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7262/500000]
+ train/ActionL1Loss=0.0401
+ throughput/total_tokens=1,394,304,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7263/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,394,496,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7264/500000]
+ train/ActionL1Loss=0.0466
+ throughput/total_tokens=1,394,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7265/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,394,880,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7266/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,395,072,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7267/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,395,264,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7268/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,395,456,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7269/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,395,648,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7270/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,395,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7271/500000]
+ train/ActionL1Loss=0.0567
+ throughput/total_tokens=1,396,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7272/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=1,396,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7273/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,396,416,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7274/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,396,608,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7275/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,396,800,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7276/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,396,992,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7277/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,397,184,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7278/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=1,397,376,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7279/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,397,568,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7280/500000]
+ optim/total_grad_norm=7.779
+ train/ActionL1Loss=0.0470
+ throughput/total_tokens=1,397,760,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7281/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,397,952,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7282/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,398,144,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7283/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,398,336,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7284/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,398,528,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7285/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,398,720,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7286/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=1,398,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7287/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,399,104,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7288/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,399,296,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7289/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,399,488,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7290/500000]
+ train/ActionL1Loss=0.0452
+ throughput/total_tokens=1,399,680,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7291/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,399,872,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7292/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,400,064,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7293/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=1,400,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7294/500000]
+ train/ActionL1Loss=0.0477
+ throughput/total_tokens=1,400,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7295/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,400,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7296/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,400,832,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7297/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,401,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7298/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,401,216,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7299/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,401,408,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7300/500000]
+ optim/total_grad_norm=6.383
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,401,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7301/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,401,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7302/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,401,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7303/500000]
+ train/ActionL1Loss=0.0741
+ throughput/total_tokens=1,402,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7304/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,402,368,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7305/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,402,560,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7306/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,402,752,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7307/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,402,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7308/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,403,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7309/500000]
+ train/ActionL1Loss=0.0481
+ throughput/total_tokens=1,403,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7310/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,403,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7311/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,403,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7312/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,403,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7313/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,404,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7314/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,404,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7315/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,404,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7316/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,404,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7317/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,404,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7318/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,405,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7319/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,405,248,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7320/500000]
+ optim/total_grad_norm=7.936
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=1,405,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7321/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,405,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7322/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,405,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7323/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,406,016,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7324/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,406,208,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7325/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,406,400,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7326/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,406,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7327/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,406,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7328/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,406,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7329/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,407,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7330/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,407,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7331/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,407,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7332/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,407,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7333/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,407,936,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7334/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,408,128,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7335/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,408,320,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7336/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,408,512,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7337/500000]
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=1,408,704,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7338/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,408,896,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7339/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,409,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7340/500000]
+ optim/total_grad_norm=5.910
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,409,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7341/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,409,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7342/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,409,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7343/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,409,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7344/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,410,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7345/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,410,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7346/500000]
+ train/ActionL1Loss=0.0439
+ throughput/total_tokens=1,410,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7347/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,410,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7348/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,410,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7349/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,411,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7350/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,411,200,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7351/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=1,411,392,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7352/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,411,584,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7353/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,411,776,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7354/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,411,968,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7355/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=1,412,160,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7356/500000]
+ train/ActionL1Loss=0.0783
+ throughput/total_tokens=1,412,352,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7357/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,412,544,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7358/500000]
+ train/ActionL1Loss=0.0432
+ throughput/total_tokens=1,412,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7359/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,412,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7360/500000]
+ optim/total_grad_norm=4.925
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,413,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7361/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,413,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7362/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,413,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7363/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,413,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7364/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,413,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7365/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,414,080,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7366/500000]
+ train/ActionL1Loss=0.0780
+ throughput/total_tokens=1,414,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7367/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,414,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7368/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,414,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7369/500000]
+ train/ActionL1Loss=0.0755
+ throughput/total_tokens=1,414,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7370/500000]
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,415,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7371/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,415,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7372/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,415,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7373/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,415,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7374/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,415,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7375/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,416,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7376/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,416,192,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7377/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,416,384,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7378/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,416,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7379/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,416,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7380/500000]
+ optim/total_grad_norm=5.884
+ train/ActionL1Loss=0.0973
+ throughput/total_tokens=1,416,960,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7381/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,417,152,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7382/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,417,344,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7383/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,417,536,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7384/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,417,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7385/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,417,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7386/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,418,112,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7387/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,418,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7388/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,418,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7389/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,418,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7390/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,418,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7391/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,419,072,000
+ throughput/device/tokens_per_second=1,126
+ throughput/device/batches_per_second=0.0470
+[step=7392/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,419,264,000
+ throughput/device/tokens_per_second=1,126
+ throughput/device/batches_per_second=0.0470
+[step=7393/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,419,456,000
+ throughput/device/tokens_per_second=1,126
+ throughput/device/batches_per_second=0.0469
+[step=7394/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,419,648,000
+ throughput/device/tokens_per_second=1,126
+ throughput/device/batches_per_second=0.0469
+[step=7395/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,419,840,000
+ throughput/device/tokens_per_second=1,125
+ throughput/device/batches_per_second=0.0469
+[step=7396/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,420,032,000
+ throughput/device/tokens_per_second=1,124
+ throughput/device/batches_per_second=0.0469
+[step=7397/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,420,224,000
+ throughput/device/tokens_per_second=1,124
+ throughput/device/batches_per_second=0.0469
+[step=7398/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,420,416,000
+ throughput/device/tokens_per_second=1,124
+ throughput/device/batches_per_second=0.0469
+[step=7399/500000]
+ train/ActionL1Loss=0.0699
+ throughput/total_tokens=1,420,608,000
+ throughput/device/tokens_per_second=1,125
+ throughput/device/batches_per_second=0.0469
+[step=7400/500000]
+ optim/total_grad_norm=5.472
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=1,420,800,000
+ throughput/device/tokens_per_second=1,125
+ throughput/device/batches_per_second=0.0469
+ System/Peak GPU Memory (MB)=46,917
+[step=7401/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,420,992,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7402/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,421,184,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7403/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,421,376,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7404/500000]
+ train/ActionL1Loss=0.0408
+ throughput/total_tokens=1,421,568,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7405/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,421,760,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7406/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,421,952,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7407/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,422,144,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7408/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,422,336,000
+ throughput/device/tokens_per_second=1,123
+ throughput/device/batches_per_second=0.0468
+[step=7409/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,422,528,000
+ throughput/device/tokens_per_second=1,122
+ throughput/device/batches_per_second=0.0468
+[step=7410/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,422,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7411/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,422,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7412/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,423,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7413/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,423,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7414/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,423,488,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7415/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,423,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7416/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,423,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7417/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,424,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7418/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,424,256,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7419/500000]
+ train/ActionL1Loss=0.0738
+ throughput/total_tokens=1,424,448,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7420/500000]
+ optim/total_grad_norm=6.026
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,424,640,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7421/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,424,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7422/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,425,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7423/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,425,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7424/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,425,408,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7425/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=1,425,600,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7426/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,425,792,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7427/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,425,984,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7428/500000]
+ train/ActionL1Loss=0.0774
+ throughput/total_tokens=1,426,176,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7429/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,426,368,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7430/500000]
+ train/ActionL1Loss=0.0506
+ throughput/total_tokens=1,426,560,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7431/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,426,752,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7432/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,426,944,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7433/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,427,136,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7434/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,427,328,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7435/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,427,520,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7436/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,427,712,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7437/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,427,904,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7438/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,428,096,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7439/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,428,288,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7440/500000]
+ optim/total_grad_norm=4.235
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=1,428,480,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7441/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,428,672,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7442/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,428,864,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7443/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,429,056,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7444/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,429,248,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7445/500000]
+ train/ActionL1Loss=0.0862
+ throughput/total_tokens=1,429,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7446/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,429,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7447/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,429,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7448/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,430,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7449/500000]
+ train/ActionL1Loss=0.0826
+ throughput/total_tokens=1,430,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7450/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,430,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7451/500000]
+ train/ActionL1Loss=0.0513
+ throughput/total_tokens=1,430,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7452/500000]
+ train/ActionL1Loss=0.0477
+ throughput/total_tokens=1,430,784,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7453/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,430,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7454/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,431,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7455/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,431,360,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7456/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,431,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7457/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,431,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7458/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,431,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7459/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,432,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7460/500000]
+ optim/total_grad_norm=6.887
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,432,320,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7461/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,432,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7462/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,432,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7463/500000]
+ train/ActionL1Loss=0.0821
+ throughput/total_tokens=1,432,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7464/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,433,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7465/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,433,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7466/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,433,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7467/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,433,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7468/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,433,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7469/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,434,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7470/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=1,434,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7471/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=1,434,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7472/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,434,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7473/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,434,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7474/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,435,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7475/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,435,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7476/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,435,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7477/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,435,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7478/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,435,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7479/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,435,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7480/500000]
+ optim/total_grad_norm=5.998
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,436,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7481/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,436,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7482/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,436,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7483/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,436,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7484/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,436,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7485/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,437,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7486/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,437,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7487/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,437,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7488/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,437,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7489/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,437,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7490/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,438,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7491/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,438,272,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7492/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,438,464,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7493/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,438,656,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7494/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,438,848,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7495/500000]
+ train/ActionL1Loss=0.0459
+ throughput/total_tokens=1,439,040,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7496/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,439,232,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7497/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,439,424,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7498/500000]
+ train/ActionL1Loss=0.0471
+ throughput/total_tokens=1,439,616,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7499/500000]
+ train/ActionL1Loss=0.0868
+ throughput/total_tokens=1,439,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7500/500000]
+ optim/total_grad_norm=7.588
+ train/ActionL1Loss=0.0482
+ throughput/total_tokens=1,440,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+10/02 [09:22:51] INFO | >> Saving config... checkpoint.py:608
+10/02 [09:23:31] INFO | >> Saving model state... checkpoint.py:796
+10/02 [09:24:39] INFO | >> Saving optim state... checkpoint.py:811
+10/02 [09:26:13] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=7501/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,440,192,000
+ throughput/device/tokens_per_second=1,213
+ throughput/device/batches_per_second=0.0506
+[step=7502/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,440,384,000
+ throughput/device/tokens_per_second=1,189
+ throughput/device/batches_per_second=0.0496
+[step=7503/500000]
+ train/ActionL1Loss=0.0445
+ throughput/total_tokens=1,440,576,000
+ throughput/device/tokens_per_second=1,183
+ throughput/device/batches_per_second=0.0493
+[step=7504/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,440,768,000
+ throughput/device/tokens_per_second=1,179
+ throughput/device/batches_per_second=0.0492
+[step=7505/500000]
+ train/ActionL1Loss=0.0413
+ throughput/total_tokens=1,440,960,000
+ throughput/device/tokens_per_second=1,177
+ throughput/device/batches_per_second=0.0491
+[step=7506/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,441,152,000
+ throughput/device/tokens_per_second=1,176
+ throughput/device/batches_per_second=0.0490
+[step=7507/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,441,344,000
+ throughput/device/tokens_per_second=1,175
+ throughput/device/batches_per_second=0.0490
+[step=7508/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,441,536,000
+ throughput/device/tokens_per_second=1,174
+ throughput/device/batches_per_second=0.0490
+[step=7509/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,441,728,000
+ throughput/device/tokens_per_second=1,173
+ throughput/device/batches_per_second=0.0489
+[step=7510/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,441,920,000
+ throughput/device/tokens_per_second=1,173
+ throughput/device/batches_per_second=0.0489
+ System/Peak GPU Memory (MB)=46,917
+[step=7511/500000]
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,442,112,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=7512/500000]
+ train/ActionL1Loss=0.0468
+ throughput/total_tokens=1,442,304,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0489
+[step=7513/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,442,496,000
+ throughput/device/tokens_per_second=1,172
+ throughput/device/batches_per_second=0.0488
+[step=7514/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,442,688,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=7515/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,442,880,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=7516/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,443,072,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=7517/500000]
+ train/ActionL1Loss=0.0410
+ throughput/total_tokens=1,443,264,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=7518/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,443,456,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=7519/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,443,648,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=7520/500000]
+ optim/total_grad_norm=6.610
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=1,443,840,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+ System/Peak GPU Memory (MB)=46,917
+[step=7521/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,444,032,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=7522/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,444,224,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=7523/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,444,416,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7524/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,444,608,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7525/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=1,444,800,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7526/500000]
+ train/ActionL1Loss=0.0494
+ throughput/total_tokens=1,444,992,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7527/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,445,184,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7528/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,445,376,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7529/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,445,568,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7530/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,445,760,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7531/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,445,952,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7532/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,446,144,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7533/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,446,336,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7534/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,446,528,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7535/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,446,720,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7536/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,446,912,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7537/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,447,104,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7538/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,447,296,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7539/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,447,488,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7540/500000]
+ optim/total_grad_norm=8.261
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,447,680,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7541/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,447,872,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7542/500000]
+ train/ActionL1Loss=0.0725
+ throughput/total_tokens=1,448,064,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7543/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,448,256,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7544/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,448,448,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7545/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,448,640,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7546/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,448,832,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7547/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,449,024,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7548/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,449,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7549/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,449,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7550/500000]
+ train/ActionL1Loss=0.0430
+ throughput/total_tokens=1,449,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7551/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,449,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7552/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,449,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7553/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,450,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7554/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,450,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7555/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,450,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7556/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,450,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7557/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,450,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7558/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,451,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7559/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,451,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7560/500000]
+ optim/total_grad_norm=6.637
+ train/ActionL1Loss=0.0563
+ throughput/total_tokens=1,451,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7561/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,451,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7562/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,451,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7563/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,452,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7564/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,452,288,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7565/500000]
+ train/ActionL1Loss=0.1020
+ throughput/total_tokens=1,452,480,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7566/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,452,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7567/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,452,864,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7568/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,453,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7569/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,453,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7570/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,453,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7571/500000]
+ train/ActionL1Loss=0.0817
+ throughput/total_tokens=1,453,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7572/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,453,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7573/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,454,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7574/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=1,454,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7575/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,454,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7576/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,454,592,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7577/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,454,784,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7578/500000]
+ train/ActionL1Loss=0.0730
+ throughput/total_tokens=1,454,976,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7579/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,455,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7580/500000]
+ optim/total_grad_norm=6.462
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,455,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7581/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,455,552,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7582/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,455,744,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7583/500000]
+ train/ActionL1Loss=0.0853
+ throughput/total_tokens=1,455,936,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7584/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,456,128,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7585/500000]
+ train/ActionL1Loss=0.0508
+ throughput/total_tokens=1,456,320,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7586/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,456,512,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7587/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,456,704,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7588/500000]
+ train/ActionL1Loss=0.0876
+ throughput/total_tokens=1,456,896,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7589/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,457,088,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7590/500000]
+ train/ActionL1Loss=0.0744
+ throughput/total_tokens=1,457,280,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7591/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,457,472,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7592/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,457,664,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7593/500000]
+ train/ActionL1Loss=0.0403
+ throughput/total_tokens=1,457,856,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7594/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,458,048,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7595/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,458,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7596/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,458,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7597/500000]
+ train/ActionL1Loss=0.0756
+ throughput/total_tokens=1,458,624,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7598/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,458,816,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7599/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,459,008,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7600/500000]
+ optim/total_grad_norm=7.400
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,459,200,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7601/500000]
+ train/ActionL1Loss=0.0710
+ throughput/total_tokens=1,459,392,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7602/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,459,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7603/500000]
+ train/ActionL1Loss=0.0435
+ throughput/total_tokens=1,459,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7604/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,459,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7605/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=1,460,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7606/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,460,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7607/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,460,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7608/500000]
+ train/ActionL1Loss=0.0800
+ throughput/total_tokens=1,460,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7609/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,460,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7610/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,461,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7611/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,461,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7612/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,461,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7613/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=1,461,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7614/500000]
+ train/ActionL1Loss=0.0886
+ throughput/total_tokens=1,461,888,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7615/500000]
+ train/ActionL1Loss=0.0716
+ throughput/total_tokens=1,462,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7616/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,462,272,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7617/500000]
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=1,462,464,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7618/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,462,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7619/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,462,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7620/500000]
+ optim/total_grad_norm=6.446
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,463,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7621/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,463,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7622/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=1,463,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7623/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,463,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7624/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,463,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7625/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,464,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7626/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,464,192,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7627/500000]
+ train/ActionL1Loss=0.0469
+ throughput/total_tokens=1,464,384,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7628/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,464,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7629/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,464,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7630/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,464,960,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7631/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,465,152,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7632/500000]
+ train/ActionL1Loss=0.0911
+ throughput/total_tokens=1,465,344,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7633/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,465,536,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7634/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,465,728,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7635/500000]
+ train/ActionL1Loss=0.0489
+ throughput/total_tokens=1,465,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7636/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,466,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7637/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,466,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7638/500000]
+ train/ActionL1Loss=0.0482
+ throughput/total_tokens=1,466,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7639/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,466,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7640/500000]
+ optim/total_grad_norm=6.145
+ train/ActionL1Loss=0.0471
+ throughput/total_tokens=1,466,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7641/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,467,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7642/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,467,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7643/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,467,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7644/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,467,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7645/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,467,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7646/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,468,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7647/500000]
+ train/ActionL1Loss=0.0595
+ throughput/total_tokens=1,468,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7648/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,468,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7649/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,468,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7650/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,468,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7651/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,468,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7652/500000]
+ train/ActionL1Loss=0.0788
+ throughput/total_tokens=1,469,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7653/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,469,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7654/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,469,568,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=7655/500000]
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,469,760,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7656/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,469,952,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=7657/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,470,144,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7658/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,470,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7659/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,470,528,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7660/500000]
+ optim/total_grad_norm=6.801
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,470,720,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7661/500000]
+ train/ActionL1Loss=0.0421
+ throughput/total_tokens=1,470,912,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7662/500000]
+ train/ActionL1Loss=0.0465
+ throughput/total_tokens=1,471,104,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7663/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,471,296,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7664/500000]
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,471,488,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7665/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,471,680,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7666/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,471,872,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7667/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,472,064,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7668/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,472,256,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7669/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,472,448,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7670/500000]
+ train/ActionL1Loss=0.0393
+ throughput/total_tokens=1,472,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7671/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,472,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7672/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,473,024,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7673/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,473,216,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7674/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,473,408,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7675/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,473,600,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7676/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,473,792,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7677/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,473,984,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7678/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,474,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7679/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,474,368,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7680/500000]
+ optim/total_grad_norm=8.626
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,474,560,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7681/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=1,474,752,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7682/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,474,944,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7683/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,475,136,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7684/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,475,328,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7685/500000]
+ train/ActionL1Loss=0.0753
+ throughput/total_tokens=1,475,520,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7686/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,475,712,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7687/500000]
+ train/ActionL1Loss=0.0472
+ throughput/total_tokens=1,475,904,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7688/500000]
+ train/ActionL1Loss=0.0409
+ throughput/total_tokens=1,476,096,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7689/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,476,288,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7690/500000]
+ train/ActionL1Loss=0.0518
+ throughput/total_tokens=1,476,480,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7691/500000]
+ train/ActionL1Loss=0.0513
+ throughput/total_tokens=1,476,672,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7692/500000]
+ train/ActionL1Loss=0.0677
+ throughput/total_tokens=1,476,864,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7693/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,477,056,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7694/500000]
+ train/ActionL1Loss=0.0474
+ throughput/total_tokens=1,477,248,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7695/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,477,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7696/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,477,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7697/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,477,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7698/500000]
+ train/ActionL1Loss=0.0498
+ throughput/total_tokens=1,478,016,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7699/500000]
+ train/ActionL1Loss=0.0457
+ throughput/total_tokens=1,478,208,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7700/500000]
+ optim/total_grad_norm=5.611
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,478,400,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7701/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,478,592,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7702/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,478,784,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7703/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,478,976,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7704/500000]
+ train/ActionL1Loss=0.0508
+ throughput/total_tokens=1,479,168,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7705/500000]
+ train/ActionL1Loss=0.0594
+ throughput/total_tokens=1,479,360,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7706/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,479,552,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7707/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,479,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7708/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,479,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7709/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,480,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7710/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,480,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7711/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,480,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7712/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,480,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7713/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,480,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7714/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,481,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7715/500000]
+ train/ActionL1Loss=0.0806
+ throughput/total_tokens=1,481,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7716/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,481,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7717/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,481,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7718/500000]
+ train/ActionL1Loss=0.0859
+ throughput/total_tokens=1,481,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7719/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,482,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7720/500000]
+ optim/total_grad_norm=4.346
+ train/ActionL1Loss=0.0449
+ throughput/total_tokens=1,482,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7721/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,482,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7722/500000]
+ train/ActionL1Loss=0.0875
+ throughput/total_tokens=1,482,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7723/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,482,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7724/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,483,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7725/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,483,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7726/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,483,392,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7727/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,483,584,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7728/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=1,483,776,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7729/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=1,483,968,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7730/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,484,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7731/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,484,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7732/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,484,544,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7733/500000]
+ train/ActionL1Loss=0.0419
+ throughput/total_tokens=1,484,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7734/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,484,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7735/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,485,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7736/500000]
+ train/ActionL1Loss=0.0881
+ throughput/total_tokens=1,485,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7737/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,485,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7738/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,485,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7739/500000]
+ train/ActionL1Loss=0.0471
+ throughput/total_tokens=1,485,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7740/500000]
+ optim/total_grad_norm=5.096
+ train/ActionL1Loss=0.0940
+ throughput/total_tokens=1,486,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7741/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,486,272,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7742/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,486,464,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7743/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,486,656,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7744/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,486,848,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7745/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,487,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7746/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,487,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7747/500000]
+ train/ActionL1Loss=0.0481
+ throughput/total_tokens=1,487,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7748/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,487,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7749/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,487,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7750/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,488,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7751/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,488,192,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7752/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=1,488,384,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7753/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,488,576,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7754/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,488,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7755/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,488,960,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7756/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,489,152,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7757/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,489,344,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7758/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,489,536,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7759/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,489,728,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7760/500000]
+ optim/total_grad_norm=7.822
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,489,920,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7761/500000]
+ train/ActionL1Loss=0.0830
+ throughput/total_tokens=1,490,112,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7762/500000]
+ train/ActionL1Loss=0.0686
+ throughput/total_tokens=1,490,304,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7763/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,490,496,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7764/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,490,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7765/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,490,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7766/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,491,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7767/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,491,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7768/500000]
+ train/ActionL1Loss=0.0468
+ throughput/total_tokens=1,491,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7769/500000]
+ train/ActionL1Loss=0.0456
+ throughput/total_tokens=1,491,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7770/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,491,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7771/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,492,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7772/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=1,492,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7773/500000]
+ train/ActionL1Loss=0.0419
+ throughput/total_tokens=1,492,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7774/500000]
+ train/ActionL1Loss=0.0560
+ throughput/total_tokens=1,492,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7775/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,492,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7776/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,492,992,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7777/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,493,184,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7778/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,493,376,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7779/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,493,568,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7780/500000]
+ optim/total_grad_norm=6.279
+ train/ActionL1Loss=0.0447
+ throughput/total_tokens=1,493,760,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7781/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=1,493,952,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7782/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,494,144,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7783/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,494,336,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7784/500000]
+ train/ActionL1Loss=0.0448
+ throughput/total_tokens=1,494,528,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7785/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,494,720,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7786/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,494,912,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7787/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,495,104,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7788/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,495,296,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7789/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,495,488,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7790/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,495,680,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7791/500000]
+ train/ActionL1Loss=0.0498
+ throughput/total_tokens=1,495,872,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7792/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,496,064,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7793/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,496,256,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7794/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,496,448,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7795/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,496,640,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7796/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,496,832,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7797/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,497,024,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7798/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,497,216,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7799/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,497,408,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7800/500000]
+ optim/total_grad_norm=5.378
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,497,600,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7801/500000]
+ train/ActionL1Loss=0.0505
+ throughput/total_tokens=1,497,792,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7802/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,497,984,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7803/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,498,176,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7804/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,498,368,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7805/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,498,560,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7806/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,498,752,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7807/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,498,944,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7808/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=1,499,136,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7809/500000]
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,499,328,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7810/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,499,520,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7811/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,499,712,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7812/500000]
+ train/ActionL1Loss=0.0871
+ throughput/total_tokens=1,499,904,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7813/500000]
+ train/ActionL1Loss=0.0440
+ throughput/total_tokens=1,500,096,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7814/500000]
+ train/ActionL1Loss=0.0489
+ throughput/total_tokens=1,500,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7815/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,500,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7816/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,500,672,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7817/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=1,500,864,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7818/500000]
+ train/ActionL1Loss=0.0888
+ throughput/total_tokens=1,501,056,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7819/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,501,248,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7820/500000]
+ optim/total_grad_norm=6.834
+ train/ActionL1Loss=0.0408
+ throughput/total_tokens=1,501,440,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7821/500000]
+ train/ActionL1Loss=0.0805
+ throughput/total_tokens=1,501,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7822/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,501,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7823/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,502,016,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7824/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,502,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7825/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,502,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7826/500000]
+ train/ActionL1Loss=0.0526
+ throughput/total_tokens=1,502,592,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7827/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,502,784,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7828/500000]
+ train/ActionL1Loss=0.0680
+ throughput/total_tokens=1,502,976,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7829/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,503,168,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7830/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,503,360,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7831/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,503,552,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7832/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,503,744,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7833/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,503,936,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7834/500000]
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=1,504,128,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7835/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,504,320,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7836/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,504,512,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7837/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,504,704,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7838/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,504,896,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7839/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,505,088,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7840/500000]
+ optim/total_grad_norm=5.781
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,505,280,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7841/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,505,472,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7842/500000]
+ train/ActionL1Loss=0.0784
+ throughput/total_tokens=1,505,664,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7843/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,505,856,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7844/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,506,048,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7845/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,506,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7846/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,506,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7847/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,506,624,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7848/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,506,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7849/500000]
+ train/ActionL1Loss=0.0964
+ throughput/total_tokens=1,507,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7850/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,507,200,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7851/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,507,392,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7852/500000]
+ train/ActionL1Loss=0.0504
+ throughput/total_tokens=1,507,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7853/500000]
+ train/ActionL1Loss=0.0761
+ throughput/total_tokens=1,507,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7854/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,507,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7855/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,508,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7856/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,508,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7857/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,508,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7858/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,508,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7859/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,508,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7860/500000]
+ optim/total_grad_norm=6.353
+ train/ActionL1Loss=0.0461
+ throughput/total_tokens=1,509,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7861/500000]
+ train/ActionL1Loss=0.0655
+ throughput/total_tokens=1,509,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7862/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,509,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7863/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,509,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7864/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,509,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7865/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,510,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7866/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,510,272,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7867/500000]
+ train/ActionL1Loss=0.0791
+ throughput/total_tokens=1,510,464,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7868/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,510,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7869/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,510,848,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7870/500000]
+ train/ActionL1Loss=0.0513
+ throughput/total_tokens=1,511,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7871/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=1,511,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7872/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,511,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7873/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,511,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7874/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,511,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7875/500000]
+ train/ActionL1Loss=0.0816
+ throughput/total_tokens=1,512,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7876/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,512,192,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7877/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,512,384,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7878/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,512,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7879/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,512,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7880/500000]
+ optim/total_grad_norm=6.621
+ train/ActionL1Loss=0.0720
+ throughput/total_tokens=1,512,960,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7881/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,513,152,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7882/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,513,344,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7883/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,513,536,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7884/500000]
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,513,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7885/500000]
+ train/ActionL1Loss=0.0717
+ throughput/total_tokens=1,513,920,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7886/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,514,112,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7887/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,514,304,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7888/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,514,496,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7889/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,514,688,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7890/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,514,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7891/500000]
+ train/ActionL1Loss=0.0513
+ throughput/total_tokens=1,515,072,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7892/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,515,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7893/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,515,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7894/500000]
+ train/ActionL1Loss=0.0494
+ throughput/total_tokens=1,515,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7895/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,515,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7896/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,516,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=7897/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,516,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=7898/500000]
+ train/ActionL1Loss=0.0701
+ throughput/total_tokens=1,516,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7899/500000]
+ train/ActionL1Loss=0.0795
+ throughput/total_tokens=1,516,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7900/500000]
+ optim/total_grad_norm=7.112
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,516,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7901/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,516,992,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7902/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,517,184,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7903/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,517,376,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7904/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,517,568,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7905/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,517,760,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7906/500000]
+ train/ActionL1Loss=0.0498
+ throughput/total_tokens=1,517,952,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7907/500000]
+ train/ActionL1Loss=0.0768
+ throughput/total_tokens=1,518,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7908/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,518,336,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=7909/500000]
+ train/ActionL1Loss=0.0431
+ throughput/total_tokens=1,518,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7910/500000]
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,518,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=7911/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,518,912,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7912/500000]
+ train/ActionL1Loss=0.0470
+ throughput/total_tokens=1,519,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7913/500000]
+ train/ActionL1Loss=0.0494
+ throughput/total_tokens=1,519,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7914/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,519,488,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7915/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,519,680,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7916/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,519,872,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=7917/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,520,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7918/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,520,256,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7919/500000]
+ train/ActionL1Loss=0.0400
+ throughput/total_tokens=1,520,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=7920/500000]
+ optim/total_grad_norm=6.966
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,520,640,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7921/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,520,832,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7922/500000]
+ train/ActionL1Loss=0.0563
+ throughput/total_tokens=1,521,024,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7923/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,521,216,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7924/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,521,408,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7925/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=1,521,600,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7926/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,521,792,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7927/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,521,984,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7928/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,522,176,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7929/500000]
+ train/ActionL1Loss=0.0751
+ throughput/total_tokens=1,522,368,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7930/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,522,560,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7931/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,522,752,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7932/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,522,944,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7933/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,523,136,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7934/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,523,328,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7935/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,523,520,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7936/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,523,712,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7937/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,523,904,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7938/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,524,096,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7939/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,524,288,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7940/500000]
+ optim/total_grad_norm=5.282
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,524,480,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7941/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,524,672,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7942/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,524,864,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7943/500000]
+ train/ActionL1Loss=0.0460
+ throughput/total_tokens=1,525,056,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7944/500000]
+ train/ActionL1Loss=0.0873
+ throughput/total_tokens=1,525,248,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7945/500000]
+ train/ActionL1Loss=0.0474
+ throughput/total_tokens=1,525,440,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7946/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,525,632,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7947/500000]
+ train/ActionL1Loss=0.0706
+ throughput/total_tokens=1,525,824,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7948/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,526,016,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7949/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,526,208,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7950/500000]
+ train/ActionL1Loss=0.0728
+ throughput/total_tokens=1,526,400,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7951/500000]
+ train/ActionL1Loss=0.0670
+ throughput/total_tokens=1,526,592,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7952/500000]
+ train/ActionL1Loss=0.0745
+ throughput/total_tokens=1,526,784,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7953/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,526,976,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7954/500000]
+ train/ActionL1Loss=0.0789
+ throughput/total_tokens=1,527,168,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7955/500000]
+ train/ActionL1Loss=0.0477
+ throughput/total_tokens=1,527,360,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7956/500000]
+ train/ActionL1Loss=0.0466
+ throughput/total_tokens=1,527,552,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7957/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,527,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7958/500000]
+ train/ActionL1Loss=0.0924
+ throughput/total_tokens=1,527,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7959/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,528,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=7960/500000]
+ optim/total_grad_norm=6.486
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,528,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=7961/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,528,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7962/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,528,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7963/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,528,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7964/500000]
+ train/ActionL1Loss=0.0523
+ throughput/total_tokens=1,529,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7965/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,529,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7966/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,529,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7967/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,529,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7968/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,529,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7969/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,530,048,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=7970/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,530,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7971/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,530,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7972/500000]
+ train/ActionL1Loss=0.0757
+ throughput/total_tokens=1,530,624,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7973/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,530,816,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7974/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,531,008,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7975/500000]
+ train/ActionL1Loss=0.0533
+ throughput/total_tokens=1,531,200,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7976/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,531,392,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7977/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,531,584,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7978/500000]
+ train/ActionL1Loss=0.0922
+ throughput/total_tokens=1,531,776,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7979/500000]
+ train/ActionL1Loss=0.0487
+ throughput/total_tokens=1,531,968,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7980/500000]
+ optim/total_grad_norm=3.987
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,532,160,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7981/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,532,352,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7982/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,532,544,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7983/500000]
+ train/ActionL1Loss=0.0754
+ throughput/total_tokens=1,532,736,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=7984/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,532,928,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7985/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,533,120,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7986/500000]
+ train/ActionL1Loss=0.0563
+ throughput/total_tokens=1,533,312,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7987/500000]
+ train/ActionL1Loss=0.0459
+ throughput/total_tokens=1,533,504,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7988/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,533,696,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7989/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,533,888,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7990/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,534,080,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=7991/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,534,272,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7992/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,534,464,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7993/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,534,656,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7994/500000]
+ train/ActionL1Loss=0.0785
+ throughput/total_tokens=1,534,848,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7995/500000]
+ train/ActionL1Loss=0.0819
+ throughput/total_tokens=1,535,040,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7996/500000]
+ train/ActionL1Loss=0.0869
+ throughput/total_tokens=1,535,232,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7997/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,535,424,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7998/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=1,535,616,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=7999/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,535,808,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8000/500000]
+ optim/total_grad_norm=6.170
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,536,000,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+10/02 [12:18:58] INFO | >> Saving config... checkpoint.py:608
+10/02 [12:19:40] INFO | >> Saving model state... checkpoint.py:796
+10/02 [12:20:52] INFO | >> Saving optim state... checkpoint.py:811
+10/02 [12:22:25] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=8001/500000]
+ train/ActionL1Loss=0.0490
+ throughput/total_tokens=1,536,192,000
+ throughput/device/tokens_per_second=1,213
+ throughput/device/batches_per_second=0.0506
+[step=8002/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,536,384,000
+ throughput/device/tokens_per_second=1,186
+ throughput/device/batches_per_second=0.0495
+[step=8003/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,536,576,000
+ throughput/device/tokens_per_second=1,175
+ throughput/device/batches_per_second=0.0490
+[step=8004/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,536,768,000
+ throughput/device/tokens_per_second=1,171
+ throughput/device/batches_per_second=0.0488
+[step=8005/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,536,960,000
+ throughput/device/tokens_per_second=1,170
+ throughput/device/batches_per_second=0.0488
+[step=8006/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,537,152,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8007/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,537,344,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8008/500000]
+ train/ActionL1Loss=0.0539
+ throughput/total_tokens=1,537,536,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8009/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,537,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8010/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,537,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8011/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,538,112,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8012/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,538,304,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8013/500000]
+ train/ActionL1Loss=0.0740
+ throughput/total_tokens=1,538,496,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8014/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,538,688,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8015/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,538,880,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8016/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,539,072,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8017/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,539,264,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8018/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,539,456,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8019/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,539,648,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8020/500000]
+ optim/total_grad_norm=6.306
+ train/ActionL1Loss=0.0395
+ throughput/total_tokens=1,539,840,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8021/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,540,032,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8022/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,540,224,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8023/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,540,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8024/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,540,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8025/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,540,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8026/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,540,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8027/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,541,184,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8028/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,541,376,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8029/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,541,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8030/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,541,760,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8031/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,541,952,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8032/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,542,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8033/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,542,336,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8034/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,542,528,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8035/500000]
+ train/ActionL1Loss=0.0470
+ throughput/total_tokens=1,542,720,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8036/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,542,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8037/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,543,104,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8038/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,543,296,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8039/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,543,488,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8040/500000]
+ optim/total_grad_norm=7.374
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,543,680,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8041/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,543,872,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8042/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,544,064,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8043/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,544,256,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8044/500000]
+ train/ActionL1Loss=0.0527
+ throughput/total_tokens=1,544,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8045/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,544,640,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8046/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,544,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8047/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,545,024,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8048/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,545,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8049/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,545,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8050/500000]
+ train/ActionL1Loss=0.0497
+ throughput/total_tokens=1,545,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8051/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,545,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8052/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,545,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8053/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,546,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8054/500000]
+ train/ActionL1Loss=0.0641
+ throughput/total_tokens=1,546,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8055/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,546,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8056/500000]
+ train/ActionL1Loss=0.0904
+ throughput/total_tokens=1,546,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8057/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,546,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8058/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,547,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8059/500000]
+ train/ActionL1Loss=0.0820
+ throughput/total_tokens=1,547,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8060/500000]
+ optim/total_grad_norm=7.330
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,547,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8061/500000]
+ train/ActionL1Loss=0.0799
+ throughput/total_tokens=1,547,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8062/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,547,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8063/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,548,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8064/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,548,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8065/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,548,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8066/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,548,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8067/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,548,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8068/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,549,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8069/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=1,549,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8070/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,549,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8071/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,549,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8072/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,549,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8073/500000]
+ train/ActionL1Loss=0.0685
+ throughput/total_tokens=1,550,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8074/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,550,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8075/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,550,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8076/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,550,592,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8077/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,550,784,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8078/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,550,976,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8079/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,551,168,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8080/500000]
+ optim/total_grad_norm=7.223
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,551,360,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8081/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,551,552,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8082/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,551,744,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8083/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,551,936,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8084/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,552,128,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8085/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,552,320,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8086/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,552,512,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8087/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,552,704,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8088/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,552,896,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8089/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,553,088,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8090/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,553,280,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8091/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,553,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8092/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,553,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8093/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,553,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8094/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,554,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8095/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,554,240,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8096/500000]
+ train/ActionL1Loss=0.0691
+ throughput/total_tokens=1,554,432,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8097/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,554,624,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8098/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,554,816,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8099/500000]
+ train/ActionL1Loss=0.0847
+ throughput/total_tokens=1,555,008,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8100/500000]
+ optim/total_grad_norm=5.448
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,555,200,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8101/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,555,392,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=8102/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,555,584,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=8103/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=1,555,776,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=8104/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,555,968,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=8105/500000]
+ train/ActionL1Loss=0.0423
+ throughput/total_tokens=1,556,160,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=8106/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,556,352,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=8107/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,556,544,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0485
+[step=8108/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,556,736,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8109/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=1,556,928,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8110/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,557,120,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8111/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,557,312,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8112/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,557,504,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8113/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,557,696,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8114/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,557,888,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8115/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,558,080,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8116/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,558,272,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8117/500000]
+ train/ActionL1Loss=0.0673
+ throughput/total_tokens=1,558,464,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8118/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,558,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8119/500000]
+ train/ActionL1Loss=0.0493
+ throughput/total_tokens=1,558,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8120/500000]
+ optim/total_grad_norm=5.159
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,559,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8121/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,559,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8122/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,559,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8123/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,559,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8124/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,559,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8125/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,560,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8126/500000]
+ train/ActionL1Loss=0.0647
+ throughput/total_tokens=1,560,192,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8127/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,560,384,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8128/500000]
+ train/ActionL1Loss=0.0696
+ throughput/total_tokens=1,560,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8129/500000]
+ train/ActionL1Loss=0.0450
+ throughput/total_tokens=1,560,768,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8130/500000]
+ train/ActionL1Loss=0.0407
+ throughput/total_tokens=1,560,960,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8131/500000]
+ train/ActionL1Loss=0.0737
+ throughput/total_tokens=1,561,152,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8132/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,561,344,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8133/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,561,536,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8134/500000]
+ train/ActionL1Loss=0.0472
+ throughput/total_tokens=1,561,728,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8135/500000]
+ train/ActionL1Loss=0.0480
+ throughput/total_tokens=1,561,920,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8136/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=1,562,112,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8137/500000]
+ train/ActionL1Loss=0.0479
+ throughput/total_tokens=1,562,304,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8138/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,562,496,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8139/500000]
+ train/ActionL1Loss=0.0722
+ throughput/total_tokens=1,562,688,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8140/500000]
+ optim/total_grad_norm=5.230
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,562,880,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8141/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,563,072,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8142/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,563,264,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8143/500000]
+ train/ActionL1Loss=0.0462
+ throughput/total_tokens=1,563,456,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8144/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,563,648,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8145/500000]
+ train/ActionL1Loss=0.0469
+ throughput/total_tokens=1,563,840,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8146/500000]
+ train/ActionL1Loss=0.0607
+ throughput/total_tokens=1,564,032,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8147/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,564,224,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8148/500000]
+ train/ActionL1Loss=0.0515
+ throughput/total_tokens=1,564,416,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8149/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,564,608,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8150/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,564,800,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8151/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,564,992,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8152/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,565,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8153/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,565,376,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8154/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,565,568,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8155/500000]
+ train/ActionL1Loss=0.0446
+ throughput/total_tokens=1,565,760,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8156/500000]
+ train/ActionL1Loss=0.0480
+ throughput/total_tokens=1,565,952,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8157/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,566,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8158/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,566,336,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8159/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,566,528,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8160/500000]
+ optim/total_grad_norm=6.170
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,566,720,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8161/500000]
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,566,912,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8162/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,567,104,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8163/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,567,296,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8164/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,567,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8165/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,567,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8166/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,567,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8167/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,568,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8168/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,568,256,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8169/500000]
+ train/ActionL1Loss=0.0463
+ throughput/total_tokens=1,568,448,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8170/500000]
+ train/ActionL1Loss=0.0798
+ throughput/total_tokens=1,568,640,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8171/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,568,832,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8172/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,569,024,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8173/500000]
+ train/ActionL1Loss=0.0752
+ throughput/total_tokens=1,569,216,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8174/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,569,408,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8175/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,569,600,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8176/500000]
+ train/ActionL1Loss=0.0437
+ throughput/total_tokens=1,569,792,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8177/500000]
+ train/ActionL1Loss=0.0429
+ throughput/total_tokens=1,569,984,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8178/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,570,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8179/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,570,368,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8180/500000]
+ optim/total_grad_norm=6.299
+ train/ActionL1Loss=0.0547
+ throughput/total_tokens=1,570,560,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8181/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,570,752,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8182/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,570,944,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8183/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,571,136,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8184/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,571,328,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8185/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,571,520,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8186/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,571,712,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8187/500000]
+ train/ActionL1Loss=0.0804
+ throughput/total_tokens=1,571,904,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8188/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,572,096,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8189/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,572,288,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8190/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,572,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8191/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,572,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8192/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,572,864,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8193/500000]
+ train/ActionL1Loss=0.0461
+ throughput/total_tokens=1,573,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8194/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,573,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8195/500000]
+ train/ActionL1Loss=0.0689
+ throughput/total_tokens=1,573,440,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8196/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,573,632,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8197/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,573,824,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8198/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,574,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8199/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,574,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8200/500000]
+ optim/total_grad_norm=5.098
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,574,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8201/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,574,592,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8202/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,574,784,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8203/500000]
+ train/ActionL1Loss=0.0777
+ throughput/total_tokens=1,574,976,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8204/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,575,168,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8205/500000]
+ train/ActionL1Loss=0.0515
+ throughput/total_tokens=1,575,360,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8206/500000]
+ train/ActionL1Loss=0.0707
+ throughput/total_tokens=1,575,552,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8207/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,575,744,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8208/500000]
+ train/ActionL1Loss=0.0731
+ throughput/total_tokens=1,575,936,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8209/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,576,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8210/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,576,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8211/500000]
+ train/ActionL1Loss=0.0657
+ throughput/total_tokens=1,576,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8212/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,576,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8213/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,576,896,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8214/500000]
+ train/ActionL1Loss=0.0477
+ throughput/total_tokens=1,577,088,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8215/500000]
+ train/ActionL1Loss=0.0658
+ throughput/total_tokens=1,577,280,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8216/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,577,472,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8217/500000]
+ train/ActionL1Loss=0.0597
+ throughput/total_tokens=1,577,664,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8218/500000]
+ train/ActionL1Loss=0.0378
+ throughput/total_tokens=1,577,856,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8219/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,578,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8220/500000]
+ optim/total_grad_norm=5.415
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,578,240,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8221/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,578,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8222/500000]
+ train/ActionL1Loss=0.0406
+ throughput/total_tokens=1,578,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8223/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,578,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8224/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,579,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8225/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,579,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8226/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,579,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8227/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,579,584,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8228/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,579,776,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8229/500000]
+ train/ActionL1Loss=0.0702
+ throughput/total_tokens=1,579,968,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8230/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,580,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8231/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,580,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8232/500000]
+ train/ActionL1Loss=0.0526
+ throughput/total_tokens=1,580,544,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8233/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,580,736,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8234/500000]
+ train/ActionL1Loss=0.0631
+ throughput/total_tokens=1,580,928,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8235/500000]
+ train/ActionL1Loss=0.0708
+ throughput/total_tokens=1,581,120,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8236/500000]
+ train/ActionL1Loss=0.0502
+ throughput/total_tokens=1,581,312,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8237/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,581,504,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8238/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,581,696,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8239/500000]
+ train/ActionL1Loss=0.0383
+ throughput/total_tokens=1,581,888,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8240/500000]
+ optim/total_grad_norm=5.537
+ train/ActionL1Loss=0.0660
+ throughput/total_tokens=1,582,080,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8241/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,582,272,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8242/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,582,464,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8243/500000]
+ train/ActionL1Loss=0.0811
+ throughput/total_tokens=1,582,656,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8244/500000]
+ train/ActionL1Loss=0.0736
+ throughput/total_tokens=1,582,848,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8245/500000]
+ train/ActionL1Loss=0.0500
+ throughput/total_tokens=1,583,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8246/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,583,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8247/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=1,583,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8248/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,583,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8249/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,583,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8250/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,584,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8251/500000]
+ train/ActionL1Loss=0.0794
+ throughput/total_tokens=1,584,192,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8252/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,584,384,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8253/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,584,576,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8254/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,584,768,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8255/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,584,960,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8256/500000]
+ train/ActionL1Loss=0.0474
+ throughput/total_tokens=1,585,152,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8257/500000]
+ train/ActionL1Loss=0.0418
+ throughput/total_tokens=1,585,344,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8258/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,585,536,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8259/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,585,728,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8260/500000]
+ optim/total_grad_norm=5.470
+ train/ActionL1Loss=0.0652
+ throughput/total_tokens=1,585,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8261/500000]
+ train/ActionL1Loss=0.0552
+ throughput/total_tokens=1,586,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8262/500000]
+ train/ActionL1Loss=0.0662
+ throughput/total_tokens=1,586,304,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8263/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,586,496,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8264/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,586,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8265/500000]
+ train/ActionL1Loss=0.0514
+ throughput/total_tokens=1,586,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8266/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,587,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8267/500000]
+ train/ActionL1Loss=0.0427
+ throughput/total_tokens=1,587,264,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8268/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,587,456,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8269/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,587,648,000
+ throughput/device/tokens_per_second=1,163
+ throughput/device/batches_per_second=0.0485
+[step=8270/500000]
+ train/ActionL1Loss=0.0545
+ throughput/total_tokens=1,587,840,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8271/500000]
+ train/ActionL1Loss=0.0675
+ throughput/total_tokens=1,588,032,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8272/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,588,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8273/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,588,416,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8274/500000]
+ train/ActionL1Loss=0.0444
+ throughput/total_tokens=1,588,608,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8275/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,588,800,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8276/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,588,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8277/500000]
+ train/ActionL1Loss=0.0394
+ throughput/total_tokens=1,589,184,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8278/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,589,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8279/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,589,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8280/500000]
+ optim/total_grad_norm=6.555
+ train/ActionL1Loss=0.0638
+ throughput/total_tokens=1,589,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8281/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,589,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8282/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,590,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8283/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,590,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8284/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,590,528,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8285/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,590,720,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8286/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,590,912,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8287/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,591,104,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8288/500000]
+ train/ActionL1Loss=0.0464
+ throughput/total_tokens=1,591,296,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8289/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,591,488,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8290/500000]
+ train/ActionL1Loss=0.0492
+ throughput/total_tokens=1,591,680,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8291/500000]
+ train/ActionL1Loss=0.0429
+ throughput/total_tokens=1,591,872,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8292/500000]
+ train/ActionL1Loss=0.0527
+ throughput/total_tokens=1,592,064,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8293/500000]
+ train/ActionL1Loss=0.0477
+ throughput/total_tokens=1,592,256,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8294/500000]
+ train/ActionL1Loss=0.0665
+ throughput/total_tokens=1,592,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8295/500000]
+ train/ActionL1Loss=0.0484
+ throughput/total_tokens=1,592,640,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8296/500000]
+ train/ActionL1Loss=0.0502
+ throughput/total_tokens=1,592,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8297/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,593,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8298/500000]
+ train/ActionL1Loss=0.0559
+ throughput/total_tokens=1,593,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8299/500000]
+ train/ActionL1Loss=0.0636
+ throughput/total_tokens=1,593,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8300/500000]
+ optim/total_grad_norm=4.815
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,593,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8301/500000]
+ train/ActionL1Loss=0.0596
+ throughput/total_tokens=1,593,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8302/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,593,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8303/500000]
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,594,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8304/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,594,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8305/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,594,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8306/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,594,752,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8307/500000]
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,594,944,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8308/500000]
+ train/ActionL1Loss=0.0453
+ throughput/total_tokens=1,595,136,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8309/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,595,328,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8310/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,595,520,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8311/500000]
+ train/ActionL1Loss=0.0748
+ throughput/total_tokens=1,595,712,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8312/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=1,595,904,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8313/500000]
+ train/ActionL1Loss=0.0502
+ throughput/total_tokens=1,596,096,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8314/500000]
+ train/ActionL1Loss=0.0374
+ throughput/total_tokens=1,596,288,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8315/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,596,480,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8316/500000]
+ train/ActionL1Loss=0.0563
+ throughput/total_tokens=1,596,672,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8317/500000]
+ train/ActionL1Loss=0.0389
+ throughput/total_tokens=1,596,864,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8318/500000]
+ train/ActionL1Loss=0.0460
+ throughput/total_tokens=1,597,056,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8319/500000]
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=1,597,248,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8320/500000]
+ optim/total_grad_norm=5.330
+ train/ActionL1Loss=0.0771
+ throughput/total_tokens=1,597,440,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8321/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,597,632,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8322/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,597,824,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8323/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,598,016,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8324/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,598,208,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8325/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,598,400,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8326/500000]
+ train/ActionL1Loss=0.0724
+ throughput/total_tokens=1,598,592,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8327/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,598,784,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8328/500000]
+ train/ActionL1Loss=0.0642
+ throughput/total_tokens=1,598,976,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8329/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,599,168,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8330/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,599,360,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8331/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,599,552,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8332/500000]
+ train/ActionL1Loss=0.0468
+ throughput/total_tokens=1,599,744,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8333/500000]
+ train/ActionL1Loss=0.0678
+ throughput/total_tokens=1,599,936,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8334/500000]
+ train/ActionL1Loss=0.0445
+ throughput/total_tokens=1,600,128,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8335/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,600,320,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8336/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,600,512,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8337/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=1,600,704,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8338/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,600,896,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8339/500000]
+ train/ActionL1Loss=0.0769
+ throughput/total_tokens=1,601,088,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8340/500000]
+ optim/total_grad_norm=6.980
+ train/ActionL1Loss=0.0569
+ throughput/total_tokens=1,601,280,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8341/500000]
+ train/ActionL1Loss=0.0721
+ throughput/total_tokens=1,601,472,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8342/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,601,664,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8343/500000]
+ train/ActionL1Loss=0.0719
+ throughput/total_tokens=1,601,856,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8344/500000]
+ train/ActionL1Loss=0.0434
+ throughput/total_tokens=1,602,048,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8345/500000]
+ train/ActionL1Loss=0.0812
+ throughput/total_tokens=1,602,240,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8346/500000]
+ train/ActionL1Loss=0.0538
+ throughput/total_tokens=1,602,432,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8347/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,602,624,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8348/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,602,816,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8349/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,603,008,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8350/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,603,200,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8351/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,603,392,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8352/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,603,584,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8353/500000]
+ train/ActionL1Loss=0.0733
+ throughput/total_tokens=1,603,776,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8354/500000]
+ train/ActionL1Loss=0.0471
+ throughput/total_tokens=1,603,968,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8355/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,604,160,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8356/500000]
+ train/ActionL1Loss=0.0464
+ throughput/total_tokens=1,604,352,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8357/500000]
+ train/ActionL1Loss=0.0551
+ throughput/total_tokens=1,604,544,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8358/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=1,604,736,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8359/500000]
+ train/ActionL1Loss=0.0463
+ throughput/total_tokens=1,604,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8360/500000]
+ optim/total_grad_norm=6.697
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,605,120,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8361/500000]
+ train/ActionL1Loss=0.0894
+ throughput/total_tokens=1,605,312,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8362/500000]
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,605,504,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8363/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,605,696,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8364/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,605,888,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8365/500000]
+ train/ActionL1Loss=0.0617
+ throughput/total_tokens=1,606,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8366/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,606,272,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8367/500000]
+ train/ActionL1Loss=0.0472
+ throughput/total_tokens=1,606,464,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8368/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,606,656,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8369/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,606,848,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8370/500000]
+ train/ActionL1Loss=0.0760
+ throughput/total_tokens=1,607,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8371/500000]
+ train/ActionL1Loss=0.0471
+ throughput/total_tokens=1,607,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8372/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,607,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8373/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,607,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8374/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,607,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8375/500000]
+ train/ActionL1Loss=0.0808
+ throughput/total_tokens=1,608,000,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8376/500000]
+ train/ActionL1Loss=0.0868
+ throughput/total_tokens=1,608,192,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8377/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,608,384,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8378/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,608,576,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8379/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,608,768,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8380/500000]
+ optim/total_grad_norm=5.955
+ train/ActionL1Loss=0.0429
+ throughput/total_tokens=1,608,960,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8381/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,609,152,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8382/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,609,344,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8383/500000]
+ train/ActionL1Loss=0.0513
+ throughput/total_tokens=1,609,536,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8384/500000]
+ train/ActionL1Loss=0.0533
+ throughput/total_tokens=1,609,728,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8385/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,609,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8386/500000]
+ train/ActionL1Loss=0.0626
+ throughput/total_tokens=1,610,112,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8387/500000]
+ train/ActionL1Loss=0.0502
+ throughput/total_tokens=1,610,304,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8388/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,610,496,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8389/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,610,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8390/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,610,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8391/500000]
+ train/ActionL1Loss=0.0612
+ throughput/total_tokens=1,611,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8392/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,611,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8393/500000]
+ train/ActionL1Loss=0.0623
+ throughput/total_tokens=1,611,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8394/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=1,611,648,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8395/500000]
+ train/ActionL1Loss=0.0543
+ throughput/total_tokens=1,611,840,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8396/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,612,032,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8397/500000]
+ train/ActionL1Loss=0.0505
+ throughput/total_tokens=1,612,224,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8398/500000]
+ train/ActionL1Loss=0.0442
+ throughput/total_tokens=1,612,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8399/500000]
+ train/ActionL1Loss=0.0490
+ throughput/total_tokens=1,612,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8400/500000]
+ optim/total_grad_norm=6.614
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,612,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8401/500000]
+ train/ActionL1Loss=0.0639
+ throughput/total_tokens=1,612,992,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8402/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,613,184,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8403/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,613,376,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8404/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,613,568,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8405/500000]
+ train/ActionL1Loss=0.0718
+ throughput/total_tokens=1,613,760,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8406/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,613,952,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8407/500000]
+ train/ActionL1Loss=0.0598
+ throughput/total_tokens=1,614,144,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8408/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,614,336,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8409/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,614,528,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8410/500000]
+ train/ActionL1Loss=0.0407
+ throughput/total_tokens=1,614,720,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8411/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,614,912,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8412/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,615,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8413/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,615,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8414/500000]
+ train/ActionL1Loss=0.0764
+ throughput/total_tokens=1,615,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8415/500000]
+ train/ActionL1Loss=0.0688
+ throughput/total_tokens=1,615,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8416/500000]
+ train/ActionL1Loss=0.0583
+ throughput/total_tokens=1,615,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8417/500000]
+ train/ActionL1Loss=0.0535
+ throughput/total_tokens=1,616,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8418/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,616,256,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8419/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,616,448,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8420/500000]
+ optim/total_grad_norm=6.694
+ train/ActionL1Loss=0.0646
+ throughput/total_tokens=1,616,640,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8421/500000]
+ train/ActionL1Loss=0.0592
+ throughput/total_tokens=1,616,832,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8422/500000]
+ train/ActionL1Loss=0.0418
+ throughput/total_tokens=1,617,024,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8423/500000]
+ train/ActionL1Loss=0.0772
+ throughput/total_tokens=1,617,216,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8424/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,617,408,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8425/500000]
+ train/ActionL1Loss=0.0802
+ throughput/total_tokens=1,617,600,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8426/500000]
+ train/ActionL1Loss=0.0500
+ throughput/total_tokens=1,617,792,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8427/500000]
+ train/ActionL1Loss=0.0664
+ throughput/total_tokens=1,617,984,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8428/500000]
+ train/ActionL1Loss=0.0618
+ throughput/total_tokens=1,618,176,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8429/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,618,368,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8430/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,618,560,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8431/500000]
+ train/ActionL1Loss=0.0763
+ throughput/total_tokens=1,618,752,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8432/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,618,944,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8433/500000]
+ train/ActionL1Loss=0.0835
+ throughput/total_tokens=1,619,136,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8434/500000]
+ train/ActionL1Loss=0.0575
+ throughput/total_tokens=1,619,328,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8435/500000]
+ train/ActionL1Loss=0.0467
+ throughput/total_tokens=1,619,520,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8436/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,619,712,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8437/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,619,904,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8438/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,620,096,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8439/500000]
+ train/ActionL1Loss=0.0496
+ throughput/total_tokens=1,620,288,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8440/500000]
+ optim/total_grad_norm=5.799
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,620,480,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8441/500000]
+ train/ActionL1Loss=0.0520
+ throughput/total_tokens=1,620,672,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8442/500000]
+ train/ActionL1Loss=0.0443
+ throughput/total_tokens=1,620,864,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8443/500000]
+ train/ActionL1Loss=0.0676
+ throughput/total_tokens=1,621,056,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8444/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,621,248,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8445/500000]
+ train/ActionL1Loss=0.0650
+ throughput/total_tokens=1,621,440,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8446/500000]
+ train/ActionL1Loss=0.0553
+ throughput/total_tokens=1,621,632,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8447/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,621,824,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8448/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,622,016,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8449/500000]
+ train/ActionL1Loss=0.0640
+ throughput/total_tokens=1,622,208,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8450/500000]
+ train/ActionL1Loss=0.0530
+ throughput/total_tokens=1,622,400,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8451/500000]
+ train/ActionL1Loss=0.0457
+ throughput/total_tokens=1,622,592,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8452/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,622,784,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8453/500000]
+ train/ActionL1Loss=0.0723
+ throughput/total_tokens=1,622,976,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8454/500000]
+ train/ActionL1Loss=0.0649
+ throughput/total_tokens=1,623,168,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8455/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,623,360,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8456/500000]
+ train/ActionL1Loss=0.0498
+ throughput/total_tokens=1,623,552,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8457/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,623,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8458/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,623,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8459/500000]
+ train/ActionL1Loss=0.0565
+ throughput/total_tokens=1,624,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8460/500000]
+ optim/total_grad_norm=5.515
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,624,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8461/500000]
+ train/ActionL1Loss=0.0620
+ throughput/total_tokens=1,624,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8462/500000]
+ train/ActionL1Loss=0.0533
+ throughput/total_tokens=1,624,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8463/500000]
+ train/ActionL1Loss=0.0440
+ throughput/total_tokens=1,624,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8464/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,625,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8465/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,625,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8466/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,625,472,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8467/500000]
+ train/ActionL1Loss=0.0683
+ throughput/total_tokens=1,625,664,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8468/500000]
+ train/ActionL1Loss=0.0407
+ throughput/total_tokens=1,625,856,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8469/500000]
+ train/ActionL1Loss=0.0579
+ throughput/total_tokens=1,626,048,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8470/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,626,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8471/500000]
+ train/ActionL1Loss=0.0548
+ throughput/total_tokens=1,626,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8472/500000]
+ train/ActionL1Loss=0.0704
+ throughput/total_tokens=1,626,624,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8473/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,626,816,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8474/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=1,627,008,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8475/500000]
+ train/ActionL1Loss=0.0713
+ throughput/total_tokens=1,627,200,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8476/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,627,392,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8477/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,627,584,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8478/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,627,776,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8479/500000]
+ train/ActionL1Loss=0.0580
+ throughput/total_tokens=1,627,968,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8480/500000]
+ optim/total_grad_norm=7.240
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,628,160,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8481/500000]
+ train/ActionL1Loss=0.0711
+ throughput/total_tokens=1,628,352,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8482/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,628,544,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8483/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=1,628,736,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8484/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,628,928,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8485/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,629,120,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8486/500000]
+ train/ActionL1Loss=0.0695
+ throughput/total_tokens=1,629,312,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8487/500000]
+ train/ActionL1Loss=0.0424
+ throughput/total_tokens=1,629,504,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8488/500000]
+ train/ActionL1Loss=0.0822
+ throughput/total_tokens=1,629,696,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8489/500000]
+ train/ActionL1Loss=0.0727
+ throughput/total_tokens=1,629,888,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8490/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,630,080,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8491/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,630,272,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8492/500000]
+ train/ActionL1Loss=0.0570
+ throughput/total_tokens=1,630,464,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8493/500000]
+ train/ActionL1Loss=0.0705
+ throughput/total_tokens=1,630,656,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8494/500000]
+ train/ActionL1Loss=0.0566
+ throughput/total_tokens=1,630,848,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8495/500000]
+ train/ActionL1Loss=0.0426
+ throughput/total_tokens=1,631,040,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8496/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,631,232,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8497/500000]
+ train/ActionL1Loss=0.0500
+ throughput/total_tokens=1,631,424,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8498/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,631,616,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8499/500000]
+ train/ActionL1Loss=0.0829
+ throughput/total_tokens=1,631,808,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8500/500000]
+ optim/total_grad_norm=5.669
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,632,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+10/02 [15:15:12] INFO | >> Saving config... checkpoint.py:608
+10/02 [15:16:07] INFO | >> Saving model state... checkpoint.py:796
+10/02 [15:17:19] INFO | >> Saving optim state... checkpoint.py:811
+10/02 [15:18:51] INFO | >> Saving trainer state... checkpoint.py:670
+ INFO | >> Saving config... checkpoint.py:608
+[step=8501/500000]
+ train/ActionL1Loss=0.0632
+ throughput/total_tokens=1,632,192,000
+ throughput/device/tokens_per_second=1,209
+ throughput/device/batches_per_second=0.0504
+[step=8502/500000]
+ train/ActionL1Loss=0.0503
+ throughput/total_tokens=1,632,384,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8503/500000]
+ train/ActionL1Loss=0.0437
+ throughput/total_tokens=1,632,576,000
+ throughput/device/tokens_per_second=1,162
+ throughput/device/batches_per_second=0.0484
+[step=8504/500000]
+ train/ActionL1Loss=0.0491
+ throughput/total_tokens=1,632,768,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8505/500000]
+ train/ActionL1Loss=0.0663
+ throughput/total_tokens=1,632,960,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8506/500000]
+ train/ActionL1Loss=0.0815
+ throughput/total_tokens=1,633,152,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8507/500000]
+ train/ActionL1Loss=0.0818
+ throughput/total_tokens=1,633,344,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8508/500000]
+ train/ActionL1Loss=0.0605
+ throughput/total_tokens=1,633,536,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8509/500000]
+ train/ActionL1Loss=0.0554
+ throughput/total_tokens=1,633,728,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8510/500000]
+ train/ActionL1Loss=0.0742
+ throughput/total_tokens=1,633,920,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8511/500000]
+ train/ActionL1Loss=0.0684
+ throughput/total_tokens=1,634,112,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8512/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,634,304,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8513/500000]
+ train/ActionL1Loss=0.0457
+ throughput/total_tokens=1,634,496,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8514/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,634,688,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8515/500000]
+ train/ActionL1Loss=0.0567
+ throughput/total_tokens=1,634,880,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8516/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,635,072,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8517/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,635,264,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8518/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,635,456,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8519/500000]
+ train/ActionL1Loss=0.0440
+ throughput/total_tokens=1,635,648,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8520/500000]
+ optim/total_grad_norm=7.000
+ train/ActionL1Loss=0.0487
+ throughput/total_tokens=1,635,840,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8521/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,636,032,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8522/500000]
+ train/ActionL1Loss=0.0524
+ throughput/total_tokens=1,636,224,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8523/500000]
+ train/ActionL1Loss=0.0544
+ throughput/total_tokens=1,636,416,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8524/500000]
+ train/ActionL1Loss=0.0809
+ throughput/total_tokens=1,636,608,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8525/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,636,800,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8526/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,636,992,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8527/500000]
+ train/ActionL1Loss=0.0672
+ throughput/total_tokens=1,637,184,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8528/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,637,376,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8529/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,637,568,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8530/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,637,760,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8531/500000]
+ train/ActionL1Loss=0.0568
+ throughput/total_tokens=1,637,952,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8532/500000]
+ train/ActionL1Loss=0.0546
+ throughput/total_tokens=1,638,144,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8533/500000]
+ train/ActionL1Loss=0.0767
+ throughput/total_tokens=1,638,336,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8534/500000]
+ train/ActionL1Loss=0.0674
+ throughput/total_tokens=1,638,528,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8535/500000]
+ train/ActionL1Loss=0.0601
+ throughput/total_tokens=1,638,720,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8536/500000]
+ train/ActionL1Loss=0.0417
+ throughput/total_tokens=1,638,912,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8537/500000]
+ train/ActionL1Loss=0.0690
+ throughput/total_tokens=1,639,104,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8538/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,639,296,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8539/500000]
+ train/ActionL1Loss=0.0635
+ throughput/total_tokens=1,639,488,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8540/500000]
+ optim/total_grad_norm=5.603
+ train/ActionL1Loss=0.0471
+ throughput/total_tokens=1,639,680,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8541/500000]
+ train/ActionL1Loss=0.0506
+ throughput/total_tokens=1,639,872,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8542/500000]
+ train/ActionL1Loss=0.0656
+ throughput/total_tokens=1,640,064,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8543/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,640,256,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8544/500000]
+ train/ActionL1Loss=0.0630
+ throughput/total_tokens=1,640,448,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8545/500000]
+ train/ActionL1Loss=0.0507
+ throughput/total_tokens=1,640,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8546/500000]
+ train/ActionL1Loss=0.0627
+ throughput/total_tokens=1,640,832,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8547/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,641,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8548/500000]
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,641,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8549/500000]
+ train/ActionL1Loss=0.0510
+ throughput/total_tokens=1,641,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8550/500000]
+ train/ActionL1Loss=0.0587
+ throughput/total_tokens=1,641,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8551/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,641,792,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8552/500000]
+ train/ActionL1Loss=0.0542
+ throughput/total_tokens=1,641,984,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8553/500000]
+ train/ActionL1Loss=0.0404
+ throughput/total_tokens=1,642,176,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8554/500000]
+ train/ActionL1Loss=0.0682
+ throughput/total_tokens=1,642,368,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8555/500000]
+ train/ActionL1Loss=0.0402
+ throughput/total_tokens=1,642,560,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8556/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,642,752,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8557/500000]
+ train/ActionL1Loss=0.0865
+ throughput/total_tokens=1,642,944,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8558/500000]
+ train/ActionL1Loss=0.0516
+ throughput/total_tokens=1,643,136,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8559/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,643,328,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8560/500000]
+ optim/total_grad_norm=6.421
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,643,520,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8561/500000]
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=1,643,712,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8562/500000]
+ train/ActionL1Loss=0.0451
+ throughput/total_tokens=1,643,904,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8563/500000]
+ train/ActionL1Loss=0.0593
+ throughput/total_tokens=1,644,096,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8564/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,644,288,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8565/500000]
+ train/ActionL1Loss=0.0715
+ throughput/total_tokens=1,644,480,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8566/500000]
+ train/ActionL1Loss=0.0485
+ throughput/total_tokens=1,644,672,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8567/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,644,864,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8568/500000]
+ train/ActionL1Loss=0.0556
+ throughput/total_tokens=1,645,056,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8569/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,645,248,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8570/500000]
+ train/ActionL1Loss=0.0453
+ throughput/total_tokens=1,645,440,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8571/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,645,632,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8572/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,645,824,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8573/500000]
+ train/ActionL1Loss=0.0509
+ throughput/total_tokens=1,646,016,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8574/500000]
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,646,208,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8575/500000]
+ train/ActionL1Loss=0.0469
+ throughput/total_tokens=1,646,400,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8576/500000]
+ train/ActionL1Loss=0.0494
+ throughput/total_tokens=1,646,592,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8577/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,646,784,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8578/500000]
+ train/ActionL1Loss=0.0661
+ throughput/total_tokens=1,646,976,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8579/500000]
+ train/ActionL1Loss=0.0499
+ throughput/total_tokens=1,647,168,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8580/500000]
+ optim/total_grad_norm=4.187
+ train/ActionL1Loss=0.0700
+ throughput/total_tokens=1,647,360,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8581/500000]
+ train/ActionL1Loss=0.0604
+ throughput/total_tokens=1,647,552,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8582/500000]
+ train/ActionL1Loss=0.0574
+ throughput/total_tokens=1,647,744,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8583/500000]
+ train/ActionL1Loss=0.0562
+ throughput/total_tokens=1,647,936,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8584/500000]
+ train/ActionL1Loss=0.0644
+ throughput/total_tokens=1,648,128,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8585/500000]
+ train/ActionL1Loss=0.0557
+ throughput/total_tokens=1,648,320,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8586/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,648,512,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8587/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,648,704,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8588/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,648,896,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8589/500000]
+ train/ActionL1Loss=0.0528
+ throughput/total_tokens=1,649,088,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8590/500000]
+ train/ActionL1Loss=0.0426
+ throughput/total_tokens=1,649,280,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8591/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,649,472,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8592/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,649,664,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8593/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,649,856,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8594/500000]
+ train/ActionL1Loss=0.0668
+ throughput/total_tokens=1,650,048,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8595/500000]
+ train/ActionL1Loss=0.0483
+ throughput/total_tokens=1,650,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8596/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,650,432,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8597/500000]
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,650,624,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8598/500000]
+ train/ActionL1Loss=0.0537
+ throughput/total_tokens=1,650,816,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8599/500000]
+ train/ActionL1Loss=0.0443
+ throughput/total_tokens=1,651,008,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8600/500000]
+ optim/total_grad_norm=4.654
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,651,200,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8601/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,651,392,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8602/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,651,584,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8603/500000]
+ train/ActionL1Loss=0.0489
+ throughput/total_tokens=1,651,776,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8604/500000]
+ train/ActionL1Loss=0.0369
+ throughput/total_tokens=1,651,968,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8605/500000]
+ train/ActionL1Loss=0.0541
+ throughput/total_tokens=1,652,160,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8606/500000]
+ train/ActionL1Loss=0.0669
+ throughput/total_tokens=1,652,352,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8607/500000]
+ train/ActionL1Loss=0.0511
+ throughput/total_tokens=1,652,544,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8608/500000]
+ train/ActionL1Loss=0.0455
+ throughput/total_tokens=1,652,736,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8609/500000]
+ train/ActionL1Loss=0.0521
+ throughput/total_tokens=1,652,928,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8610/500000]
+ train/ActionL1Loss=0.0487
+ throughput/total_tokens=1,653,120,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8611/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,653,312,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8612/500000]
+ train/ActionL1Loss=0.0714
+ throughput/total_tokens=1,653,504,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8613/500000]
+ train/ActionL1Loss=0.0787
+ throughput/total_tokens=1,653,696,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8614/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,653,888,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8615/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,654,080,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8616/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,654,272,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8617/500000]
+ train/ActionL1Loss=0.0697
+ throughput/total_tokens=1,654,464,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8618/500000]
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,654,656,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8619/500000]
+ train/ActionL1Loss=0.0600
+ throughput/total_tokens=1,654,848,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8620/500000]
+ optim/total_grad_norm=5.179
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,655,040,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8621/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,655,232,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8622/500000]
+ train/ActionL1Loss=0.0582
+ throughput/total_tokens=1,655,424,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8623/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,655,616,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8624/500000]
+ train/ActionL1Loss=0.0698
+ throughput/total_tokens=1,655,808,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8625/500000]
+ train/ActionL1Loss=0.0629
+ throughput/total_tokens=1,656,000,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8626/500000]
+ train/ActionL1Loss=0.0609
+ throughput/total_tokens=1,656,192,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8627/500000]
+ train/ActionL1Loss=0.0834
+ throughput/total_tokens=1,656,384,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8628/500000]
+ train/ActionL1Loss=0.0734
+ throughput/total_tokens=1,656,576,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8629/500000]
+ train/ActionL1Loss=0.0534
+ throughput/total_tokens=1,656,768,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8630/500000]
+ train/ActionL1Loss=0.0392
+ throughput/total_tokens=1,656,960,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8631/500000]
+ train/ActionL1Loss=0.0466
+ throughput/total_tokens=1,657,152,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8632/500000]
+ train/ActionL1Loss=0.0659
+ throughput/total_tokens=1,657,344,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8633/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,657,536,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8634/500000]
+ train/ActionL1Loss=0.0814
+ throughput/total_tokens=1,657,728,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8635/500000]
+ train/ActionL1Loss=0.0549
+ throughput/total_tokens=1,657,920,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8636/500000]
+ train/ActionL1Loss=0.0726
+ throughput/total_tokens=1,658,112,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8637/500000]
+ train/ActionL1Loss=0.0643
+ throughput/total_tokens=1,658,304,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8638/500000]
+ train/ActionL1Loss=0.0431
+ throughput/total_tokens=1,658,496,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8639/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,658,688,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8640/500000]
+ optim/total_grad_norm=6.998
+ train/ActionL1Loss=0.0577
+ throughput/total_tokens=1,658,880,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8641/500000]
+ train/ActionL1Loss=0.0773
+ throughput/total_tokens=1,659,072,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8642/500000]
+ train/ActionL1Loss=0.0487
+ throughput/total_tokens=1,659,264,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8643/500000]
+ train/ActionL1Loss=0.0540
+ throughput/total_tokens=1,659,456,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8644/500000]
+ train/ActionL1Loss=0.0572
+ throughput/total_tokens=1,659,648,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8645/500000]
+ train/ActionL1Loss=0.0573
+ throughput/total_tokens=1,659,840,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8646/500000]
+ train/ActionL1Loss=0.0766
+ throughput/total_tokens=1,660,032,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8647/500000]
+ train/ActionL1Loss=0.0475
+ throughput/total_tokens=1,660,224,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8648/500000]
+ train/ActionL1Loss=0.0536
+ throughput/total_tokens=1,660,416,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8649/500000]
+ train/ActionL1Loss=0.0590
+ throughput/total_tokens=1,660,608,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+[step=8650/500000]
+ train/ActionL1Loss=0.0525
+ throughput/total_tokens=1,660,800,000
+ throughput/device/tokens_per_second=1,169
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8651/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,660,992,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8652/500000]
+ train/ActionL1Loss=0.0648
+ throughput/total_tokens=1,661,184,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8653/500000]
+ train/ActionL1Loss=0.0410
+ throughput/total_tokens=1,661,376,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8654/500000]
+ train/ActionL1Loss=0.0749
+ throughput/total_tokens=1,661,568,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8655/500000]
+ train/ActionL1Loss=0.0478
+ throughput/total_tokens=1,661,760,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8656/500000]
+ train/ActionL1Loss=0.0473
+ throughput/total_tokens=1,661,952,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8657/500000]
+ train/ActionL1Loss=0.0501
+ throughput/total_tokens=1,662,144,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8658/500000]
+ train/ActionL1Loss=0.0489
+ throughput/total_tokens=1,662,336,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8659/500000]
+ train/ActionL1Loss=0.0464
+ throughput/total_tokens=1,662,528,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8660/500000]
+ optim/total_grad_norm=5.296
+ train/ActionL1Loss=0.0550
+ throughput/total_tokens=1,662,720,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8661/500000]
+ train/ActionL1Loss=0.0492
+ throughput/total_tokens=1,662,912,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8662/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,663,104,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8663/500000]
+ train/ActionL1Loss=0.0428
+ throughput/total_tokens=1,663,296,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8664/500000]
+ train/ActionL1Loss=0.0476
+ throughput/total_tokens=1,663,488,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8665/500000]
+ train/ActionL1Loss=0.0513
+ throughput/total_tokens=1,663,680,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8666/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,663,872,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8667/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,664,064,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8668/500000]
+ train/ActionL1Loss=0.0578
+ throughput/total_tokens=1,664,256,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8669/500000]
+ train/ActionL1Loss=0.0423
+ throughput/total_tokens=1,664,448,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8670/500000]
+ train/ActionL1Loss=0.0495
+ throughput/total_tokens=1,664,640,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8671/500000]
+ train/ActionL1Loss=0.0602
+ throughput/total_tokens=1,664,832,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8672/500000]
+ train/ActionL1Loss=0.0610
+ throughput/total_tokens=1,665,024,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8673/500000]
+ train/ActionL1Loss=0.0735
+ throughput/total_tokens=1,665,216,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8674/500000]
+ train/ActionL1Loss=0.0671
+ throughput/total_tokens=1,665,408,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8675/500000]
+ train/ActionL1Loss=0.0654
+ throughput/total_tokens=1,665,600,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8676/500000]
+ train/ActionL1Loss=0.0486
+ throughput/total_tokens=1,665,792,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8677/500000]
+ train/ActionL1Loss=0.0584
+ throughput/total_tokens=1,665,984,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8678/500000]
+ train/ActionL1Loss=0.0729
+ throughput/total_tokens=1,666,176,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8679/500000]
+ train/ActionL1Loss=0.0633
+ throughput/total_tokens=1,666,368,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8680/500000]
+ optim/total_grad_norm=5.683
+ train/ActionL1Loss=0.0624
+ throughput/total_tokens=1,666,560,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8681/500000]
+ train/ActionL1Loss=0.0628
+ throughput/total_tokens=1,666,752,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8682/500000]
+ train/ActionL1Loss=0.0666
+ throughput/total_tokens=1,666,944,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8683/500000]
+ train/ActionL1Loss=0.0709
+ throughput/total_tokens=1,667,136,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8684/500000]
+ train/ActionL1Loss=0.0693
+ throughput/total_tokens=1,667,328,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8685/500000]
+ train/ActionL1Loss=0.0616
+ throughput/total_tokens=1,667,520,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8686/500000]
+ train/ActionL1Loss=0.0588
+ throughput/total_tokens=1,667,712,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8687/500000]
+ train/ActionL1Loss=0.0634
+ throughput/total_tokens=1,667,904,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8688/500000]
+ train/ActionL1Loss=0.0681
+ throughput/total_tokens=1,668,096,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8689/500000]
+ train/ActionL1Loss=0.0500
+ throughput/total_tokens=1,668,288,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8690/500000]
+ train/ActionL1Loss=0.0589
+ throughput/total_tokens=1,668,480,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8691/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,668,672,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8692/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,668,864,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8693/500000]
+ train/ActionL1Loss=0.0653
+ throughput/total_tokens=1,669,056,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8694/500000]
+ train/ActionL1Loss=0.0637
+ throughput/total_tokens=1,669,248,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8695/500000]
+ train/ActionL1Loss=0.0487
+ throughput/total_tokens=1,669,440,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8696/500000]
+ train/ActionL1Loss=0.0793
+ throughput/total_tokens=1,669,632,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8697/500000]
+ train/ActionL1Loss=0.0606
+ throughput/total_tokens=1,669,824,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8698/500000]
+ train/ActionL1Loss=0.0531
+ throughput/total_tokens=1,670,016,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8699/500000]
+ train/ActionL1Loss=0.0621
+ throughput/total_tokens=1,670,208,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8700/500000]
+ optim/total_grad_norm=7.272
+ train/ActionL1Loss=0.0599
+ throughput/total_tokens=1,670,400,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8701/500000]
+ train/ActionL1Loss=0.0622
+ throughput/total_tokens=1,670,592,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8702/500000]
+ train/ActionL1Loss=0.0770
+ throughput/total_tokens=1,670,784,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8703/500000]
+ train/ActionL1Loss=0.0608
+ throughput/total_tokens=1,670,976,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8704/500000]
+ train/ActionL1Loss=0.0581
+ throughput/total_tokens=1,671,168,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8705/500000]
+ train/ActionL1Loss=0.0619
+ throughput/total_tokens=1,671,360,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8706/500000]
+ train/ActionL1Loss=0.0625
+ throughput/total_tokens=1,671,552,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8707/500000]
+ train/ActionL1Loss=0.0576
+ throughput/total_tokens=1,671,744,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8708/500000]
+ train/ActionL1Loss=0.0615
+ throughput/total_tokens=1,671,936,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8709/500000]
+ train/ActionL1Loss=0.0687
+ throughput/total_tokens=1,672,128,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8710/500000]
+ train/ActionL1Loss=0.0703
+ throughput/total_tokens=1,672,320,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+ System/Peak GPU Memory (MB)=46,917
+[step=8711/500000]
+ train/ActionL1Loss=0.0571
+ throughput/total_tokens=1,672,512,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8712/500000]
+ train/ActionL1Loss=0.0746
+ throughput/total_tokens=1,672,704,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8713/500000]
+ train/ActionL1Loss=0.0645
+ throughput/total_tokens=1,672,896,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8714/500000]
+ train/ActionL1Loss=0.0667
+ throughput/total_tokens=1,673,088,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8715/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,673,280,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8716/500000]
+ train/ActionL1Loss=0.0591
+ throughput/total_tokens=1,673,472,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0486
+[step=8717/500000]
+ train/ActionL1Loss=0.0586
+ throughput/total_tokens=1,673,664,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8718/500000]
+ train/ActionL1Loss=0.0679
+ throughput/total_tokens=1,673,856,000
+ throughput/device/tokens_per_second=1,165
+ throughput/device/batches_per_second=0.0485
+[step=8719/500000]
+ train/ActionL1Loss=0.0732
+ throughput/total_tokens=1,674,048,000
+ throughput/device/tokens_per_second=1,164
+ throughput/device/batches_per_second=0.0485
+[step=8720/500000]
+ optim/total_grad_norm=5.450
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,674,240,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8721/500000]
+ train/ActionL1Loss=0.0694
+ throughput/total_tokens=1,674,432,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8722/500000]
+ train/ActionL1Loss=0.0614
+ throughput/total_tokens=1,674,624,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8723/500000]
+ train/ActionL1Loss=0.0558
+ throughput/total_tokens=1,674,816,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8724/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,675,008,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8725/500000]
+ train/ActionL1Loss=0.0611
+ throughput/total_tokens=1,675,200,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8726/500000]
+ train/ActionL1Loss=0.0555
+ throughput/total_tokens=1,675,392,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+[step=8727/500000]
+ train/ActionL1Loss=0.0603
+ throughput/total_tokens=1,675,584,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8728/500000]
+ train/ActionL1Loss=0.0561
+ throughput/total_tokens=1,675,776,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8729/500000]
+ train/ActionL1Loss=0.0519
+ throughput/total_tokens=1,675,968,000
+ throughput/device/tokens_per_second=1,168
+ throughput/device/batches_per_second=0.0487
+[step=8730/500000]
+ train/ActionL1Loss=0.0613
+ throughput/total_tokens=1,676,160,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0487
+ System/Peak GPU Memory (MB)=46,917
+[step=8731/500000]
+ train/ActionL1Loss=0.0522
+ throughput/total_tokens=1,676,352,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8732/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,676,544,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8733/500000]
+ train/ActionL1Loss=0.0529
+ throughput/total_tokens=1,676,736,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8734/500000]
+ train/ActionL1Loss=0.0532
+ throughput/total_tokens=1,676,928,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8735/500000]
+ train/ActionL1Loss=0.0564
+ throughput/total_tokens=1,677,120,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8736/500000]
+ train/ActionL1Loss=0.0585
+ throughput/total_tokens=1,677,312,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8737/500000]
+ train/ActionL1Loss=0.0651
+ throughput/total_tokens=1,677,504,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8738/500000]
+ train/ActionL1Loss=0.0477
+ throughput/total_tokens=1,677,696,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8739/500000]
+ train/ActionL1Loss=0.0759
+ throughput/total_tokens=1,677,888,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8740/500000]
+ optim/total_grad_norm=7.025
+ train/ActionL1Loss=0.0444
+ throughput/total_tokens=1,678,080,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+ System/Peak GPU Memory (MB)=46,917
+[step=8741/500000]
+ train/ActionL1Loss=0.0512
+ throughput/total_tokens=1,678,272,000
+ throughput/device/tokens_per_second=1,167
+ throughput/device/batches_per_second=0.0486
+[step=8742/500000]
+ train/ActionL1Loss=0.0517
+ throughput/total_tokens=1,678,464,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
+[step=8743/500000]
+ train/ActionL1Loss=0.0468
+ throughput/total_tokens=1,678,656,000
+ throughput/device/tokens_per_second=1,166
+ throughput/device/batches_per_second=0.0486
diff --git a/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/requirements.txt b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5a5e971a5d9f3b6f6a75248e203682c5fbcd69e8
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/requirements.txt
@@ -0,0 +1,286 @@
+ai2-molmo==0.0.0
+astunparse==1.6.3
+flatbuffers==25.2.10
+gast==0.6.0
+google-pasta==0.2.0
+h5py==3.14.0
+libclang==18.1.1
+Markdown==3.9
+namex==0.1.0
+opt_einsum==3.4.0
+optree==0.17.0
+tensorboard-data-server==0.7.2
+tensorflow-io-gcs-filesystem==0.37.1
+termcolor==3.1.0
+Werkzeug==3.1.3
+Brotli==1.1.0
+Farama-Notifications==0.0.4
+MarkupSafe==2.1.5
+PyYAML==6.0.2
+absl-py==2.3.1
+accelerate==1.10.1
+ai2-molmo==0.0.0
+aiofiles==24.1.0
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.15
+aiosignal==1.4.0
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+anyio==4.10.0
+array_record==0.8.1
+async-timeout==5.0.1
+attrs==25.3.0
+av==15.1.0
+backports.tarfile==1.2.0
+beaker-gantry==3.2.0
+beaker-py==2.5.0
+black==23.12.1
+blinker==1.9.0
+boltons==25.0.0
+boto3==1.40.33
+botocore==1.40.33
+build==1.3.0
+cached_path==1.7.3
+cached-property==2.0.1
+cachetools==5.5.2
+certifi==2025.8.3
+cffi==2.0.0
+charset-normalizer==3.4.3
+click==8.2.1
+click-help-colors==0.9.4
+click-option-group==0.5.7
+cloudpickle==3.1.1
+cmake==4.1.0
+contourpy==1.3.2
+cryptography==46.0.1
+cycler==0.12.1
+dataclass-extensions==0.2.3
+datasets==3.6.0
+decorator==5.2.1
+deepdiff==8.6.1
+diffusers==0.35.1
+dill==0.3.8
+distro==1.9.0
+dlimp==0.0.1
+dm-tree==0.1.9
+docutils==0.22.1
+draccus==0.10.0
+editdistance==0.8.1
+einops==0.8.1
+einops-exts==0.0.4
+et_xmlfile==2.0.0
+etils==1.13.0
+evdev==1.9.2
+exceptiongroup==1.3.0
+face==24.0.0
+fastapi==0.116.2
+ffmpy==0.6.1
+fiddle==0.3.0
+filelock==3.13.1
+Flask==3.1.2
+fonttools==4.60.0
+frozenlist==1.7.0
+fsspec==2023.9.2
+ftfy==6.3.1
+gcsfs==2023.9.2
+gitdb==4.0.12
+GitPython==3.1.45
+glom==24.11.0
+google-api-core==2.25.1
+google-auth==2.40.3
+google-auth-oauthlib==1.2.2
+google-cloud-core==2.4.3
+google-cloud-storage==2.19.0
+google-crc32c==1.7.1
+google-resumable-media==2.7.2
+googleapis-common-protos==1.70.0
+gradio==5.46.0
+gradio_client==1.13.0
+graphviz==0.21
+groovy==0.1.2
+grpcio==1.75.0
+gymnasium==0.29.1
+h11==0.16.0
+hf_transfer==0.1.9
+hf-xet==1.1.10
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.35.0
+id==1.5.0
+idna==3.10
+imageio==2.37.0
+imageio-ffmpeg==0.6.0
+importlib_metadata==8.7.0
+importlib_resources==6.5.2
+iniconfig==2.1.0
+inquirerpy==0.3.4
+isort==5.12.0
+itsdangerous==2.2.0
+jaraco.classes==3.4.0
+jaraco.context==6.0.1
+jaraco.functools==4.3.0
+jeepney==0.9.0
+Jinja2==3.1.4
+jiter==0.11.0
+jmespath==1.0.1
+joblib==1.5.2
+jsonlines==4.0.0
+keras==2.15.0
+keyring==25.6.0
+kiwisolver==1.4.9
+latex2sympy2_extended==1.10.2
+lerobot==0.3.4
+Levenshtein==0.27.1
+libcst==1.8.4
+lightning-utilities==0.15.2
+markdown-it-py==4.0.0
+math-verify==0.8.0
+matplotlib==3.10.6
+mdurl==0.1.2
+mergedeep==1.3.4
+ml-dtypes==0.2.0
+ml_dtypes==0.5.3
+more-itertools==10.8.0
+mpmath==1.3.0
+msgspec==0.19.0
+multidict==6.6.4
+multiprocess==0.70.16
+mypy==1.3.0
+mypy_extensions==1.1.0
+necessary==0.4.3
+networkx==3.3
+nh3==0.3.0
+nltk==3.9.1
+numpy==1.26.4
+oauthlib==3.3.1
+omegaconf==2.3.0
+openai==1.108.0
+opencv-python-headless==4.12.0.88
+OpenEXR==3.4.0
+openpyxl==3.1.5
+orderly-set==5.5.0
+orjson==3.11.3
+packaging==25.0
+pandas==2.3.2
+pathspec==0.12.1
+petname==2.6
+pfzy==0.3.4
+pillow==11.0.0
+pip==25.2
+platformdirs==4.4.0
+pluggy==1.6.0
+promise==2.3
+prompt_toolkit==3.0.52
+propcache==0.3.2
+proto-plus==1.26.1
+protobuf==4.21.12
+protobuf==6.32.1
+psutil==7.1.0
+pyarrow==21.0.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.23
+pydantic==2.11.9
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pynput==1.8.1
+pyparsing==3.2.4
+pyproject_hooks==1.2.0
+pyserial==3.5
+pytest==8.4.2
+pytest-sphinx==0.6.3
+python-dateutil==2.9.0.post0
+python-Levenshtein==0.27.1
+python-multipart==0.0.20
+python-xlib==0.33
+pytorch-triton-rocm==3.4.0
+pytz==2025.2
+pyyaml-include==1.4.1
+RapidFuzz==3.14.1
+readme_renderer==44.0
+regex==2025.9.1
+requests==2.32.5
+requests-oauthlib==2.0.0
+requests-toolbelt==1.0.0
+requirements-parser==0.13.0
+rerun-sdk==0.22.1
+rfc3986==2.0.0
+rich==13.9.4
+rsa==4.9.1
+ruff==0.13.0
+s3transfer==0.14.0
+safehttpx==0.1.6
+safetensors==0.6.2
+scikit-learn==1.7.2
+scipy==1.15.3
+SecretStorage==3.4.0
+semantic-version==2.10.0
+sentencepiece==0.2.1
+sentry-sdk==2.38.0
+setuptools==78.1.1
+shellingham==1.5.4
+six==1.17.0
+smart_open==7.3.1
+smashed==0.21.5
+smmap==5.0.2
+sniffio==1.3.1
+starlette==0.48.0
+sympy==1.13.3
+tensorboard==2.15.2
+tensorboard==2.19.0
+tensorflow==2.15.0
+tensorflow-addons==0.23.0
+tensorflow-datasets==4.9.3
+tensorflow-estimator==2.15.0
+tensorflow-graphics==2021.12.3
+tensorflow-metadata==1.17.2
+threadpoolctl==3.6.0
+timm==1.0.19
+tokenizers==0.22.0
+toml==0.10.2
+tomli==2.2.1
+tomlkit==0.13.3
+torch==2.8.0+rocm6.4
+torchcodec==0.5
+torchmetrics==1.8.2
+torchvision==0.23.0+rocm6.4
+tqdm==4.67.1
+transformers==4.56.1
+trimesh==4.8.2
+trouting==0.3.3
+twine==6.2.0
+typeguard==2.13.3
+typer==0.17.4
+typing_extensions==4.15.0
+typing-inspect==0.9.0
+typing-inspection==0.4.1
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+wandb==0.21.4
+wcwidth==0.2.13
+websockets==15.0.1
+wheel==0.45.1
+wrapt==1.14.2
+xxhash==3.5.0
+yarl==1.20.1
+zipp==3.23.0
+lerobot==0.3.4
+minLoRA==0.1.0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2
diff --git a/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/wandb-metadata.json b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/wandb-metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..1bf5af562d1ceeed09c0950a4120295c46a7bf44
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/files/wandb-metadata.json
@@ -0,0 +1,202 @@
+{
+ "os": "Linux-5.15.0-140-generic-x86_64-with-glibc2.35",
+ "python": "CPython 3.10.18",
+ "startedAt": "2025-09-30T13:12:50.512180Z",
+ "args": [
+ "qwen2_7b",
+ "save_folder=/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1",
+ "--vision_backbone",
+ "openai",
+ "--action_head",
+ "l1_regression",
+ "--seq_len",
+ "1600",
+ "--ft_llm",
+ "--checkpoint",
+ "/vast/users/xiaodan/zhangjian/molmo_data/Molmo-7B-D-0924",
+ "--device_train_microbatch_size",
+ "16",
+ "--global_batch_size",
+ "126",
+ "--dataset",
+ "vla_dataset_realworld",
+ "--llm_learning_rate",
+ "5e-5",
+ "--wandb_entity",
+ "henryeap",
+ "--wandb_project",
+ "a1-realworld",
+ "--wandb_run_name",
+ "realworld",
+ "--save_overwrite"
+ ],
+ "program": "/vast/users/xiaodan/zhangjian/A1/launch_scripts/train_vla.py",
+ "codePath": "launch_scripts/train_vla.py",
+ "codePathLocal": "launch_scripts/train_vla.py",
+ "git": {
+ "remote": "https://github.com/Spatialtemporal-AI/A1.git",
+ "commit": "f2afcc15e05f491a8e50add64395fc1db0a1188d"
+ },
+ "email": "ihenrykwok@outlook.com",
+ "root": "/vast/users/xiaodan/workspace/minghao.guo/warehouse_a1/ckpt/all_l1/wandb",
+ "host": "auh7-1b-gpu-266",
+ "executable": "/vast/users/xiaodan/miniconda3/envs/a1/bin/python3.10",
+ "cpu_count": 64,
+ "cpu_count_logical": 128,
+ "gpu": "Instinct MI210",
+ "gpu_count": 8,
+ "disk": {
+ "/": {
+ "total": "470343073792",
+ "used": "50365865984"
+ }
+ },
+ "memory": {
+ "total": "2434606915584"
+ },
+ "gpu_amd": [
+ {
+ "id": "5",
+ "uniqueId": "0x55366988951d07ad",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "6",
+ "uniqueId": "0x38b61e086b79539",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "7",
+ "uniqueId": "0xec87d8b9ba4d4aa9",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "2",
+ "uniqueId": "0x433fd00772eb3fb7",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "3",
+ "uniqueId": "0xdaf8107e479aecf0",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "1",
+ "uniqueId": "0x1633d5497b099b7f",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "4",
+ "uniqueId": "0xe7ce40a1ea1cbac2",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ },
+ {
+ "id": "0",
+ "uniqueId": "0xef820a5c24b0c740",
+ "vbiosVersion": "113-D67301V-073",
+ "performanceLevel": "auto",
+ "maxPower": "300.0",
+ "series": "Instinct MI210",
+ "model": "0x740f",
+ "vendor": "Advanced Micro Devices, Inc. [AMD/ATI]",
+ "sku": "D67301V",
+ "sclkRange": "500Mhz - 1700Mhz",
+ "mclkRange": "400Mhz - 1600Mhz"
+ }
+ ],
+ "slurm": {
+ "cluster_name": "ai-04r",
+ "conf": "/etc/slurm/slurm.conf",
+ "cpus_on_node": "128",
+ "gpus_on_node": "8",
+ "gtids": "0",
+ "job_account": "faculty-acc",
+ "job_cpus_per_node": "128",
+ "job_end_time": "1759497126",
+ "job_gid": "2000",
+ "job_gpus": "0,1,2,3,4,5,6,7",
+ "job_id": "2020",
+ "job_name": "mh_realworld",
+ "job_nodelist": "auh7-1b-gpu-266",
+ "job_num_nodes": "1",
+ "job_partition": "faculty",
+ "job_qos": "xdqos",
+ "job_start_time": "1759237926",
+ "job_uid": "2013",
+ "job_user": "xiaodan",
+ "jobid": "2020",
+ "localid": "0",
+ "nnodes": "1",
+ "nodeid": "0",
+ "nodelist": "auh7-1b-gpu-266",
+ "nprocs": "1",
+ "ntasks": "1",
+ "ntasks_per_node": "1",
+ "oom_kill_step": "0",
+ "prio_process": "0",
+ "procid": "0",
+ "submit_dir": "/vast/users/xiaodan/zhangjian/A1/launch_scripts",
+ "submit_host": "auh-1b-cpu-login-001",
+ "task_pid": "3355381",
+ "tasks_per_node": "1",
+ "topology_addr": "auh7-1b-gpu-266",
+ "topology_addr_pattern": "node"
+ },
+ "writerId": "br57kuzpd4cnjl4v1nt32cpk941kjn1w"
+}
\ No newline at end of file
diff --git a/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug-core.log b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug-core.log
new file mode 100644
index 0000000000000000000000000000000000000000..578a394a3ea8135822fdea1f1a024e6b4cbd66c4
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug-core.log
@@ -0,0 +1,6 @@
+{"time":"2025-09-30T13:12:50.569260504Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp7ktuvs87/port-3355558.txt","pid":3355558,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2025-09-30T13:12:50.569851234Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3355558}
+{"time":"2025-09-30T13:12:50.569857124Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3355558-3355728-1383521146/socket","Net":"unix"}}
+{"time":"2025-09-30T13:12:50.74547283Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2025-09-30T13:12:50.752005291Z","level":"INFO","msg":"handleInformInit: received","streamId":"ea1k0g3y","id":"1(@)"}
+{"time":"2025-09-30T13:12:51.778015112Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"ea1k0g3y","id":"1(@)"}
diff --git a/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug-internal.log b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug-internal.log
new file mode 100644
index 0000000000000000000000000000000000000000..f830e5e77134044528636ac1e7a7cedc908690b4
--- /dev/null
+++ b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug-internal.log
@@ -0,0 +1,9 @@
+{"time":"2025-09-30T13:12:50.753807221Z","level":"INFO","msg":"stream: starting","core version":"0.21.4"}
+{"time":"2025-09-30T13:12:51.777951571Z","level":"INFO","msg":"stream: created new stream","id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778008702Z","level":"INFO","msg":"stream: started","id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778019413Z","level":"INFO","msg":"handler: started","stream_id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778022263Z","level":"INFO","msg":"sender: started","stream_id":"ea1k0g3y"}
+{"time":"2025-09-30T13:12:51.778055373Z","level":"INFO","msg":"writer: started","stream_id":"ea1k0g3y"}
+{"time":"2025-10-01T17:48:53.058127813Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/henryeap/a1-realworld/ea1k0g3y/file_stream","body":"\n\n\n502 Server Error\n\n\nError: Server Error
\nThe server encountered a temporary error and could not complete your request.
Please try again in 30 seconds.\n
\n\n"}
+{"time":"2025-10-02T06:50:26.681725611Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/henryeap/a1-realworld/ea1k0g3y/file_stream","body":"\n\n\n502 Server Error\n\n\nError: Server Error
\nThe server encountered a temporary error and could not complete your request.
Please try again in 30 seconds.\n
\n\n"}
+{"time":"2025-10-02T07:20:37.728905409Z","level":"INFO","msg":"api: retrying error","error":"Post \"https://api.wandb.ai/graphql\": net/http: request canceled (Client.Timeout exceeded while awaiting headers)"}
diff --git a/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug.log b/all_l1/wandb/wandb/run-20250930_131250-ea1k0g3y/logs/debug.log
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/cleandesk50_flow_matching/wandb/wandb/run-20251008_163834-quokv8gn/run-quokv8gn.wandb b/cleandesk50_flow_matching/wandb/wandb/run-20251008_163834-quokv8gn/run-quokv8gn.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..ab7d3edeb78080f304eb9e5feae5958bb8bf7270
--- /dev/null
+++ b/cleandesk50_flow_matching/wandb/wandb/run-20251008_163834-quokv8gn/run-quokv8gn.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a745d4f17f11679cdf5bf92aec764de48409f778dd66d1a9d82556acc272c36
+size 212789
diff --git a/cleandesk50_l1_regression/wandb/wandb/run-20251008_163831-fqdwkc8m/run-fqdwkc8m.wandb b/cleandesk50_l1_regression/wandb/wandb/run-20251008_163831-fqdwkc8m/run-fqdwkc8m.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..5715f3ef97ddf3f48b4946ae665b0ca8a67383b9
--- /dev/null
+++ b/cleandesk50_l1_regression/wandb/wandb/run-20251008_163831-fqdwkc8m/run-fqdwkc8m.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcd41307edd91290339717d666254857fafd5f8d6cb82e9ad0e0ea1e6c93bfb5
+size 206094
diff --git a/cleandesk_flow_matching/step12000-action-head/metadata.pt b/cleandesk_flow_matching/step12000-action-head/metadata.pt
new file mode 100644
index 0000000000000000000000000000000000000000..752c8b547c4cac7f78746ac5066032167b1cf3c1
--- /dev/null
+++ b/cleandesk_flow_matching/step12000-action-head/metadata.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:995307502120af3866f237cd0bc484fc848a652539d28e53cbea882abc16ba6b
+size 1331
diff --git a/cleandesk_flow_matching/step12000-unsharded/lora.pt b/cleandesk_flow_matching/step12000-unsharded/lora.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d3eb8fc48e155d8229340760896a20ba9fc413e2
--- /dev/null
+++ b/cleandesk_flow_matching/step12000-unsharded/lora.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b09055f15a54dd092b4dd30833406731057005822da0c55c16231cf2e68f7f6
+size 1243
diff --git a/cleandesk_flow_matching/step12000-unsharded/train.pt b/cleandesk_flow_matching/step12000-unsharded/train.pt
new file mode 100644
index 0000000000000000000000000000000000000000..067eeb392ccd2fc0c39b2b3aad15ff7da8cc27e7
--- /dev/null
+++ b/cleandesk_flow_matching/step12000-unsharded/train.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:878ba9d72f994a81ed1a71b220badd89c2e842c5be4ff48e26d49c71f262847b
+size 15061
diff --git a/glue_flow_matching/wandb/wandb/run-20251002_163728-tmwli25x/run-tmwli25x.wandb b/glue_flow_matching/wandb/wandb/run-20251002_163728-tmwli25x/run-tmwli25x.wandb
new file mode 100644
index 0000000000000000000000000000000000000000..11aee9be87edb54d1f40db5064f016098eb1c351
--- /dev/null
+++ b/glue_flow_matching/wandb/wandb/run-20251002_163728-tmwli25x/run-tmwli25x.wandb
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de648494435cefb93fe9231d01df14e2e77c0a85d968f930b4d3fba3df52ff32
+size 740786176